ThesisPDF Available

No One Can Whistle a Symphony: Essays on Congressional Campaign Teams and Talent

Authors:

Abstract and Figures

Political power is gathered, fashioned, and tested inside campaign organizations. Here, future legislators learn the political ropes, political coalitions and agendas form, and many Americans participate in politics for the first time. Yet while U.S. elections are closely studied, largely absent from this research is an investigation of campaign organizations themselves or the people they comprise. This dissertation examines how campaign teams form, and their impact on U.S. elections and policy. I argue that campaigns are most usefully understood as project networks, a type of temporary organization studied by management scholars. They are time-bounded structures operating on fixed-length projects inside of a larger network of permanent organizations that includes political action committees, consulting firms, and candidate support organizations. Through case studies of eight U.S. House campaigns during the 2020 campaign cycle, I find tensions between temporary campaigns and permanent organizations shape every facet of the campaign, including the distinct ways Democratic and Republican candidates structure their campaigns. At the center of the Democratic campaign network are the Democratic Congressional Campaign Committee and a small number of political action committees. At the heart of the Republican network are large, privately owned consulting shops. How do these tensions shape political campaign team formation? I find that while campaigns report hiring for professional experience, personal networks and ad hoc decision-making drive team formation in political campaigns. One result of this network-driven hiring is that campaigns tend to hire candidates that come from similar groups and backgrounds, to the exclusion of other groups, including women and people of color. To take a closer look at campaign team experience and prior shared work experience, I build a database of all campaign personnel 2003-2016 from publicly available data on campaign expenditures. I find that hiring personnel based on shared prior experience mostly served U.S. House candidates well in elections from 2012-2016, earning them more votes and---at least for Republican candidates---more money. Hiring more experienced personnel, by contrast, was not associated with greater vote share or more money raised except in the case of Republican incumbents hiring Republican consultants. What happens to these teams after the election? Conventional wisdom holds that political hacks make bad policy wonks. I combine the data on campaign personnel with existing data on legislative staff and find the opposite: From 2005-2016, representatives who hired campaign staff into their congressional offices were no less effective as legislators and did no worse at the ballot box in subsequent elections. These findings advance our understanding of U.S. elections, legislative staff, and teams and organizations in politics. They contribute to our understanding of temporary organizations by extending the theory to a new, non-market setting. Most importantly, these findings help us understand an important component of the U.S. political system: the teams that work to determine who governs, and whose work shapes the attention of the electorate and the priorities of legislators.
Content may be subject to copyright.
Copyright
by
Mark Clayton Hand
2021
The Dissertation Committee for Mark Clayton Hand
certifies that this is the approved version of the following dissertation:
No One Can Whistle a Symphony:
Essays on Congressional Campaign Teams and Talent
Committee:
Varun Rai, Supervisor
Erin Lentz
William Spelman
Alison Craig
Sekou Bermiss
No One Can Whistle a Symphony:
Essays on Congressional Campaign Teams and Talent
by
Mark Clayton Hand
DISSERTATION
Presented to the Faculty of the Graduate School of
The University of Texas at Austin
in Partial Fulfillment
of the Requirements
for the Degree of
DOCTOR OF PHILOSOPHY
THE UNIVERSITY OF TEXAS AT AUSTIN
December 2021
For the public, with all its problems and its possibilities.
Acknowledgments
Before beginning a Ph.D. program, I spent six years investing in en-
trepreneurs. I was not a natural. When I started, in fact, I did not know
that equity had any meaning other than fairness. Having previously worked
in community development and social work, I was less concerned with our
return-on-investment than with the teams and leaders we invested in.
I carried this interest with me into graduate school. Having been ex-
posed to the academic study of entrepreneurship during my MBA program, I
came to Texas hunting for research on entrepreneurial behavior and teams in
political science and public policy. To my surprise, I found both disciplines so
committed to the study of institutions and structures that individual decision-
making seemed to have been lost in the shuffle. I hope this dissertation will
contribute to the re-introduction of individual agency to the study of policy
and politics by shedding light on a curiously overlooked set of actors in the
political arena: the campaign staff and consultants that do the work of get-
ting elected officials elected. Who are these people? Why do they do this?
Are they good at it? Given my background financing startup companies, I
initially approached the universe of campaigns as an industry. I now believe
political campaigns to be better understood as a “field of action” than just an
v
industry.1As luck would have it, it also happens to be one with a rich set of
publicly available data.
Academic curiosity and technical feasibility, however, do not tell the
whole story—not for me, at least. I moved from an early career in community
development into social impact investing and finally into studying politics out
of a compulsion to understand and change the systems that shape how we
live, and that crush so many people underfoot in the process. Elected officials
write those rules; and I hope that by illuminating the process by which they
get into the position to write rules for the rest of us, I can make just a bit
more transparent how communities kept out of power might feasibly widen
one path into political participation and, therefore, political power.
“No one can whistle a symphony,” American theologian Halford E.
Luccock is alleged to have said. “It takes a whole orchestra to play it.” I
will not claim this dissertation to be symphonic, but it did require an orches-
tra’s worth of characters for it to come into being, including Dr. Varun Rai
and my other committee members, and their patience with multiple drafts
and consults; Marc Ventresca and Tim Hannigan, who introduced me to so-
ciological understandings of entrepreneurship, organizations, and life; Bryan
Jones, Sean Theriault and the other members of the Policy Agendas Project
in the UT-Austin Government Department who welcomed me as an adjunct
member; Jesse Crosson, Alan Wiseman and Geoff Lorenz, for early feedback;
1See [Fligstein and McAdam, 2012].
vi
Meeta Kothare, Steve Gray and Andrew Caldwell Marquez, the Texas Mc-
Combs students and faculty who treated me as one of their own; Jeremi Suri,
who among other things cast the first vote of confidence in this interdisci-
plinary topic; Vivek Shastry, Cale Reeves, Bryan Frizzelle, Matt Worthington,
Xue Gao, the Clements Center crew and my other public policy classmates
who taught me theory and methods and models and code; Clare Zutz, Vic-
toria Rodriguez and David Springer, who smoothed my path into the PhD
program; Marv Hackert and Elizabeth Korves, my Graduate School support
system; Chet Polson, who walked me across one particular coding bridge I was
ready to burn down; and John Hardman, Matt Scott, Zoe Schlag, and Molly
Alexander, leaders whose examples I strive to follow.
I am fortunate to have too much family to thank individually, but here
are a few: my mother, who consistently encourages my reading and puzzling
and questioning, edits my writing (including this dissertation!), and made
certain I understood early on that being good at school made me no better
and no worse than anyone else; my father, who constantly reminded us that
we’re all on the same team; my brother, from whose self-reliance I drew the
clarity to start a Ph.D.; my sister and fellow pata de perro, whose spirit is a
natural reminder not to take myself too seriously; my paternal grandparents
Albert (“Doc”) Hand and Beverly King Hand, who always encouraged—even
demanded—curiosity; to my maternal grandparents Dorothy (“D.D.”) and
Bobby Aillet, who knew and lived out that “it is always about the people;”
and to my adopted Webb family, for their fierce and loyal support.
vii
I would also like to thank the candidates and staffers who offered me
their time and trust in the middle of an intense, pandemic-ridden 2020 elec-
tion season without even the promise of seeing their names in this dissertation;
and the political professionals, including those from the Truman Project for
National Security, who helped me wrangle them into interviews: JS, LH, KR,
MG, SvM, KJ, and AT. I am grateful, too, to those who have funded my edu-
cation along the way, including my parents, the Skoll Foundation, the Ingram
family, Joanne Fleming-Hayes, the Rotary Foundation, the Harrington family,
the Dirksen Congressional Center, and the UT-Austin Strauss Center; to Josh
Basseches, one of many whose randomly well-timed pieces of advice (“Don’t
lose sleep about that part...yet.”) smoothed the rough edges of this process;
to all those whose publicly available data and code this is built upon, includ-
ing the Federal Election Commission, Adam Bonica, Ballotpedia.org, and the
#Rstats community; the Provident1898 crew, which welcomed me in all my
end-of-dissertation mania; and to the friends I have made across multiple con-
tinents whose memories remind me of my obligation to use whatever position
I have to help push power out to the powerless.
Lastly, I would like to thank Dr. Ashlyn (“Lynnie”) Saylors Webb
Hand, who has heard more articulations of this text than any human should
be subjected to and who never—not once—allowed me to doubt my ability to
get it over the line. I have been on many, many teams in my life and career.
There is no team I am more grateful to be on than yours.
viii
No One Can Whistle a Symphony:
Essays on Congressional Campaign Teams and Talent
Publication No.
Mark Clayton Hand, Ph.D.
The University of Texas at Austin, 2021
Supervisor: Varun Rai
Political power is gathered, fashioned, and tested in campaign organiza-
tions. Here, future legislators learn the political ropes, political coalitions and
agendas form, and many Americans participate in politics for the first time.
Yet while U.S. elections are closely studied, largely absent from this research
is an investigation of campaign organizations themselves or the people they
comprise. This dissertation examines how campaign teams form, and their
impact on U.S. elections and policy.
I argue that campaigns are most usefully understood as project net-
works, a type of temporary organization studied by management scholars.
They are time-bounded structures operating on fixed-length projects inside
of a larger network of permanent organizations that includes political action
committees, consulting firms, and candidate support organizations. Through
ix
case studies of eight U.S. House campaigns during the 2020 campaign cycle,
I find tensions between temporary campaigns and permanent organizations
shape every facet of the campaign, including the distinct ways Democratic
and Republican candidates structure their campaigns. At the center of the
Democratic campaign network are the Democratic Congressional Campaign
Committee and a small number of political action committees. At the heart
of the Republican network are large, privately owned consulting shops.
How do these tensions shape political campaign team formation? I
find that while campaigns report hiring for professional experience, personal
networks and ad hoc decision-making drive team formation in political cam-
paigns. One result of this network-driven hiring is that campaigns tend to hire
candidates that come from similar groups and backgrounds, to the exclusion
of other groups, including women and people of color.
To take a closer look at campaign team experience and prior shared
work experience, I build a database of all campaign personnel 2003-2016 from
publicly available data on campaign expenditures. I find that hiring personnel
based on shared prior experience mostly served U.S. House candidates well in
elections from 2012-2016, earning them more votes and—at least for Republi-
can candidates—more money. Hiring more experienced personnel, by contrast,
was not associated with greater vote share or more money raised except in the
case of Republican incumbents hiring Republican consultants.
What happens to these teams after the election? Conventional wisdom
holds that political hacks make bad policy wonks. I combine the data on
x
campaign personnel with existing data on legislative staff and find that from
2005-2016, representatives who hired campaign staff into their congressional
offices were no less effective as legislators and did no worse at the ballot box
in subsequent elections.
These findings advance our understanding of U.S. elections, legislative
staff, and teams and organizations in politics. They contribute to our un-
derstanding of temporary organizations by extending the theory to a new,
non-market setting. Most importantly, these findings help us understand an
important component of the U.S. political system: the teams that work to
determine who governs, and whose work shapes the attention of the electorate
and the priorities of legislators.
xi
Table of Contents
Abstract ix
List of Tables xvi
List of Figures xvii
Chapter 1. Introduction 1
1.1 A note on definitions . . . . . . . . . . . . . . . . . . . . . . . 6
Chapter 2. Literature Review 8
2.1 Political Science Research on Campaigns, Teams, and Legisla-
tiveEectiveness ......................... 8
2.1.1 Campaign Staff and Campaign Organizations . . . . . . 9
2.1.2 Campaigns and Elections . . . . . . . . . . . . . . . . . 12
2.1.3 Teams and Staff in Politics . . . . . . . . . . . . . . . . 12
2.1.4 Legislative Effectiveness . . . . . . . . . . . . . . . . . . 15
2.2 Temporary Organizations in Management Theory . . . . . . . 15
2.3 Political Campaigns as Temporary Organizations . . . . . . . . 18
2.3.1 The Driving Characteristics of Political Campaigns . . . 19
2.3.1.1 Impermanence . . . . . . . . . . . . . . . . . . 19
2.3.1.2 Win/Loss Dynamic . . . . . . . . . . . . . . . . 21
2.3.1.3 Uncertainty . . . . . . . . . . . . . . . . . . . . 21
2.3.1.4 Embeddedness . . . . . . . . . . . . . . . . . . 23
2.3.2 Organizational Processes in Political Campaigns . . . . 24
2.3.2.1 Team Formation . . . . . . . . . . . . . . . . . 25
2.3.2.2 Structuration . . . . . . . . . . . . . . . . . . . 26
2.3.2.3 Strategic Decision-Making . . . . . . . . . . . . 28
2.3.2.4 Performance Management . . . . . . . . . . . . 29
2.3.2.5 Organizational Learning . . . . . . . . . . . . . 30
xii
Chapter 3. The Impermanent Campaign: Team Formation in
the Project Networks of Political Campaigns 32
3.1 Abstract .............................. 32
3.2 Introduction............................ 33
3.3 Theory............................... 34
3.3.1 Temporary Organizations and Project Networks . . . . 34
3.3.2 Campaigns as Temporary Project Networks . . . . . . . 36
3.3.3 Team Formation in Political Campaigns . . . . . . . . . 37
3.4 Data and Methodology . . . . . . . . . . . . . . . . . . . . . . 39
3.5 Results............................... 42
3.5.1 The Zeroth Hire: The Candidate . . . . . . . . . . . . . 43
3.5.2 The First Hire . . . . . . . . . . . . . . . . . . . . . . . 44
3.5.3 Assembling the Team: The Methods . . . . . . . . . . . 47
3.5.4 Assembling the Team: The Motivations . . . . . . . . . 51
3.5.5 The Shifting Demographics of Campaign Teams . . . . . 55
3.5.6 Motivations on the Other Side of the Table . . . . . . . 57
3.5.7 Supply and Demand for Labor in Political Campaigns . 60
3.6 Discussion ............................. 62
3.6.1 The Structure of Political Campaign Networks . . . . . 62
3.6.2 The Structure of Political Campaign Teams . . . . . . . 66
3.6.3 Team Formation in Project Networks . . . . . . . . . . 70
3.6.4 The Five Tensions of Project Networks . . . . . . . . . . 71
3.6.5 Conclusions and Future Work . . . . . . . . . . . . . . . 76
Chapter 4. The Devil You Know: Team Experience, Team Fa-
miliarity, and Performance in U.S. House Campaigns
from 2012-2016 80
4.1 Abstract .............................. 80
4.2 Introduction............................ 81
4.2.1 Hypotheses......................... 85
4.3 Data ................................ 87
4.3.1 Dependent Variables . . . . . . . . . . . . . . . . . . . . 88
4.3.2 Constructing Study Variables . . . . . . . . . . . . . . . 88
xiii
4.4 Methods and Results . . . . . . . . . . . . . . . . . . . . . . . 92
4.4.1 Vote Share, Team Experience, and Team Familiarity . . 94
4.4.2 Fundraising, Team Experience, and Team Familiarity . . 99
4.5 Discussion ............................. 104
4.5.1 Team Experience . . . . . . . . . . . . . . . . . . . . . . 104
4.5.2 Team Familiarity . . . . . . . . . . . . . . . . . . . . . . 105
4.5.3 Moving Up the Ladder . . . . . . . . . . . . . . . . . . . 106
4.5.4 The Devil You Know . . . . . . . . . . . . . . . . . . . 108
4.5.5 Conclusions, Contributions, and Future Work . . . . . . 110
Chapter 5. Hacks and Wonks: The Effects of Hiring Campaign
Staff into Congressional Offices from 2005-2016 114
5.1 Abstract .............................. 114
5.2 Introduction............................ 115
5.3 Theory and Hypotheses . . . . . . . . . . . . . . . . . . . . . . 116
5.3.1 Legislative Staff in the U.S. Congress . . . . . . . . . . . 116
5.3.2 Hypotheses......................... 119
5.4 Data ................................ 120
5.4.1 Congressional Staff Data . . . . . . . . . . . . . . . . . 120
5.4.2 Building a Measure of Staff Crossover . . . . . . . . . . 120
5.4.3 Legislative Effectiveness . . . . . . . . . . . . . . . . . . 121
5.4.4 Electoral Outcomes . . . . . . . . . . . . . . . . . . . . 123
5.5 Methods and Results . . . . . . . . . . . . . . . . . . . . . . . 124
5.5.1 Staff Crossover and Legislative Effectiveness . . . . . . . 124
5.5.2 Staff Crossover and Electoral Outcomes . . . . . . . . . 126
5.6 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . 126
Chapter 6. Conclusion 133
6.1 Findings and Contributions . . . . . . . . . . . . . . . . . . . . 134
6.1.1 Contributions to American Politics . . . . . . . . . . . . 138
6.1.2 Contributions to Public Policy Literature . . . . . . . . 139
6.1.3 Contributions to Management Theory . . . . . . . . . . 140
6.2 Future Research Directions . . . . . . . . . . . . . . . . . . . . 141
6.3 Practical Observations and Recommendations . . . . . . . . . 142
xiv
Appendices 148
Appendix A. Candidate and Personnel Interview Template 149
Appendix B. Building a Database of Campaign Personnel 155
Appendix C. Additional Models: Chapter 4 159
C.1 Fundraising and Familiarity by Party . . . . . . . . . . . . . . 160
C.2 Small-Dollar Fundraising Models . . . . . . . . . . . . . . . . . 162
Appendix D. Additional Models: Chapter 5 163
Bibliography 166
xv
List of Tables
4.1 Summary Statistics: Team Experience and Familiarity . . . . 92
4.2 Vote Share, Team Experience, and Team Familiarity . . . . . 98
4.3 Vote Share, Team Experience, and Team Familiarity by Party
andIncumbency.......................... 100
4.4 Fundraising, Team Experience, and Team Familiarity . . . . . 102
4.5 Funds Raised, Team Experience, and Team Familiarity by Party
andIncumbency.......................... 103
5.1 Summary Statistics: Staff Experience and Crossover . . . . . . 121
5.2 Staff Crossover and Legislative Effectiveness . . . . . . . . . . 127
5.3 Staff Crossover and Subsequent Elections . . . . . . . . . . . . 128
C.1 Funds Raised, Team Experience, and Team Familiarity by Party
andIncumbency(2) ....................... 160
C.2 Funds Raised, Team Experience, and Team Familiarity by Party
andIncumbency(3) ....................... 161
C.3 Examining Number of Donors and Vote Share . . . . . . . . . 162
D.1 Staff Crossover and Legislative Effectiveness . . . . . . . . . . 164
D.2 Staff Crossover and Subsequent Elections . . . . . . . . . . . . 165
xvi
List of Figures
1.1 Chapters 3-5 in the Life of a Congressional Campaign . . . . . 6
3.1 Breakdown of Cases by Party, Gender, and Other Characteristics 41
3.2 How Two Parties Manage the Characteristics and Tensions of
Campaigns............................. 64
3.3 Stylized Structure of a Democratic Campaign Team . . . . . . 68
3.4 Stylized Structure of a Republican “Strong Consultant” Cam-
paignTeam ............................ 69
3.5 Tensions in Democratic and Republican Project Networks . . 72
4.1 Team Experience, Vote Share, and Fundraising . . . . . . . . . 91
4.2 Team Familiarity, Vote Share, and Fundraising . . . . . . . . . 92
5.1 Staff Crossover and Legislative Effectiveness . . . . . . . . . . 122
5.2 Staff Crossover and Vote Share . . . . . . . . . . . . . . . . . 124
xvii
Chapter 1
Introduction
“No one can whistle a symphony. It takes a whole orchestra to play
it.” - Halford Luccock, date unknown
“There is something wrong with these people.” - A campaign staffer
in 2020
Campaign teams are a critical component of the U.S. political system.
They where political power is gathered, fashioned and tested. They are where
future legislators learn the political ropes, where many Americans’ active par-
ticipation in politics begins, and where political coalitions and agendas are
shaped and reshaped. This project asks three sets of questions: (1) How do
political campaign teams form? (2) Do the characteristics of those campaign
personnel have an impact on campaigns’ ability to raise funds and win elec-
tions? and (3) When campaign staff are hired into legislative offices, what
impact do they have on representatives’ legislative effectiveness and subse-
quent electoral success? In answering these questions, this project expands
scholarly understanding of the role of staff and teams in the political process
and provides policymakers with actionable, evidence-based guidance on hiring
in political campaigns and legislative offices.
1
Despite their importance, campaign teams and personnel are a set of
political actors that academic researchers, the general public, and even political
operatives themselves know a puzzlingly small amount about. As one veteran
political consultant told me:
You don’t really get a peek into this world unless you’re Karl Rove
or David Axelrod...you typically don’t know who the hell these
people are or what they do. I mean, my mom still doesn’t. To this
day, she’ll ask questions like ‘Did you have something to do with
that?’ ‘Yes, Mom.’1
In the 2020 election cycle, some 14 billion dollars flowed to and through
this opaque network of full-time staff and consultants, an amount larger than
the annual budgets twelve U.S. States [Goldmacher, 2020, Foundation, 2021],
and a record likely to stand only until the next presidential election. In elec-
tions for the U.S. House of Representatives, the focus of this dissertation,
serious candidates hire anywhere from 10-150 staff responsible for raising and
spending an average of 1.6 million over the course of 12-15 months, an amount
that climbs each cycle [Thurber, 2018]. Disbursement data from the Federal
Election Commission suggests that in a typical House race, 20-30% of those
dollars will be spent on staff or consultants. In the 2018 election cycle, for
1Rove and Axelrod managed campaigns for George W. Bush and Barack Obama, respec-
tively; together, they recorded a class on campaign strategy in 2018 for the online learning
company Masterclass.
2
example, out of the 1.63 billion spent by House candidates, 414 million was
spent on personnel.2
Still, understanding political campaign organizations is important not
primarily because of the money they spend. Rather, campaign organizations
merit study by virtue of the central role they play in the political process. First,
these teams are at least partially responsible for getting elected officials elected.
Their mistakes and successes help determine who represents Americans in
Congress and who makes the laws that shape our lives. Second, candidates are
humans, too, affected and influenced by the people around them. This means
campaign personnel serve not just agents executing on campaign strategy;
they are organizations that mediate the relationship between candidates and
voters, shaping the beliefs, agendas and preferences of candidates that hope to
become legislators. As James Q. Wilson put it, organizations “are not neutral
devices...they are social systems that modify as well as express preferences”
[Wilson, 1974]. Lastly, campaigns are a common entry point into professional
politics, and it is worth considering who can, and who cannot, access this path
and the influence that comes with it.
The chapters to follow include a review of relevant literature, three sub-
2In their 2005 book Freakonomics, Levitt and Dubner claimed that Americans spent
about as much on political campaigns as on bubble gum. Perhaps unintentionally, the
authors repeated an factually incorrect talking point of Mitch McConnell’s: 2.2 billion
was spent on the 2004 campaign, twice what the authors claimed. More appropriate to
the business of politics, perhaps, is John Boehner’s (also incorrect) comparison of campaign
spending to spending on antacids [Lardner, 1997, Levitt and Dubner, 2014, Greenberg,
2015, opensecrets.org, 2021].
3
stantive chapters designed as standalone papers, and a conclusion. In Chapter
2, I describe the various strands of academic literature I draw from in order
to understand campaign personnel and organizations. First, I discuss existing
political science work on campaigns and elections, to which this dissertation
contributes most explicitly. That includes work on legislative effectiveness,
political parties and agenda-setting. I also lay out prior management liter-
ature on temporary organizations, organizational networks, human resource
management, and entrepreneurial team formation. All of these carry insights
that I introduce (and sometimes re-introduce) to political science and public
policy literature.
Chapter 3, the first of the three paper-style chapters, is a qualitative
comparative case study of eight congressional campaigns during the 2020 cycle.
Drawing primarily from management literature on temporary organizations,
I argue that political campaigns are best understood as project networks,
one type of temporary organization. The paper describes and explains po-
litical campaign team formation: How and why campaigns hire the staff and
consultants they do, how and why those individuals find and join particular
campaigns. I find that while campaigns talk about political experience as a
driver of hiring, personal networks dominate team formation processes. This
network-driven hiring happens inside of two partisan networks that are distinct
and shaped quite differently, based on the action of a few key actors in each
party: The Democratic Congressional Campaign Committee and Emily’s List
on one side, and large, private, for-profit consulting shops on the Republican
4
side.
In Chapter 4 I examine whether campaign personnel have an impact
on election outcomes. It begins with two characteristics of personnel that
have proven determinative in other industries: (1) team experience and (2)
team familiarity, or how much experience a team has working together in the
past. I build measures of both from publicly available data on U.S. House
elections from 2012-2016, then use multilinear regression to examine their as-
sociations with a candidate’s ability to raise campaign funds and earn vote
share. Contrary to the expectations of management scholarship and to vet-
eran political operatives, team experience is not associated with greater vote
share or fundraising ability, except in one instance: Republican incumbents
with more experienced teams of consultants earn more votes. Team familiar-
ity, on the other hand, is associated with greater vote share for candidates
from both parties, and with higher fundraising totals.
Chapter 5 tests the truism among political operatives that campaign
staff (sometimes called political “hacks”) make for poor legislative staff (known
as policy “wonks”). This lay theory lines up with management literature that
would describe campaigns and legislatures as different task environments that
should be costly to switch between. Unexpectedly, I find that from 2006-2016,
legislators whose offices included higher percentages of campaign staff were no
less effective at getting legislation passed, and did no better or worse at the
ballot in subsequent elections.
Each of these three chapters covers one period in or after the campaign;
5
Figure 1.1 shows where in the timeline of a campaign each chapter falls. In
Chapter 6, I conclude. I discuss what these results mean for scholarship on
political campaigns and elections and describe future work on campaign orga-
nizations. I also discuss the implications of these results for candidates, staff
and consultants.
Figure 1.1: Chapters 3-5 in the Life of a Congressional Campaign
1.1 A note on definitions
In interviews with campaign staff and consultants, it became clear that
the word “consultant” had three distinct meanings:
A general consultant (sometimes just called “the consultant”) who over-
sees a campaign
A consultant dedicated to a specific task, e.g. print mailing or fundrais-
ing
Someone paid as a consultant, but working full-time for the candidate
In fact, in one exploratory interview a state-level candidate could not
recall whether one member of their campaign team had been a consultant or
6
a full-time staffer. To the candidate, the distinction didn’t matter. What
mattered was that the staffer was working for that candidate and only that
candidate. As a result, they remembered him as operating more like a full-
time staffer than a general consultant or a task-specific consultant that might
be working for multiple campaigns in one cycle.
In light of this and in hopes of preventing unnecessary confusion, I use
the following definitions throughout the remainder of this dissertation:
Consultant refers to either a general consultant or task-specific consul-
tant, which often work for multiple campaigns
Staff refers to full-time or part-time staff that are working for a single
candidate as part of that candidate’s team
Personnel and Team include anyone paid to work for the campaign,
whether as a consultant or staff, even where consultants are not always
considered part of a candidate’s core team
Not discussed here—and also not raised by more than a small handful
of people I interviewed—are volunteers. In public, political campaigns speak
regularly about the importance of volunteers. They may be important, and
are certainly worthy of future study. But their absence in these conversa-
tions points to a clear divide between volunteers and those personnel whom
candidates consider part of the campaign organization.
7
Chapter 2
Literature Review
“[It is] very hard to find good written material on [hiring]. The net-
work for staff is very word of mouth still. People tend to work to-
gether on a campaign and then pull one another into future races.”
- Democratic campaign manager Anna Caprara by tweet, April
2019
This dissertation comes in two disciplinary parts. The first part (Chap-
ter 3) is a qualitative paper, borrowing from and designed to speak to manage-
ment scholars. It examines political phenomena from an organizational lens.
The second part (Chapters 4 and 5) are quantitative papers housed in political
science literature and written with political scientists in mind. In Chapter 6,
I conclude by speaking more broadly to policy scholars and the public policy
literature.
2.1 Political Science Research on Campaigns, Teams,
and Legislative Effectiveness
In the sections below, I begin with what political scientists know about
campaign staff and campaign organizations. I then situate that discussion
8
inside of the broader political science literature before turning to the man-
agement literature on temporary organizations that sets the foundation for
Chapter 3.
2.1.1 Campaign Staff and Campaign Organizations
In 1978, political scientist Xandra Kayden published Campaign Orga-
nizations, a study of three statewide political campaigns in Tennessee, Mas-
sachusetts and Indiana [Kayden, 1978]. For scholars of organizations, the
1970s and early 1980s were a particularly rich period that saw the publication
of foundational texts [DiMaggio and Powell, 1983, Meyer and Rowan, 1977]
and the formation of these scholars’ own organization inside the Academy of
Management [OMT, 2021].
Unfortunately for Xandra Kayden, political scientists and public policy
scholars did not follow in the organizational footsteps of their business school
peers. As the field of organizational theory in management grew, political
scientists and public policy scholars, by and large, built fields focused on the
study of institutions, interest groups, parties and the state [Skocpol et al.,
1985, Steinmo et al., 1992, Aldrich, 1995, Koelble, 1995]. Even the study of
social movements, built upon the foundations of organizational theory, was
driven primarily by sociologists, rather than scholars of politics and policy
(see [Benford and Snow, 2000]). And the most well-cited work on political
organizations uses the phrase political organizations narrowly to mean “vol-
untary organizations” such as the American Association of Retired Persons or
9
the National Rifle Association, rather than as an examination of organizations
and organizing throughout the political system [Wilson, 1974].
Kayden was not completely alone in her interest in how political cam-
paigns operate. Ten years earlier, Lamb and Smith had argued Lyndon B.
Johnson owed his success in the 1964 presidential election to his incrementalist
campaign strategy [Lamb and Smith, 1968]. His opponent Barry Goldwater,
by contrast, had an impressively organized campaign implementing a strategy
that proved as inflexible as it was comprehensive. In 1976, Robert Agranoff
published both a textbook on campaign management and a second edition of
essays on campaign management [Agranoff, 1976, Stillman, 1978]. These and
other texts from the late 1970s, wrote one reviewer at the time, reflected a
real-world shift from powerful political parties in the wake of Watergate to
“candidate-centric campaigns,” making the study of campaign management
strategies, procedures and experts all the more necessary [Stillman, 1978].
Necessary or not, little work on the internal management of political cam-
paigns followed.
In the years following, U.S. political parties continued their decline
[Wattenberg, 2009], and the study of elections grew, building up an under-
standing of candidates [Ansolabehere and Snyder Jr, 2002], campaign finance
[Ansolabehere et al., 2003], political communication [McNair, 2011] and the
electorate [Mueller, 2003]. Only recently, however, have scholars turned their
attention back to campaign organizations themselves, with a focus on cam-
paign consultants and their relationships to campaigns and parties [Sheingate,
10
2016, Medvic, 2005, Dulio, 2006].
What we do know about the organization of modern political cam-
paigns is mostly a function of popular writing and the descriptive work of
Stephen Medvic, whose work has described the basic organizational structures
of modern political campaigns [Medvic, 2005, Medvic, 2018] and the impact of
campaign technology on their management and organization [Medvic, 2011].
At the heart of campaigns, according to Medvic, is a campaign committee
made up of volunteer advisors, with larger campaigns adding other commit-
tees (e.g. finance, outreach) and sometimes a separate “kitchen cabinet” of
close personal advisors [Medvic, 2005]. All but the most local campaigns have
a paid campaign manager. Then, depending on the size of the campaign, a
candidate may also hire other functional directors in charge of finance, po-
litical, field, communications, press, and research; or consultants dedicated
to strategy, polling, media (advertising), mail, and fundraising. Starting in
the late 2000s, some campaigns also began hiring specific digital directors and
digital consultants [Kreiss, 2012, Kreiss and McGregor, 2018].
As Medvic points out, however, we know little about how campaign
organizations are structured and less about how they operate [Medvic, 2011].
And even these descriptive observations are broad and general enough that
we know little about how they vary across campaigns. Perhaps, as Xandra
Kayden argued, “we pay little attention to the internal politics of campaigns
because we are so concerned about their outcomes” [Kayden, 1978].
11
2.1.2 Campaigns and Elections
Few electoral contests are studied more closely than those for the U.S.
Congress. A robust body of research on political campaigns has considered
the characteristics of individual politicians, their constituencies, and the sys-
tems that contribute to successful campaigns. Upwards of three quarters of the
variation in electoral outcomes in the U.S. House of Representatives can be pre-
dicted from just a few factors: which party holds a seat, whether an incumbent
is running, and how a district voted in the most recent presidential election
[Jacobson, 2015]. But it is the remaining variation that determines who wins
close elections and ultimately which party controls the legislature. Research
has sought to explain this variation by looking at candidate characteristics,
including a candidate’s ideological positioning relative to the preferences of
the district and their opponents [Ansolabehere et al., 2001]; the support a
candidate receive from extended party networks [Desmarais et al., 2015]; and
candidate valence, a dual measure of the character of the candidate and the
effectiveness of their campaign [Stone and Simas, 2010].
2.1.3 Teams and Staff in Politics
Largely absent from this conversation is an examination of the indi-
viduals who work to get those politicians elected. There is growing academic
interest, however, in the teams they form after election day. Scholars have
shown that staff have an impact on legislators’ effectiveness and policy po-
sitions [Crosson et al., 2018, Hertel-Fernandez et al., 2019]. It is staff that
12
responds to constituents, drafts legislation, schedules the legislator’s time, in-
teracts with media, and helps determines the issues the legislator will focus on.
They are, in short, critical actors in the creation of public policy [Romzek and
Utter, 1997, Weissert and Weissert, 2000]. And yet only recently have political
scientists begun to demonstrate just how important those staff members are to
democratic representation and legislative effectiveness [Hertel-Fernandez et al.,
2019, McCrain, 2018, Montgomery and Nyhan, 2017].
Political scientists have tended to understand legislative staff as im-
perfect extensions of the legislator whose behavior could be modeled with a
principal-agent approach [Romzek and Utter, 1997, Rosenthal and Bell, 2003].
This has not always been the case: In his 1971 book Essence of Decision, Gra-
ham Allison examined policymaking through multiple lenses, previewing the
way that management scholars would later come to understand organizations
as variably rational, natural or open systems [Allison, 1971, Scott and Davis,
2015].
But as Bell and Rosenthal argue and others have demonstrated, “on a
day-to-day basis, there is nothing that members do in the modern legislature
that is not affected by their staffs” [Montgomery and Nyhan, 2017, Rosenthal
and Bell, 2003, Rundquist et al., 1992]. A recent wave of scholarship takes
these actors seriously, finding, for example, that Representatives hire staffers
in part to exploit their professional networks [Burgat, 2020]; and that (at the
state level) greater legislative professionalization leads to less racial discrimina-
tion [Landgrave and Weller, 2020]. Much research on legislative staff, however,
13
has focused on the problems these staff members face and create. Congres-
sional staff capacity is on the decline [LaPira et al., 2020], and one study argues
that staff collect and present politically biased information to their represen-
tatives [Furnas, 2019]. Staff are not all treated equally, either: female staffers
are systematically underpaid [McCrain and Palmer, 2019, Calcagno and Mont-
gomery, 2020], and minority staff are disproportionately placed in constituent
service roles [Ziniel, 2020]. In Chapter 5, this dissertation extends that nascent
work by examining campaign staffers that cross over into legislative offices. It
examines their effect on policymakers’ effectiveness and electoral success, two
dimensions of performance studied by political scientists and policy scholars.
Staff on political campaigns have received substantially less attention,
and the existing scholarship has focused on the paid consultants who work
for campaigns. In Building a Business of Politics, Adam Sheingate traces
the development of political consulting into an industry that determines the
strategic directions of campaigns [Sheingate, 2016]. Candidates who rely on
professional political consultants raise more money [Herrnson, 1992] and are
more willing to engage in negative campaigning on issues, if not personal
attacks [Francia and Herrnson, 2007]. The world of campaign consultants
is highly interconnected, with campaigns that share consultants engaging in
similar patterns of risk-taking and issue-ownership in their campaigns [Nyhan
and Montgomery, 2015].
14
2.1.4 Legislative Effectiveness
Introduced by Volden and Wiseman in a 2009 working paper and then
codified in a 2014 book and 2013 article, Legislative Effectiveness Scores are a
measure of a legislator’s ability to introduce substantive legislation and shep-
herd it through the legislative process [Volden and Wiseman, 2009, Volden
et al., 2013, Volden and Wiseman, 2014]. Scholars have since used these scores
to underscore structural differences between the House and Senate [Volden and
Wiseman, 2018]; to measure the importance of committee chairs [Lewallen,
2020]; to quantify the impact of social connections on legislative effectiveness
[Battaglini et al., 2020]; and to demonstrate that legislators anticipating com-
petitive primary elections are less effective legislators [Barber and Schmidt,
2019]. Other scholars have built alternate measures of effectiveness [Casas
et al., 2020]. In Chapter 5, I examine the correlation between these LES scores
and the percentage of their staff that a legislator hired from the campaign trail.
2.2 Temporary Organizations in Management Theory
Prior to publishing Campaign Organizations in 1978, political scientist
Xandra Kayden identified three major factors she described as peculiar to po-
litical campaigns: the uncertainty derived from the difficulty of measuring the
contribution of campaign activities to campaign outcomes; the lack of a hier-
archical structure resulting from the limited organizational life of campaigns;
and the win/lose payoff of campaigns [Kayden, 1973].
The “limited organizational life” that Kayden identified in political
15
campaigns is not unique to campaigns.1Far from it. Beginning in 1995, or-
ganizational scholars began to develop a theory of temporary organizations
as distinct from organizations without a determined end point [Lundin and
oderholm, 1995, Burke and Morley, 2016]. That research grew into a wider
consideration of temporary organizing that takes place in both temporary or-
ganizations and “permanent” organizations with an indeterminate endpoint
[Bakker et al., 2016]. The prevalence of such organizations is increasing [Ja-
cobsson et al., 2015] in response to the decreased organizational transaction
costs associated with the rapid development of communication technologies
starting in the late 20th and early 21st centuries. This has led some project
scholars to go so far as to herald the coming “projectification of everything”
[Jensen et al., 2016]. Permanent organizations may prove more resilient to
such predictions, but the expansion of temporary organizing in permanent
contexts will continue [Lundin et al., 2017].
Temporary organizations (or TOs), according to one recent review,
comprise four forms, delineated primarily according to their relationship to
permanent organizations. Intra-organizational TOs are temporary projects
internal to a permanent organization, such as a change management initia-
tive. Inter-organizational TOs, such as a multi-firm construction project, are
temporary agreements among multiple permanent organizations. Project-based
organizations are permanent organizations whose work is regularly organized
1Also not unique to political campaigns are Uncertainty [Knight, 1921] or win/loss dy-
namics [Ely, 1900], dimensions which also merit future study.
16
around temporary projects; advertising agencies and consulting firms often
fit this description. And project-based firms (or enterprises) are fully tempo-
rary organizations such as theater productions that form and disband with
the creation and completion of the project [Burke and Morley, 2016]. Another
view understands project work as occurring in project-supported organiza-
tions (PSOs, roughly mapping onto intra-organizational TOs), project-based
organizations (PBOs), or project networks (PNWs, which could include both
inter-organizational forms and project-based firms) [Lundin et al., 2015].
This categorization of temporary organizations is challenged by another
2016 review on temporary organizing, a shift in focus from whether and how
organizations are categorized as temporary to whether the process of orga-
nizing is different when some organizational goals, structures, tasks, or work
force are time-limited [Bakker et al., 2016]. To the degree that Burke and
Morley’s “project-based firms” dissolve completely, these scholars would de-
scribe these organizations as truly temporary, or what March called disposable
[March, 1995]. Otherwise they, like the other temporary organizations listed
by Burke and Morley, are in practice “semi-temporary,” with relationships and
structures outlasting the temporary organizations [Bakker et al., 2016].
Bakker et al.’s distinction between temporary organizations and tem-
porary organizing allows for a shift in focus from whether the organizations
themselves are categorized as more or less temporary, to whether the internal
processes they generate are time-limited. This shift can have two results: first,
the expansion of the study of temporary organizations and projects to include
17
everything with any temporary component [Lundin et al., 2015]; and second,
the expansion of focus from the individual project to the context in which a
project emerges, including its norms and values [Engwall, 2003].
The developments in research on project networks and project ecologies
reflect this second shift. DeFillippi and Sydow, adapting a 1995 framework
by Hellgren and Stjernberg’s, argue that project networks have three defin-
ing characteristics: “(1) a set of relations, where no single actor may act as
a legitimate authority for the network as a whole; (2) a situation where the
network is open in the sense that there are no definite criteria by which the
boundary of the network may be identified and controlled; and (3) an envi-
ronment where the network is temporally limited, dynamically changing, and
(partially) reconstructed from one project to the next” [Hellgren and Stjern-
berg, 1995, DeFillippi and Sydow, 2016].
2.3 Political Campaigns as Temporary Organizations
In the sections below, and as further background to Chapter 3, I gather
together those authors’ collective assertions into four driving characteristics of
political campaigns: impermanence2, the win/loss dynamic, uncertainty, and
embeddedness. I then identify how five organizational processes should be
shaped by those four drivers: team formation (the topic of Chapter 3), struc-
turation, strategic decision-making, performance management, and learning.
2...in place of “temporariness,” which is unfortunately for this dissertation not yet an
English word.
18
2.3.1 The Driving Characteristics of Political Campaigns
2.3.1.1 Impermanence
Planned organizational impermanence brings drawbacks and advan-
tages, responses to both, and open questions. First, the drawbacks: temporary
organizations are intense and stressful [Burke and Morley, 2016]. They place
high demands on individuals and can become total institutions that isolate
them from the outside world. Their impermanence generates extra ambigu-
ity and uncertainty. As a result, temporary organization members are more
likely to be younger, and therefore temporary teams bring less relevant his-
torical knowledge to a task than more experienced teams might. They can
innovate but are unlikely to metabolize knowledge into more permanent or-
ganizations. Even when they do, that knowledge may not be generalizable,
having been built up for a single task rather than routinized for others’ benefit.
As a result, temporary organizations often suffer from “organizational amne-
sia.” They often depend on parent organizations for resources, and struggle to
develop, evaluate, or compensate their members.
Prior to most of this formal work on temporary organizations, Xan-
dra Kayden observed and predicted some of these effects of impermanence
in political campaigns in the 1970s. She observed the relative intensity and
informality of political campaigns relative to the hierarchy and bureaucracy
of party organizations, as well as the tendency to orient work around specific
tasks [Kayden, 1978]. She saw that impermanence lowered managers’ ability
to sanction or fire staff; and that to the extent that staff felt bound to the
19
campaign, it would be “not out of habit, identity, or money, but because of
prior relationships” [Kayden, 1978].
Why, then, would one ever build a temporary organization? They can
be more dynamic, creative, and innovative. Isolated from the bureaucracy and
politics of permanent organizations, they can be less combative, hyper-efficient
and high-performing [Burke and Morley, 2016]. Their impermanence, and the
need to get a task done, can prevent the development of a “hegemonic view”
of a problem.
In response to these issues and opportunities, leaders of temporary
organizations tend—rightly or wrongly—to be task-focused and participatory
in their decision-making [Burke and Morley, 2016]. They also tend to be more
successful if their leadership style is relationship-oriented. They make strategic
hiring choices, as well, recruiting for a diversity of task-specific knowledge—
though others found that they hired for their interpersonal skills [Morley and
Silver, 1977]. And they respond to the problem of learning by launching series
of similar projects to take advantage of “economics of repetition” [Burke and
Morley, 2016].
This is some of what we know about the impact of impermanence on or-
ganizations, but there is much that remains to be explored [Burke and Morley,
2016]. Internally, we know little about temporary team formation in tempo-
rary organizations, or how mental models are contested or deployed. As for
the individuals who make up these temporary teams, there is little research on
membership in multiple temporary organizations, the impact of individuals’
20
mobility across project teams, how they build career capital, or how divergent
understandings of time might impact team dynamics.
2.3.1.2 Win/Loss Dynamic
If impermanence is the defining feature of political campaigns barreling
toward election day, the second is the nature of the contest: In most elections,
there is no second place [Kayden, 1973]. This win/loss dynamic, the second
of Kayden’s three defining factors of the political campaign, has a particular
effect on campaigns at the end of the race. She argues that campaigns will
behave in a more strategic manner at the outset but then, as the end of the
campaign approaches, begin to grasp at whatever advantage they can. This
final scramble is driven by uncertainty about what efforts will impact voters.
It stands to reason that this kind of zero-sum conflict, which has not yet
been addressed in temporary organizations literature, has a number of other
effects on the campaign. Those may include how candidates choose races to
enter, how staff choose races to join, which races get attention and resources
from parties and other organizations in the wider project network.
2.3.1.3 Uncertainty
The third of Kayden’s driving characteristics of campaigns is uncer-
tainty [Kayden, 1973]. For Kayden, that uncertainty is specific to the inability
of campaigns to know which of their activities—if any—will move sympathetic
voters to the polls or persuadable voters toward the candidate [Kayden, 1978].
21
This lack of feedback leads campaigns to isolate themselves from external cir-
cumstances and events, including the activities of opposing campaigns. It
also leads campaigns to divide into two camps. One camp is responsible for
tasks that lend themselves to measurement, and another for tasks that don’t
show a visible return. Staff sort themselves accordingly. The tasks without a
visible return are often riskier and only possible in the absence of negative feed-
back that comes with uncertainty. Uncertainty leads campaigns to measure
performance by inputs and outputs (hours worked, calls made) rather than
unmeasurable outcomes. Paradoxically, this makes them reliant on tradition
and routine but also open to new ideas. It pushes them to minimize internal
uncertainty, which often means failing to hold employees accountable for late
work (Was it really their fault? Would it have mattered anyway?) and avoid,
at all costs, firing staff or volunteers.
Temporary organizations also face uncertainty in their task and their
environment [Burke and Morley, 2016]. They face uncertainty within the team,
requiring them to build trust quickly. They often face task uncertainty, hav-
ing to establish their own goals in response to a broad mandate. They face
uncertainty in their relationship to permanent organizations and how those
organizations will judge their performance. Temporary organizations manage
this in two ways: first, by engaging in role-based interaction, where trust is
reduced to a judgment about team members’ ability to deliver in a particu-
lar role; and second, by stronger intra-team interdependence, given that team
members are forced to work together toward a solution in a short time frame.
22
2.3.1.4 Embeddedness
To Xandra Kayden’s three driving characteristics of campaign organi-
zations, project networks literature suggests a fourth: the embeddedness of the
campaign in a broader network of political actors. Kayden was not blind to
this, of course. She noted that the coalitional nature of the Democratic Party
would have an impact on Democratic campaign hiring practices, for exam-
ple, and that over time campaigns should become uniform in their operations
[Kayden, 1978]. Even though she believed finance reforms of the 1970s might
boost party structures by limiting donations to campaigns, she saw more forces
pulling in the other direction. The growth of primaries, the increasing focus
on voter turnout over persuasion, and the rise of a professional campaign class
would all push campaigns to build increasingly robust organizations, as they
decreased their dependence on party structures [Wattenberg, 2013]. More re-
cent work suggests this has borne out. By and large, parties do not control
campaigns, though in their (strong, sometimes conditioned) recommendations
of staff and lists of party-approved consultants, they continue to try [Medvic,
2005].3
But “no project is an island” [Engwall, 2003], and campaigns, like
temporary organizations, can only be understood in their wider context [Burke
3Two examples: one former fundraiser for a congressional campaign reported being told
by the party that their public support was conditional upon their hiring a party-approved—
and eventually disastrous—campaign manager. And from 2019-2021, the DCCC main-
tainted a poorly-enforced blacklist of consultants that had worked with primary challengers
to Democratic incumbents (see New York Times, “The D.C.C.C. Blacklist Is No More”,
March 09, 2021).
23
and Morley, 2016]. For most temporary organizations that context is the
permanent organization that launched and houses them. This relationship to
the permanent organization offers the opportunity for learning and “economies
of repetition” across serialized projects and gives staff a bounded network
within which to advance their career. But that relationship can also create
resource dependence, goal misalignment and conflict, split loyalties among
staff between the temporary and permanent organization, shallow (or “single-
loop”) learning, and the establishment of unproductive routines and rigidities
[Burke and Morley, 2016].
Viewed in this wider context, a number of open questions about tempo-
rary organizations remain unexplored [Burke and Morley, 2016]. They include
how temporary organizations are bounded or integrated into permanent or-
ganizations, how they relate to other temporary organizations, how they are
managed or controlled by permanent organizations, how they balance delib-
erate and emergent strategy, how they balance decision-making power with
permanent organizations, and how their success is evaluated in the context of
the wider organization or network they in which they are embedded.
2.3.2 Organizational Processes in Political Campaigns
The literature above on campaign organizations [Lamb and Smith,
1968, Kayden, 1978, Medvic, 2005] and temporary organizations [Burke and
Morley, 2016, DeFillippi and Sydow, 2016] offer a set of expectations—some
explicit, others implicit—about modern political campaign organizations when
24
viewed as temporary project networks. In the next section, I compile them
into a set of expectations about how campaigns as temporary organizations
should behave across five key organizational processes: (1) Team Formation,
(2) Structuration, (3) Strategic Decision-Making, (4) Performance Manage-
ment, and (5) Organizational Learning.
2.3.2.1 Team Formation
One of the first and most important tasks of the candidate is to build
a team around them [Medvic, 2005]. Prior research on campaigns provide
clues as to both why and how candidates choose team members, and why and
how team members join campaigns. Candidates look for candor, irreverence,
honesty, and loyalty in campaign staff [Kayden, 1978]. Democratic campaigns,
based on their coalitional nature, will build teams whose diversity represents
the different groups they need to motivate to vote. Republican campaigns will
feel less pressure to do so and more likely to hire “copies” of the candidate.
They will hire consultants for specific functions, though one of those functions
might be a general strategy role [Medvic, 2005]. Staff motivations vary widely.
Core campaign staff (which Kayden calls “organizers”) are committed not to
the candidate, for example, but to the building of their own influence and
to the political process. The candidate “is a vehicle for that commitment”
[Kayden, 1978].
In terms of how campaigns hire, candidates tend to hire campaign man-
agers they already know and trust, or else an experienced campaign manager
25
recommended by national parties or activists [Medvic, 2005]. The same will
be true of consultants, with parties providing lists of approved vendors. No-
tably absent from descriptions of campaigns are any references to formal hiring
processes such as job postings or professionalized interviews.
The temporary organization literature itself offers more questions than
answers about team formation in temporary organizations [Burke and Morley,
2016] or how team formation processes might be dependent on an organiza-
tion’s environment. What we can expect to see, according to this literature,
is for temporary organizations to hire for task-relevant knowledge into specific
roles.
Based on the tensions identified by project networks, we can also expect
personnel’s understanding of future work and their divided loyalties to lead to
a gap between how candidates and personnel think about team formation and
hiring [DeFillippi and Sydow, 2016]. Candidates work toward a stable role
as a representative, while staff have much less certainty—even in victory—
about a secure job following the campaign. Meanwhile consultants, given
the increasing professionalization of campaigns, will consider their temporary
relationship to the campaign in the context of their permanent role in the
campaign ecosystem.
2.3.2.2 Structuration
Prior work lays out a clear structure of political campaigns. First,
there is a central campaign committee made up of donors and political allies;
26
a kitchen cabinet of trusted advisors; a campaign manager; and a strategy
consultant [Medvic, 2005]. This core team is augmented, depending on the
size of campaign, with staff dedicated to clearly demarcated roles, including
finance (fundraising), communications, field operations, research (on policy
and the opposition), PR, and digital. Added to that are consultants dedicated
to specific functions (strategy, polling, media, mail, and fundraising). Finally,
there are volunteers and (for incumbent candidates) legislative staff techni-
cally on leave of absence from their day jobs.4The campaign manager works
most closely with the candidate, either as a tactician implementing strategy
developed by the consultant. In very large or very small races, they serve as
chief strategist themselves. Their primary task, according to one scholar, is to
attend to staff motivation and relations [Medvic, 2005].
This more-or-less uniform structure of campaign organizations was put
forward by Steven Medvic in 2005. In 1978, Kayden also predicted uniformity,
though not such clearly defined roles. In her study, impermanence in cam-
paigns led to intensity and informality, rather than hierarchy and bureaucracy
[Kayden, 1978]. What structure did exist was be driven by (a) a distinction
between tasks that had measurable outputs and those that did not; and (b) a
need to quickly provide patterns of authority and organizational predictability
in the midst of task uncertainty. Kayden also argued that the candidate is
peripheral to the campaign: its object, rather than a member of it. Literature
4This seemingly standard maneuver allows legislative staff to comply with laws prohibit-
ing their paid participation on campaigns, in letter if not in spirit.
27
on temporary organizations would predict that campaigns operate more like
the role-defined hierarchies that Medvic identifies. Uncertainty, according to
those scholars, pushes organizations into role-based interactions as a shortcut
to developing trust [Burke and Morley, 2016].
2.3.2.3 Strategic Decision-Making
Kayden and Medvic offer two views of how strategic decisions are made
in political campaigns. Medvic offers a structural view: while lower-level
decision-making might be chaotic, “larger strategic decisions are made at the
top levels of the campaign, and input into those decisions comes from only
a handful of people, including the candidate, key political consultants, and
perhaps a few close advisers” [Medvic, 2005]. This top-level strategy-making
may or may not include the campaign manager, who then coordinates among
staff members charged with executing role-specific strategy often developed
jointly with consultants.
Kayden’s vision of campaign decision-making is entirely different: “If
there is any single rule that dominates political decision making,” she writes,
“it is this: decisions are made by whoever happens to be in the room at the
time” [Kayden, 1978]. This leads to an internal battle for access to the can-
didate. In Kayden’s view, major decisions are driven by a combination of
risk-taking in the face of uncertainty, an aversion to repeating perceived mis-
takes from prior campaigns, and an inertial reliance on tradition.
Lamb and Smith’s work, in laying out two styles of campaign, may a
28
provide a clue as to how to reconcile these two views of political campaigns
[Lamb and Smith, 1968]. In key respects Medvic’s description of campaigns is
similar to that of Lamb and Smith’s “comprehensive” campaign, which is well-
organized around a clear strategy and demarcated roles. Kayden’s description
of campaigns, by contrast, echoes Lamb and Smith’s description of the more
ad hoc, flexible, “incrementalist” campaign.
The literature on temporary organizations offers more open questions
than answers about decision-making within such organizations, as well. Schol-
ars argue that decision-making in temporary organizations will be subject to
tensions between temporary and permanent organizations, and that temporary
organizations will toggle back and forth between what they call idea-generating
and decision-making modes [Burke and Morley, 2016]. The list of what we
do not know about decision-making in temporary organizations includes how
permanent organizations deploy control mechanisms to TOs’ decision-making
process, how those control mechanisms influence the strategy process, how
team composition or divergent mental models affect decision-making, and how
strategy adapts over time in response to shifting goals.
2.3.2.4 Performance Management
Temporary organizations are designed to be dynamic, high-performing,
and hyper-efficient organizations free from “organizational slack” in order to
achieve higher levels of creativity and freedom [Burke and Morley, 2016]. But
the combination of impermanence and uncertainty that political campaigns
29
face means that measuring performance is difficult at both the campaign and
the individual level [Kayden, 1978]. At the campaign level, this leads to task
divisions designed to reduce internal uncertainty rather than maximize per-
formance, and a focus on measuring inputs and outputs in the absence of
being able to measure outcomes. At the individual level, these factors plus
relationship-based hiring prevents managers from sanctioning or firing staff.
To the degree that campaigns resemble other temporary organizations,
it will also complicate the professional development, evaluation and compen-
sation of staff [Burke and Morley, 2016]. In the wider project network, it will
mean that practices from one campaign will get routinized, regardless of their
actual impact on performance [Burke and Morley, 2016]. Kayden observed
this in campaigns’ reliance on tradition, which she argued “tends to include
everything that was ever done (or at least remembered). It is almost synony-
mous with superstition” [Kayden, 1978]. In the wider campaign network, it
also stands to reason that this uncertainty will lead to a reliance on longevity
as a proxy for performance.
2.3.2.5 Organizational Learning
Temporary organizations may be better at creativity and innovation,
but they are often unable to institutionalize that knowledge into permanent
organizations [Burke and Morley, 2016]. They operate on a “situational prag-
matism” that values getting the task done over codifying knowledge, leading
to organizational amnesia. At best, they routinize practices in an un-examined
30
way, developing “core rigidities” rather than learning. They can create knowl-
edge, but not transfer it; they craft practices, but fail to standardize them
[DeFillippi and Sydow, 2016].
Kayden observed this lack of learning first-hand in campaigns, where
campaigns at best relied on tradition and at worst organized around avoiding
repeating “mistakes” from the last campaign cycle [Kayden, 1978]. In short,
we should see little effort to learn from campaign to campaign. Where such
effort does exist, the impermanence and uncertainty of campaigns should make
it mostly unsuccessful.
31
Chapter 3
The Impermanent Campaign: Team
Formation in the Project Networks of Political
Campaigns
“A political campaign is different from almost every other entity
in that there’s a date certain in which it will end. The only other
like comparable example I can think of is like an Olympic Planning
Committee...For a political campaign, time is by far the most valu-
able asset. If you’re a tech startup you can say ‘let’s just launch
our product next year, we can just get more people and money.’
That’s not the way political campaigns work.” - A 2020 congres-
sional campaign manager
3.1 Abstract
How do political campaign teams form? What drives the hiring deci-
sions of campaigns and the decisions of personnel to join a campaign? Using
eight inductive case studies of campaigns for the U.S. House of Representatives
in the U.S. South in 2020, I examine the motivations behind key moments in
the formation of political campaign teams, including candidates’ first and sub-
sequent hiring decisions, personnel’s decisions to join a campaign, and even a
32
candidate’s decision to run. I find those decisions are shaped primarily by the
structure of the extended party networks in which campaigns are embedded.
These network structures differ by party in their structure and their effects
on campaign team formation. At the center of the Democratic network are a
central party organization and an abortion rights campaign committee. At the
center of the Republican network are large, privately owned consulting shops.
I also find that on both sides of the aisle, hiring occurs primarily by
familiarity and word-of-mouth networks, in response to the absence of clear
signals about either personnel or candidate performance that might inform
decision-making. Based on these results and on prior literature on political
campaigns, I argue that campaigns are best understood as project networks, a
type of temporary organization studied by scholars of organizational theory. I
advance temporary organizations literature by studying team formation in a
temporary organizing context and add to our understanding of U.S. political
campaigns by explaining the formation of the teams that battle to determine
the makeup of the U.S. Congress.
3.2 Introduction
In her 1978 book Campaign Organizations, political scientist Xandra
Kayden recognized that campaigns are subject to different pressures than other
organizations by virtue of having a fixed end date. It would be another fifteen
years before management scholars would begin to theorize how temporary
organizations in other industries differ from permanent ones, and then another
33
twenty years before a working definition of temporary organizations emerged:
“A temporally bounded group of interdependent organizational actors, formed
to complete a complex task” [Kayden, 1978, Burke and Morley, 2016].
In this paper, I revisit Kayden’s focus on political campaigns as orga-
nizations, using temporary organizations literature. In the sections below, I
describe prior research on team formation in political campaign organizations,
temporary organizations and project networks (one type of temporary organi-
zation). I then consolidate these streams of literature into a set of expectations
for what we might expect to see in political campaign team formation. Follow-
ing that, I introduce and report results from eight case studies of congressional
campaigns and supporting interviews. I share results from those case studies
on the hiring and joining motivations of candidates, hiring managers, and cam-
paign personnel. I also explore the methods these actors use to find and select
each other, the dynamics of supply and demand for political campaign labor,
and the changing demographics of political campaigns. In the last section,
I discuss these results and their implications for the study of temporary or-
ganizations, project networks and political campaigns, and I conclude with a
description of future work.
3.3 Theory
3.3.1 Temporary Organizations and Project Networks
In the mid 1990s, organizational scholars began to formalize a theory of
temporary organizations as distinct from “permanent” organizations without
34
a determined end point [Lundin and Steinthorsson, 2003, Burke and Morley,
2016]. That research grew into a wider consideration of temporary organizing
that takes place in both temporary organizations and permanent organizations.
One subset of temporary organizations is the project network [Jacobs-
son et al., 2015]. In such networks, temporary projects occur across organiza-
tions within an existing field, or in firms created and then disbanded around a
specific project, such as theater productions or construction projects. Adapt-
ing an earlier framework, DeFillippi and Sydow argue project networks have
three defining characteristics:
(1) a set of relations, where no single actor may act as a legiti-
mate authority for the network as a whole; (2) a situation where
the network is open in the sense that there are no definite criteria
by which the boundary of the network may be identified and con-
trolled; and (3) an environment where the network is temporally
limited, dynamically changing, and (partially) reconstructed from
one project to the next [Hellgren and Stjernberg, 1995, DeFillippi
and Sydow, 2016].
In such a network, temporary projects find themselves subject to five
tensions: between temporary and permanent organizations; between knowl-
edge creation and transfer; between individual and collective identity; between
crafting and standardizing practices; and among past, present, and future
work. Project networks, they argue, manage these tensions through four gov-
35
ernance mechanisms: responsibilities, routines, roles, and relations [DeFillippi
and Sydow, 2016].
3.3.2 Campaigns as Temporary Project Networks
Campaigns are temporary organizations, subject to the same tensions
as other kinds of temporary organizations. Like other temporary organiza-
tions, they exist in a state of tension with permanent organizations—in their
case party organizations and standing political action committees. As tem-
porary organizations, they are highly task oriented [Bryman et al., 1987] and
face “high levels of environmental uncertainty” they resolve with varying de-
grees of success through defined roles and categories that build trust quickly
[Burke and Morley, 2016]. They wrestle with questions of how to help staff
build career capital across multiple projects over time, with varying degrees
of resource dependency on permanent organizations, and with different defi-
nitions of success on the part of the temporary organization (in this case, win
the election) and permanent ones (win power or push a policy agenda) [Arthur
et al., 2001, Bakker et al., 2009].
More specifically, and much more so today than when Xandra Kayden
was writing about political campaign organizations in the 1970s, campaigns
exhibit the traits of project networks, a subset of temporary organizations.
They comprise networks of actors embedded in an even wider network. That
network includes state, national and local party organizations; interest group
and incumbent candidate PACS; consultants, consulting firms and candidate
36
training support organizations; rival and friendly campaigns; donors and inde-
pendent expenditure groups (Super PACs); and informal networks of political
and policy campaign staff.1As with other project networks, none of these
actors plays the role of sole legitimate authority; the boundary of the network
cannot be controlled; and the network of actors is continually changing and
reconstructed as projects take place [DeFillippi and Sydow, 2016]. In political
campaigns as in other project networks, these realities generate a constant set
of tensions that actors in this broad, loosely bounded network must manage.
3.3.3 Team Formation in Political Campaigns
One aspect of temporary organizations that has yet to be studied is how
the those teams form, and how the team formation process is dependent on the
environment in which temporary organizations are embedded [Burke and Mor-
ley, 2016]. We can build some expectations from existing work on temporary
organizations, however. We should see teams hire based on pre-defined roles
so as to build trust early, for example, and for hiring to be heavily influenced
by the permanent organizations upon which they are resource-dependent.
Based on the tensions identified by project networks, we can also expect
to see a gap among how candidates, consultants and staff think about team
formation [DeFillippi and Sydow, 2016]. Candidates, in the simplest view,
are running campaigns to earn (or defend) their job as a legislator. Staff,
1This is precisely the kind of diffuse, opaque environment that Xandra Kayden predicted
would develop in the wake of changes to campaign finance law in the 1970s [Kayden, 1978].
37
meanwhile, are motivated by a number of considerations, not least among
them how a particular campaign fits into the rest of their career. Consultants
are often running permanent businesses that serve multiple clients, not all of
them political. And other actors, such as political action committees, have
their own financial or ideological goals.
Political science research on campaigns provides scattered clues as to
why and how candidates choose team members, and why and how team mem-
bers join campaigns. In her 1970s case studies, Kayden observed campaigns
looking for candor, irreverence, honesty, and loyalty in campaign staff [Kay-
den, 1978]. She also observed Democratic campaigns building teams whose
diversity represented the different groups they need to motivated to vote. Re-
publican campaigns, however, felt less pressure to do so and were more likely to
hire “copies” of the candidate. Staff motivations, according to Kayden, varied
widely. But core campaign staff (whom she called “organizers”) were commit-
ted to building their own influence and to the political process itself, rather
than to the candidate, who “was a vehicle for that commitment” [Kayden,
1978]. In terms of how campaigns hire, contemporary scholars of campaigns
have observed that candidates tend to hire campaign managers they already
know and trust or an experienced campaign manager recommended by na-
tional parties or activists [Herrnson, 2005]. Notably absent from descriptions
of campaigns are any references to formal hiring processes such as job postings
or professionalized interviews.
38
3.4 Data and Methodology
The first goal of this paper is simply to describe the variation in how
political campaign teams form. The second goal is to lay out how and why
political campaign teams form as they do. These are questions of internal
organizational processes, questions that lend themselves to qualitative exam-
ination through case studies [Yin, 2018]. And in an environment with little
preexisting theory about how these organizations should behave, inductive re-
search can serve as the first step in the theory-building process [Eisenhardt,
1989].
In this paper I deploy an inductive, qualitative, multiple case study ap-
proach to studying eight campaigns for the U.S. House during the 2020 cam-
paign cycle. This approach allows for examining the internal processes across
multiple organizations [Yin, 2014]. Each case consists of 2-3 semi-structured
interviews of 45-75 minutes with candidates, campaign managers, consultants
and staff, supplemented by data from the political website Ballotpedia.org and
the Federal Election Commission’s database of donations and expenditures.
In addition to the 18 interviews conducted for those eight cases, I con-
ducted supporting interviews with 12 candidates, staff, and consultants from
ten other U.S. House campaigns; and with 18 other state-level candidates, cam-
paign staff, consultants, lobbyists, legislative aides, and national party staff.2I
2I also conducted one quickly aborted interview with a childhood friend, now a current
New Orleans preacher, who happened to call from a new number just as I was waiting for an
interviewee to dial in. While confused, my friend was still quite willing to share his opinions
about the political campaigns he wasn’t involved in.
39
then thematically coded each of the 28 interviews twice, first based on the the-
oretical expectations laid out above and second according to the themes that
emerged in the interviews themselves [Braun and Clarke, 2012, Lentz et al.,
2019]. Using the networked note-taking software Roam Research, I was able
to compare and correlate themes across interviews [Eisenhardt and Graebner,
2007].
Each of the eight cases share some common context. All the campaigns
were for U.S. House seats in the 2020 campaign cycle, and all occurred in
Republican-dominated states in the U.S. South. Other than this overlap, I
chose cases and conducted supporting interviews so as to gain variation across
each of the factors that exploratory interviews suggested would prove impor-
tant. As depicted visually in Figure 3.1, those included:
Party (4 Republicans and 4 Democrats)
Gender (3 women and 5 men)3
Incumbency (1 incumbent, 2 challengers, and 5 candidates for open
seats)
Success (3 winning candidates, 4 losing primary candidates, and 1 losing
general candidate)
3Throughout the dissertation I change gender, location, and any other details necessary
so that interviewees would not be able to connect the anonymized quotes of their colleagues
to their own campaign. I do this in ways so as not to affect interpretation in those particular
section. But in sections where I consider a particular characteristic of candidates or person-
nel (e.g. gender or race). I retain those details and change others to maintain anonymity.
Where possible, I use gender neutral pronouns (they/them).
40
Competitiveness (3 candidates running in competitive primaries with
competitive general elections, 4 running in competitive primaries with
non-competitive general elections, and 1 with a non-competitive primary
in a non-competitive district)4
Population Density (5 candidates running in primarily rural districts,
and 3 candidates running in primarily urban districts)
Figure 3.1: Breakdown of Cases by Party, Gender, and Other Characteristics
In each conversation, I asked interviewees questions about their back-
ground, motivations, and path to the current campaign; how hiring occurred
4For the purposes of this paper, I did not pursue interviewing candidates running non-
competitive campaigns that they had no reasonable chance of winning, either because they
were third-party candidates, or were running against a safe incumbent from the other party,
or were running against a safe incumbent of their own party. The number of times such
candidates break through is so small that even casual observers of politics will have heard
something of those unusual victories, such as Alexandria Ocasio-Cortez’s 2018 primary win
over Joe Crowley or Dave Brat’s 2016 primary win over Eric Cantor.
41
in that campaign; how the campaign chose its strategy; and how the cam-
paign responded to change—which, in 2020, most often meant the onset of the
COVID-19 pandemic and subsequent shutdown. Appendix A includes a full
list of questions. Unfortunately, due to the pandemic, I was unable to observe
campaigns in person, and all interviews were conducted by phone or video.
3.5 Results
Evident from these conversations is that the environment in which po-
litical candidates build their teams is opaque, dynamic, and driven by multiple
and sometimes contradictory motivations. Some campaigns attempt to man-
age this chaos; others give little conscious thought to whom they hire or how.
Some campaigns hire by way of a lightly structured recruitment process; others
rely on their networks to find or filter personnel. The pathways for hiring for
and getting hired by Republicans and Democratic campaigns differ in impor-
tant ways. But they are alike in their opacity, and in the face of that opacity
most candidates tend to build their teams in an ad hoc way. Republicans tend
to rely on networks of consulting firms and Democrats on party leadership
and a small number of political action committees (PACs). The result is wide
variation, detailed below, in how candidates decide to form a team, how they
recruit and select personnel, what they look for in hiring, and the order in
which they build their team, as well as the drivers behind why personnel join
campaign teams and the nature of supply and demand in the campaign labor
market.
42
3.5.1 The Zeroth Hire: The Candidate
The process for assembling a campaign team starts before the campaign
formally begins. In fact, some candidates treat building a team as a first test
of their own viability as a candidate. One candidate interviewed called the
President and a former governor to ensure their support before announcing
a run. Another made the rounds of local activists, donors, and consulting
firms to see who they were supporting. When they weren’t excited about a
particular candidate, they decided there might be an opening, but only “after
seeing what kind of team to put together.” A candidate from a previous cycle
built a spreadsheet of all their potential advisors and team members before
deciding to run: “Who are the people that are going to go to the mat for me?”
In telling their story about the decision to run, nearly every candidate
made reference to being encouraged, asked, or religiously convicted to run.
Referring to themselves in the plural first person, one candidate reflected,
“This wasn’t something we’d planned on,” until a close family member pushed
them to consider it. The idea of being “called” to run is such a regular feature
of a candidate’s origin story that it is dismissed by some political operatives.
As one Republican campaign advisor put it: “If people say ‘the party asked
me to run,’ they’re full of it...it’s all self-selection.” Still, the absence of a
formal candidate-recruiting mechanism did not mean party elites do not have
mechanisms for hand-picking candidates. One longtime Republican consultant
cautiously described hunting for a candidate to back:
43
We knew—well, we didn’t know, but you know—kind of looking,
that the incumbent was be looking to step down at some point
in time, so we’re looking for a replacement. This candidate was
kind of the guy that was...you know, put up. No one else from the
district wanted to do it. So he stepped up and decided to run.
On the Democratic side, one consultant and former PAC-employee also
referenced recruiting candidates. Recruitment, they said, had been a hard
sell in the Republican-dominated 2000s and 2010s, resulting in what a local
journalist called “nothingburger” Democratic candidates. As the state had
become somewhat politically competitive, however, candidates were now “all
over the place.”
These actors had their own motivations for recruiting candidates, and
their perspectives on candidate recruiting hint at the differences between the
Democratic and Republican campaign universes. I return to both of these
topics below. But first, I examine how candidates go about building their
team, starting with their first hire.
3.5.2 The First Hire
A candidate’s first hire is an early indicator of the team they will build,
their understanding of the task at hand, and their view of how the organization
relates to the campaign. First-time candidates face a choice. Should their
first hire be a campaign manager, general consultant, or finance director?
Incumbents and challenger candidates with a recent history of running for
44
office come with some of that team already built in. One incumbent, for
example, had a trusted aide that served interchangeably (and potentially not
quite legally) as both their Campaign Manager and Legislative Chief of Staff.
One challenger candidate, running a second time for a seat they narrowly lost,
carried over consultants and a finance director from their previous campaign.
In interviews, each Democratic candidate answered the “first hire”
question a different way. One candidate hired a range of task-specific con-
sultants recommended by a candidate training organization before hiring a
campaign manager. Another hired a finance director first, raised money, then a
campaign manager. A third hired a “call-time manager” to help with fundrais-
ing calls and a trusted friend as a bookkeeper. Candidates on both sides of the
aisle complained about the wave of calls from consultants that came in once
they had publicly filed for election.
The first thing that happens is that every swinging you-know-what
calls you because they want to be your consultant. As soon as you
file, your phone explodes...I wish I’d-a had my campaign manager
set before that...You need to get what we call your moat dragon
to deal with your consultants.
But this presented a dilemma. If a candidate wanted to hire top talent,
the candidate explained, they needed money. And if they want to raise money,
they needed staff.
Perhaps as a way around this dilemma, the extended Republican party
network developed the idea of the general consultant. “Serious candidates,”
45
said one consultant, “should hire a general consultant. The consultant needs
to do a lot of things at the beginning. You need that checklist, like ‘don’t hire
a campaign manager right off the bat.’” The general consultant, according to
another, is responsible for helping a candidate plan the campaign.
General consultants shortcut candidates’ chicken-versus-egg, money-
versus-staff problem. In the process, consultants gain a powerful and durable
position in the network of Republican political campaign personnel. Over time,
as their own employees leave to start their own shops, they build extended
networks of friendly competition. Said one densely networked Republican
consultant:
You know, it’s kind of like football. You have a head coach. And
then you see that twenty years later all the head coaches that
worked for that head coach. It’s almost the same thing with general
consultants and managers. They almost have a lineage, a tree of
people that they have worked with that have their style or put their
spin on that style to be successful... [They work for] all these all
these different big time names—worked for them at some point—
and they are now the consultants and they are training the next
generation of Republican consultants.
None of the more than two dozen Democratic candidates or personnel
interviewed mentioned this kind of infrastructure on the Democratic side.5In
5This came as a surprise to one Democratic operative, who listed a number of Democratic
consulting shops. One likely explanation is that some of the idea of “general consulting” has
bled into Democratic politics in these Republican-controlled states, although unsuccessfully.
46
fact, when one mostly state-level Democratic consultant pitched a candidate
on hiring him as a general consultant and his colleague as a campaign manager,
the candidate told him the Democratic Congressional Campaign Committee
would not allow them to do so.6
3.5.3 Assembling the Team: The Methods
By and large, and on both sides of the aisle, candidates and person-
nel use an informal, semi-professional network to recruit and filter campaign
personnel. In four dozen interviews covering dozens more hiring processes,
all but four personnel were hired through the candidate’s extended network,
most often through other consultants, staff, or party organizations. All four of
the Republican candidates—whether incumbents, candidates that had run in
previous races, or first-time candidates—began their races drawing from deep
political connections. Three of them launched their campaigns having already
hired a general consultant or campaign manager through these connections.
From these key first hires down to low-level staffers, the use of net-
works and personal referrals dominated hiring in political campaigns. Some
of this network-based hiring was based on shared experience from a previous
campaign (explored in detail in the next chapter). Three of four Democratic
campaign managers were on one end or the other of this kind of familiarity-
driven hiring. And as one Republican campaign manager put it very clearly
6It is also possible that this was just the reason the candidate gave, and that they had
other reasons for not hiring that consultant.
47
when speaking about the consultants they managed: “Everyone on this race
I’ve worked with before, and/or they have worked with the candidate.”
Beyond these networks of shared experience, campaigns made use of
all aspects of their network to search for and evaluate potential staff: donors,
informal “kitchen cabinet” advisors, existing consultants, previous candidates,
non-political friends, and even staffers’ personal networks. This was not always
strategic. One staffer, without their manager’s knowledge, introduced a candi-
date to a media consultant recommended by another political operative—who
also happened to be the staffer’s mother.
There were three exceptions to this kind of organic, network-driven
hiring, all on the Democratic side.7One (and only one) interviewee reported
posting a Request for Proposals (RFP) for a consultant. This campaign, run
by a veteran campaign manager, was for a well-connected but first-time Demo-
cratic candidate. A second non-network pathway for hiring into Democratic
campaigns was the progressive job board Jobs That Are Left.8At least four
of the Democratic staffers, including a Field Director, a Finance Director, and
7One mid-level Republican staffer got a job by cold-calling a candidate, having heard
about them at a College Republicans event. This seemed to work better than the lone
example of a Democratic campaign manager, not part of one of these case studies, who saw
a job posting online. That relationship ended bitterly.
8All of my interviews were conducted in the summer of 2020. As of July 2021 it
appears this job board may have gone private (see https://www.gainpower.org/2020/
04/jobsthatareleftupdates/). This does not seem to have gone over well among pro-
gressive Twitter users, who have launched two competing Google Groups: Jobs For
The LEFT, “An actual job board for the LEFT” (see https://groups.google.com/
g/jobsfortheleft) and another, unrelated, Jobs That Are Left (see https://groups.
google.com/g/jobsleft).
48
two other lower-level staff came to campaigns through this list. None of the
Republican staffers interviewed hired personnel or found work through a job
board.9
A final pathway for campaign hiring emerged among Democratic cam-
paigns. The Democratic Congressional Campaign Committee (or DCCC, pro-
nounced “dee-trip”) and other extended party organizations, most notably
Emily’s List, played critical roles in the hiring of all four campaign teams.10
Three of the four campaign managers were hired through one of these two
groups, and all four candidates sought their help either in gathering resumes
or running potential personnel by their “desk” at the DCCC. This is not to
say that these groups were universally helpful. In fact, the DCCC worked
against one insurgent candidate challenging a Democratic incumbent, at one
point blacklisting any consultant or staffer that worked against a Democratic
incumbent.
Another candidate, searching specifically for a female campaign man-
ager in the middle of the heated Democratic presidential primary, could not
get qualified resumes from women from the DCCC, Emily’s List or the state
party.11 Ultimately, they poached a Finance Director from a candidate support
9One longtime campaign manager mentioned a website called Campaign Hunter,
launched by the Republican consultancy Majority Strategies in 2015 (see https:
//majoritystrategies.com/blog/big-announcement-meet-campaign-hunter/). That
service seems to have evolved into a staffing agency called Majority Hunter (see https:
//majorityhunter.com/).
10Said one candidate: “You can’t swing a dead cat without hitting an organization that
wants to help you with your campaign.”
11Emily’s List is properly, but distractingly, capitalized as EMILY’s List.
49
organization, then promoted her the Campaign Manager. One consultant to a
campaign outside of these eight case studies said they would “never trust the
DCCC’s process again,” after losing a fight against the DCCC’s demand that
the candidate hire what turned out to be a “bum steer campaign manager.”
By contrast, of all the Republican personnel I interviewed or discussed with
candidates, only one mentioned having floated a resume with the the DCCC’s
Republican counterpart, the National Republican Congressional Committee
(NRCC).
That the majority of hiring in these campaigns was network-driven
does not mean it was not strategic. The level of strategic consideration that
campaigns gave to hiring varied widely. Two campaigns (one Democratic
and one Republicans) had given little thought to the hiring process. The
Republican candidate (a successful one, it should be noted) went so far as to
say “I wasn’t really thinking about hiring” during the campaign.
On the other end of the spectrum were two Republican candidates that
were most strategic in their hiring. One candidate, who hired exclusively
via recommendation from existing consultants, grilled potential vendors for
details about the costs of different ad placements. They attributed this style
to their business background. “Business people are unique in wanting to know
the details. At the end I try to get the best work out of people.” Another
attributed their care in hiring to their military training.
50
3.5.4 Assembling the Team: The Motivations
The range in methods by which campaigns hire can be explained in part
by the even greater array of motivations driving those decisions. By far the
most common motivation for hiring overlaps with campaigns’ most common
method of hiring: shared prior experience, explored in the next chapter as team
familiarity. Candidates and hiring managers use shared prior experience not
just as a method for finding potential staff; that shared prior experience is the
motivation for hiring a particular person. As one retired consultant (not part
of one of these case studies) put it, “it’s easier and better to hire somebody
you know and trust that helps you get there than to take a chance.” Famil-
iarity means not only foreknowledge of skills, but also of potential personality
conflicts, organizational rhythms and culture.
The reasons campaigns and personnel gave for hiring based on shared
experience preview the range of reasons given for hiring more generally. These
included hiring for particular skills and experience; access to knowledge and
networks; “fit” with the campaign; race and gender diversity; diversity of opin-
ion; competitive signaling or posturing; and in one instance, their availability
to start work the next day.
The second most common reason for hiring political personnel was, by
far, prior experience in political campaigns. Interviewees used the word experi-
ence to mean a range of things, including experience in another House race, or
in a race with a similar profile, or in a particular function. In explaining why
they had hired a particular Field Director, one campaign manager explained:
51
We had a candidate who had...electric personality skills. He was a
great public speaker, he had this incredible resume, but he had no
political experience, and nobody in the district knew him...That
sounds like you know, a senator...who gets dropped into Iowa,
where not a single person knows who they are, but they have
to form relationships from ground zero. And I was like, ‘I want
somebody who’s done a presidential campaign before,’ who can
implement some of these town hall, coffee shop type tactics for
this candidate.
This hire represented a common theme in campaigns’ hiring. High-level
hires were chosen to balance out the candidate’s gaps in experience (“What
does the candidate need more hand-holding on?” asked a staffer) or the pub-
licly perceived weaknesses of the candidate. Completely absent from these
comments about hiring campaign personnel because of their experience was
any reference to individuals’ actual performance in previous campaigns. In
fact only one consultant, who had multiple years of experience running na-
tional campaigns, mentioned looking back at prior election performance as a
measure of individual performance. Otherwise, performance and promotion
were fundamentally disconnected. Explained another consultant: “If you stay
in, you move up the ladder, regardless of if you have the skills or capabilities
to do that.”
In one hotly contested Republican primary, the campaign manager fo-
cused entirely on hiring staffers with local ties in order to fend off attacks of
candidates being from outside of the district. As when campaigns hire for
52
experience, however, underneath the drive for “local” personnel is an entire
range of motivations. For one Democratic campaign manager, local experience
meant a head start in building a ground game for a candidate without a lot
of local knowledge. To another Republican candidate, a staffer’s being from
an area culturally similar to the district made the difference in whether that
staffer could connect with local supporters and voters. Lastly, one other out-
sider candidate’s efforts to hire a local campaign manager were driven entirely
by the access to networks of local funders and supporters. It is a story worth
sharing at length:
I knew I needed to meet some people in the district who would be
able to help me kind of introduce me to other people, you know,
politically influential people that control some of the politics in-
volved themselves in the district...So I reached out to a friend of
mine, who I knew had grown up one town over...I knew that they
knew people in town, so I called them and they introduced me
to one of their best friends who is a retired high school teacher
here...So a lot of other folks, you know, a lot of the influential folks
I was trying to get in with, they had taught all of them in high
school, and then taught all of their kids.
So I went down and met with them and their spouse. And they
just agreed to be my campaign manager and help me get started...
They just started driving me all around the district, introducing
me to the bankers, and the CEOs, the insurance folks and the real
estate people, just everybody. They just set up meetings all over
53
the district, every day, 3-4 meetings set up with anywhere from
5-20 people who would gather together, and I’d talk to ’em.12
Hiring managers on both sides of the aisle regularly referenced three
other important considerations in hiring. First was their willingness to work at
a punishing pace. “You hustle all year,” said one campaign manager. “It’s all
day, every day,” said another. The second consideration was “personality fit,”
a phrase that seemed primarily to mean the ability to put up with a particu-
lar candidate.13 The third consideration, mentioned separately by the heads
of two Republican campaigns, was diversity of opinion. While perhaps less
streamlined, they argued, this led to healthier discussion and better decisions.
The examples above are focused primarily on what potential personnel
bring to the team in terms of their experience, their networks, their person-
alities, and how they fit with the rest of the team. A number of campaigns,
however, mentioned making hiring decisions with an eye toward the impact
that hire might have outside the campaign. One Republican candidate hired
an entirely local team in an unsuccessful effort to highlight the contrast be-
tween their campaign and that of their outsider opponent. Another did so
to preempt negative attacks that the candidate was from outside the region.
In one Democratic campaign, a candidate who wanted to swap out campaign
12This quotation heavily edited to preserve anonymity.
13One thoughtful interviewee wondered aloud about what personality types get filtered in
and out of the campaign industry. Another had a quick-draw answer. “Campaign workers
must be people-pleasers and must be children of alcoholics, based on how hard you work to
make this person’s life easier.”
54
managers kept the previous manager on payroll through the election, with
another staffer serving as a sort of shadow campaign manager:
The number one hit from our opponents was that the candidate
wasn’t in the district and spent all their time out of state, etc., etc.
And so we didn’t want to make it seem like their campaign was
being run by a white lady from out of state.
This was one of two instances out of eight cases where campaigns
avoided hiring an underperforming team member rather than allow opponents
or their own party’s leaders, know there might be issues in the organization.
Lastly, one Republican campaign involved in a multi-candidate primary hired a
well-known fundraiser not only to be able to access their funding networks, but
also because they wanted them “off the table” for other primary opponents.
3.5.5 The Shifting Demographics of Campaign Teams
The 2020 campaign cycle, during which these interviews were con-
ducted, came in the midst of a racial reckoning following the murder of George
Floyd, a Black man killed by a white police officer. In response to public pres-
sure, both political parties’ presidential campaigns released data on the diver-
sity of their staff. In my interviews, however it was Democrats, with a single
exception, that were wrestling with the question of staff diversity. It was a lone
Republican advisor, themselves a person of color, who was willing to describe
the situation most bluntly: “There is an unbelievable amount of nepotism in
55
that space.” In the context of the rest of our conversation, I understood this
to mean not only family-based nepotism, but also a tendency to hire from ex-
isting networks, therefore systematically excluding people based on race and
gender. Multiple Democratic interviewees expressed an understanding that
hiring through existing networks, especially for nebulous criteria such as ”fit”
or ”trust,” allowed for and even encouraged exclusionary hiring.
In response, Three of the four Democratic campaigns included here
were explicit about hiring women and people of color, in what one campaign
manager described as “bench building.” One Democratic candidate was ex-
plicit:
I hired all women around me. I mean I was very purposeful in
that because if I as a candidate don’t go out and hire women,
how the hell are we going to grow up women as staff? And it’s
very hard to find really good staff for your campaigns that are gals
and especially obviously diverse women. And why? Because we
frickin’ don’t train them up purposefully. And if you don’t seek
them out, and if you don’t hire them even though they’re maybe
not as qualified, you know, give them a chance...And so that was
my goal.
This change did not come without backlash. One longtime Democratic
campaign manager angrily maintained that his fellow Democrats were “more
open to checking off intersectional boxes and creating safe spaces for people
and making people happy than winning elections...I don’t even recognize my
56
own party.” Overall, however, interviews with Democratic staffers reflected
a growing consensus that the party had a positive obligation to diversify its
makeup.
3.5.6 Motivations on the Other Side of the Table
The disgruntled Democratic staffer quoted above described (however
tactlessly) a very real tension that played out in Democratic campaigns at all
levels in the 2020 campaign cycle. For candidates interviewed here, the primary
motivation was clear: they wanted to win.14 But Democratic candidates also
expressed wanting to diversify the pool of personnel serving their campaigns,
a motivation voiced not as a means to an end but an end in itself.
This tension was just one among many reflected in the reasons personnel
chose to work on one campaign over another. Their reasons often overlapped
and reinforced each other. Campaign personnel’s choices to join a campaign
were motivated not just by winning and the desire to see more diverse candi-
dates succeed, but also because of the desire to support a particular candidate
and how a particular race might shape their own personal careers.
Winning, of course, was a key motivator for personnel, and an explicit
one for some staffers. One Democratic campaign manager picked a race be-
14Each of the candidates interviewed here, regardless of their eventual success at the
ballot box, entered the race able to convince others they had a serious chance of winning.
Why candidates might go through the effort of running a race they know they will lose is a
curiosity for a future paper. Why launch a political campaign destined to fail, when there
are other, more winnable races available?
57
cause they had run a race in 2018 against a similar opponent: “I had their
number.” Multiple Democratic staff mentioned wanting to be a part of flipping
a seat, and one joined a moderate candidate primary race into order to push
back against the progressive wing of their party. “I was an all-American, all-
state athlete,” said a manager. “I really like competition.” This comparison
to sports was echoed by another Republican general consultant:
I think the bottom line is, I was always very competitive growing
up, you know, sports, whatever, anything...It’s just a competition,
between the people you think should be there, and the people they
think should be there...I think a lot of the appeal to this is the
competition and the fight for ideals and principles that you believe
in and...hopefully working for people that share that.
Yet as this quote illustrates, the desire to win was not the only rea-
son personnel chose a particular campaign. They were also motivated by an
authentic desire to support a particular candidate and what that candidate
represented. This was voiced in a variety of ways: in terms of the diversity
of the candidate, what the candidate stood for politically, and how personally
impressive the candidate was.
The qualities that attracted personnel to candidates included their pub-
lic presentation, their knowledge of the issues, their determination in the face
of obstacles, their professional (often military) background and their energy
levels. But for Democratic staffers, the diversity of the candidate—in race,
58
gender, or sexual orientation—was often either a necessary condition or a de-
ciding factor. Mostly absent from personnel’s choices were considerations of
personal fit with the candidate. They were concerned instead about political
fit, especially in primary races. One Democratic staffer saw a race “as a battle
between the two wings of the party.” A Republican staffer said “Moderate
Republicans, or RINOs, really get under my skin. They campaign on con-
servative values, pretend to be one thing, and then go vote in a completely
different manner.”
The desire to win and to support a particular candidate were not the
only reasons personnel joined a race. They were often, and sometimes nakedly,
considering how joining a race would advance their own careers, as a “means
to an end,” admitted a Republican consultant. One Democratic manager was
similarly blunt when asked why they chose a particular race: “First, just basic
career advancement...I had never done a federal race before...Second, it was a
high-profile race. That’s important to your career.” Whether veiled or explicit,
for those personnel that considered themselves campaign operatives, where a
particular race fit into their career trajectory was top of mind. One first-
time and disaffected Republican staffer put it bluntly: “People are applying
because they think [the candidate] will win. They are not true believers...and
the manager just wants to win. He doesn’t care about the candidate.”
For Republican consultants, the ability to sell work to a campaign could
also serve as a means to the end of a promotion within the firm. When they
were hired by a particular candidate, one consultant said,
59
I was actually just working on contract with [the consulting firm].
It was my ‘get.’ It was my get through and through...It’s not like
I got a big check or anything like that but I was responsible for
bringing that client in, and of course it’s a great asset to have, to
be a closer.
Neither in this interview nor in any other were salary levels mentioned
as a motivator for joining a campaign. Where self-interest drove personnel
decision-making, it was measured in career advancement, not cash flow.
3.5.7 Supply and Demand for Labor in Political Campaigns
This lack of focus on compensation is one contributing factor to the
complexity of both the Republican and Democratic markets for campaign la-
bor. On the Republican side, the contest for labor revolved primarily around
consultants. According to one general consultant, deciding to work for a cam-
paign is a “two-way interview,” in which their joining can unlock support from
advocates and donors or even launch a campaign. They described what hap-
pened when one incumbent in a safe Republican seat announced retirement:
As soon as consultants hear that the gun is fired, they all go ab-
solutely wild looking for, ‘Who do we know out there? And if we
don’t know anybody, who’s a business leader or somebody sitting
on a big pile of money that wants to run for office?’
At the same time, there were clear signs that Republican candidates
were firmly in the driver’s seat when choosing their consultants. All four
60
Republican campaigns spoke with or interviewed multiple consultants before
making their choices about whom to hire, and every consultant spoke about
their efforts to find new business.
The exception to this rule was in fundraising consultants, who were
specific to a state or region and in high demand. In addition to one campaign
that hired a fundraising consultant solely to prevent them from working for
their primary opponents, another hired a fundraising consultant because they
were the only one left that hadn’t yet disappointed and didn’t have a person-
ality conflict with the general consultant. The labor market for Republican
staff revolved around consultants, as well. Many Republican staffers were ac-
tually on payroll at consulting firms, which then assigned them to particular
campaigns. Among other staffers, both job seekers and hiring managers said
that they mostly took what they could get, hiring the people that asked for
jobs or taking the jobs they were offered.
For Democratic candidates, the hiring of both consultants and staff
was mediated, aided, and sometimes frustrated by the DCCC and Emily’s
List. But while candidates were awash with offers from consultants, even
these powerful national organizations could not help campaigns hire staff at
the height of the 2020 Democratic presidential primary, when high-quality staff
were drawn into presidential campaigns and then larger, statewide campaigns
In one exceptional case, a campaign manager reported being highly sought
after and having to lobby hard to go work on the campaign they wanted to
join. From there, it was easy for that manager to build the team they wanted
61
even in a tight labor market, thanks to the quality of the candidate, the
candidate’s intersectionality, and the possibility of knocking off a high-profile
incumbent.
3.6 Discussion
In recent years a new wave of research has begun to examine the net-
works surrounding, supporting and shaping political candidates and policy-
makers [Ringe et al., 2017, Victor et al., 2018, Box-Steffensmeier et al., 2019].
This paper builds on that work by extending research on campaign consultants
and legislative staff to include political campaign staff, a set of political actors
that we know comparatively little about. Campaign teams are the first polit-
ical organizations candidates build and lead. They form the context in which
they develop habits for hiring and managing the personnel through whom they
win elections, pass legislation, serve constituents, and build their own careers.
This study focuses on how and why campaigns hire their personnel, and how
and why personnel join campaigns.
3.6.1 The Structure of Political Campaign Networks
I find that at every step, campaign formation decisions are heavily in-
fluenced by the network of relationships in which a campaign is embedded,
and what actors wield influence in that system. For Republicans, the most
prominent and influential set of actors are political consulting firms, large en-
tities that take on running multiple aspects of a campaign. Their central role
62
in the Republican campaign ecosystem means they influence, and sometimes
even determine, who runs for Congress on the Republican ticket, shaping the
team even before the candidate is in place. They are often Republican cam-
paigns’ first critical hire, and exert considerable influence on subsequent hires,
sometimes even being responsible for the staffing of the entire campaign. They
are the path into campaign work for many Republican staff, or where young
staff go after their first campaign.
On the Democratic side, the central players are the Democratic Con-
gressional Campaign Committee (DCCC); Emily’s List, a pro-choice political
action committee (PAC); and other candidate support organizations. These
organizations influence political campaign team formation very differently than
the Republican campaign shops. They maintain resume banks and lists of sug-
gested consulting firms. To campaigns they support, they are regular sources
of free advice and guidance. The DCCC also uses the promise of endorsement
as a way to pressure campaigns into hiring or not hiring particular personnel.
With the support of these organizations, Democratic campaigns tend to hire
either a manager or finance director first, then fill other roles with resumes
collected from and/or filtered through these organizations. Figure 3.2 lays out
how these different structures emerge from how parties manage the tensions
of project networks (explored in more detail below), tensions that flow from
the basic characteristics of campaigns.
In the section below, I examine how the structure of Democratic and
Republican campaign networks shape the structure of political campaigns. But
63
Figure 3.2: How Two Parties Manage the Characteristics and Tensions of
Campaigns
why would Republicans and Democrats have developed such different network-
level structures, when they are attempting to achieve the same task and man-
age the same tensions? One potential explanation is ideological: perhaps
modern Democrats’ stereotypical tendency toward centralization of power led
to the rise of powerful central party organizations, while more market-friendly
Republicans privatized as many campaign functions as they could.
What little history of central party organizations exists, however, sug-
gests a much more complicated story. In the 1960s and 1970s, according to
one history, Democrats did indeed take steps to consolidate power in central
party organizations [Conway, 1983]. But Republicans did, too–by centralizing
their fundraising power and then offering services to Republican candidates.
And according to another report, the Republicans’ efforts to centralize proved
more effective: By centralizing fundraising ability, the National Republican
64
Campaign Committee (NRCC) was able to consolidate power and even be-
gin supporting candidates, while the Democratic party organizations ceded
more and more fundraising power to individual candidates [Jacobson, 1985].
Without financial leverage, central Democratic party organizations also found
themselves less able to support and assist individual candidates [Herrnson,
1986].
In the 1980s, centralized Republican control began to slip, as individual
Republican legislators began directly supporting other candidates and Newt
Gingrich building up of GOPAC as an alternate path for conservative campaign
dollars [Koopman, 1996]. By the mid-1990s, the NRCC had ceded much of
its fundraising turf, and with that went its ability to influence and support
individual campaigns.
Beyond those few works (and one fairly recent paper on direct ver-
sus brokered campaign contributions), there is little other historical record of
Republican and Democratic campaign networks evolved into the structures de-
scribed in these interviews, thirty years later. What is clear is that in the wake
of the decentralization in the 1980s and 1990s described by Koopman, private
Republican consulting firms stepped into the gap to raise and spend cam-
paign dollars on behalf of candidates. On the Democratic side, meanwhile, the
DCCC’s efforts to centralize authority has been incomplete. As with Republi-
cans in the 1990s, most fundraising happens outside of the DCCC’s purview,
and they were not powerful enough to stop the rise of PACs such as Emily’s
List, which now play central (and sometimes opposing) roles in the Democratic
65
campaign network. But they have carved out a central role in the formation of
campaigns, with the power, however imperfect, to influence campaigns’ hiring
of staff and consultants. It is within these partisan networks that the next
section examines the two common, party-specific campaign structures.
3.6.2 The Structure of Political Campaign Teams
In Figure 3.3 and Figure 3.4 I present simplified visualizations of two
archetypal campaign structures: One for Democratic campaigns and another
for ”strong consultant” Republican campaigns. These are stylized and by no
means universal, but rather the result of path dependence in Democrats and
Republican project networks. They are also structures that are passed down
from campaign to campaign, rather than strategically determined by each
campaign or the parties. Campaigns on one side or the other choose these
structures not after careful consideration of the alternatives, but because that
is the way other campaigns have done it. Even those campaigns that have
built variations of these structures do not seem to have done so consciously.
In fact, they are mostly unaware that they are structurally different from other
campaigns.
In both figures, the dark lines represent lines of explicit management or
authority. The dashed lines represent implicit lines of management or author-
ity. Where these lines intersect or overlap, there lie the intra-organizational
tensions for a typical campaign from each party. A Republican campaign man-
ager, for example, might be appointed to report to a candidate, even if their
66
paycheck is coming from the consulting firm they will return to after the cam-
paign. On the Democratic side, a campaign manager is hired by the candidate,
but as Xandra Kayden pointed out, their allegiance and career hopes are with
the party.
According to two Republican consultants, Republican campaigns come
in two varieties: The “strong consultant” campaign, depicted in Figure 3.4, and
the “strong manager” campaign. In the former, the General Consultant serves
as the de facto head of the campaign, choosing and managing staff and other
consultants. Three of the Republican campaigns interviewed here were strong
consultant campaigns. The fourth was a strong manager campaign, where
the Campaign Manager served as the head of the campaign, with consultants
reporting jointly to them and the candidate. All four Democratic campaigns
closely resembled the structure reflected in Figure 3.3.
I also find inside both of these two distinct network structures, per-
sonal and professional networks drive the team formation process. Campaigns
identify, recruit and vet people through their own existing networks or the
networks of consulting shops and party organizations. Only a few candidates,
and almost exclusively those with non-political work histories, attempted to
add some formal hiring structure to that process. And there was little con-
sistency in what campaigns look for in personnel. With few clear patterns
by party or incumbency, campaigns hire for particular skills and experience,
access to knowledge and networks, “fit” with the campaign, race and gender
diversity, diversity of opinion, competitive signaling or posturing, and—in one
67
Figure 3.3: Stylized Structure of a Democratic Campaign Team
instance—their simple availability to start work the next day. Missing from
this list is the verifiable performance of personnel in prior elections. As Xan-
dra Kayden identified four decades ago, there is simply too much uncertainty
about which actions lead to which electoral outcomes for past performance to
be a clear signal of competence [Kayden, 1973, Marland and Giasson, 2020].15
In this environment, finding and filtering personnel through existing
networks may be a rational response. But it is one that leads to predictable
outcomes. Campaigns’ tendency to hire people they know and that “fit” leads
to systemic exclusion of women and people of color (see also [Ruef et al.,
15Also missing were the “candor, irreverence, honesty, and loyalty” that Kayden’s candi-
dates were looking for in the 1970s.
68
Figure 3.4: Stylized Structure of a Republican “Strong Consultant” Campaign
Team
2003]). In 2020, Democratic campaigns made this a central driver in their
hiring. They did this not to access different groups they needed to turn out to
vote, as Kayden observed Democratic campaigns doing in the 1970s, but with
diversity as its own end [Kayden, 1978].
For their part, personnel expressed a similarly broad range of motiva-
tions for joining particular campaigns, including the desire to win, to support a
particular candidate, to build their own careers, and—for Democratic staff—to
see more diverse candidates succeed. This mix of motivations, along with the
fact that they are not strongly financially motivated, adds greater complexity
to an already opaque, confusing labor market for campaign personnel.
69
3.6.3 Team Formation in Project Networks
At the outset of this paper I proposed political campaigns are best
understood as project networks, a particular type of temporary organization.
In this section I return to theory on temporary organizations and project
networks in light of the findings above.
By the standards of one commonly held definition of temporary or-
ganizations, campaigns clearly fit. They are “A temporally bounded group
of interdependent organizational actors, formed to complete a complex task.”
[Burke and Morley, 2016].16 More specifically, they bear the characteristics of
project networks, in that they are:
(1) a set of relations, where no single actor may act as a legiti-
mate authority for the network as a whole; (2) a situation where
the network is open in the sense that there are no definite criteria
by which the boundary of the network may be identified and con-
trolled; and (3) an environment where the network is temporally
limited, dynamically changing, and (partially) reconstructed from
one project to the next [Hellgren and Stjernberg, 1995, DeFillippi
and Sydow, 2016].
Evidence from these case studies complicates at least the first two of
these characteristics. First, the absence of a single legitimate authority was
16In their review, Burke and Morley do not commit to a definition of complexity, though
political campaigns pass most of the tests of the various definitions they reference.
70
was clear in both the Democratic and Republican project networks. But this
lack of legitimate authority proved a power vacuum that generated constant
contest. The DCCC worked to become such an authority, even attempting
to punish actors that worked against their preferred candidates. Republican
consulting shops did not seek the same kind of authority, but they did use their
gatekeeping powers over candidates to their advantage and worked to box out
competitors, a portent of more sharp-elbowed competition to come if a few
firms consolidate the Republican market. This is related to the second criteria
of project networks: their boundaries may not be definitively identified or
controlled. If legitimate authority is contested in project networks, these case
studies show that controlling boundaries through the team formation process
is the ground on which that contest is fought.
3.6.4 The Five Tensions of Project Networks
Project network scholars argue these three characteristics generate five
identifiable tensions project networks must manage, between temporary and
permanent organizations; between knowledge creation and transfer; between
individual and collective identity; between crafting and standardizing prac-
tices; and among past, present, and future work [DeFillippi and Sydow, 2016].
In Table 3.5, I lay out how the Republican and Democratic project networks
have resolved or continued to struggle with these tensions.
71
Figure 3.5: Tensions in Democratic and Republican Project Networks
72
The first of these five tensions, between temporary organizations (cam-
paigns) and permanent organizations (parties, PACs, and consulting firms) was
evident throughout the political campaign team formation process. Multiple
Democratic campaigns were resentful of the DCCC’s influence or incompe-
tence, a predictable result given the gap between the campaign’s narrow goal
to win an election and the party’s broader goal to win or maintain a major-
ity [Jacobson, 1985]. This was less the case on the Republican side, where
the national party exerted little control or influence except through the public
endorsement of the President.17 One Republican advisor did complain that
consultants’ profit motives did not align with the candidates’ need to win. But
by and large, Republican consultants expressed both a desire to win (align-
ing with candidates’ motives) and also a desire for repeat customers (which
likewise depended on their client winning).
Evident, too, was an active tension among past, present and future
work. In these interviews, shared prior experience was a key driver of cam-
paign hiring and joining. Prior campaigns shape hiring for current campaigns,
primarily through the presence of prior conflicts, or a history of working for
opponents; these both prevent personnel from working together on a cam-
paign. Less salient was the question of future work. Consultants certainly
hoped for repeat business and candidates knew staff were thinking about their
17The president’s explicit endorsement and candidates’ ability to pitch themselves as close
to Trump proved decisive in two of the primary elections studied here. In the same way the
absence of a powerful central party organization contributed to the rise of large Republican
for-profit consulting shops, it also contributed to the development and persistence of the
“Trump effect.”
73
next campaign. But these were not ongoing sources of tension, and staff were
more concerned that a job continue to demonstrate upward trajectory with
their past work than in preparing for unknown work to come.
Less evident in these project networks was a tension between individual
and collective identity. Candidates did not expect consultants to identify with
the campaign. They also seemed to understand staff would not, either. The
only collective mission was to win. They hoped for competence, not personal
identification with the campaign. And only one campaign manager expressed
a sense of responsibility for where their staffers might go after the campaign.
The final two of DeFillippi and Sydow’s tensions are between creating
and transferring knowledge, and crafting and standardizing practices. Nei-
ther of these were active tensions in political campaign team formation; both
Republicans and Democrats had resolved these in some way. There were ex-
ceptions, including one staffer and two candidates with a military background
who were separately considering writing books on campaign management. But
these were exceptions among campaigns that proved mostly incurious about
creating knowledge or transferring it, or about creating practices or standard-
izing them. Where questions of campaign management are examined in depth,
it is in smaller, outside organizations (such as the conservative Leadership In-
stitute and the progressive group Arena) that have taken up the question of
how to gather and disseminate that knowledge. Even still, publicly available
guidance on whom to hire and how to hire them was, at least in 2021, still
limited to a handful of podcasts and blog entries.
74
To manage these tensions, project network scholars argue, organiza-
tions rely on four governance mechanisms: responsibilities, routines, roles and
relationships.18 Most obviously at work in the project networks of campaigns
were relationships and roles. Relationships and the information that flowed
across them determined who hired whom and what campaigns were worth
joining. Roles were often clear, too: most consultants were not general con-
sultants, but specialized in a specific task. In Democratic campaigns, there
were clear rules about what positions would be filled depending on the size
and budget of the campaign. Consultants’ specialization, however, is thanks
to the privatization of campaign work, not to governance concerns [Sheingate,
2016].
Less clear was governance by responsibility or routine. Given the fun-
damental uncertainty of campaigns, individual personnel could only be held
responsible for outputs, rather than outcomes. Even then, campaigns were
often too short and outcomes too uncertain for individuals to be held respon-
sible in a meaningful way across the network [Kayden, 1978]. And the idea
of routinization of campaign work was not a salient consideration, perhaps
because of the rapidly changing nature of campaign work in the face of tech-
nological change or because the ego of individual candidates led them to resist
templated consulting services. Where this did exist, it occurred not within
the temporary organization (the campaign) but in permanent organizations
18The original authors used the word relations here rather than relationships, which I
prefer.
75
(task-specific consulting firms) and not for governance’s sake, but in order to
expand their profit margins.
3.6.5 Conclusions and Future Work
The initial, narrow goal of this paper with regard to temporary organi-
zations and project networks was to illuminate how team formation happens in
temporary organizing environments, a previously noted gap in that literature
[Burke and Morley, 2016]. To that end, I find impermanence impacts hiring in
two main ways in the project networks of political campaigns. First, imper-
manence makes assessing potential hires or bosses based on their track record
more difficult, leading both employers and employees to rely more heavily on
relationship-driven hiring. The opacity this implies means team formation it-
self becomes a ground for power struggles within the network. It is true this
may be magnified by the uncertainty that defines political campaigns, but
other project networks likely deal with comparable uncertainty.
Second, team formation in project networks is heavily influenced by
contests for power at the network level, both in the past and in the present.
In political campaigns, these played out through the efforts of the DCCC, the
candidate-making power of Republican consultants, and the influence of the
45th President. Third, this paper highlights which of the paradoxes and gov-
ernance mechanisms proposed by project network researchers were more ‘live’
than others. Future work might explore why certain tensions and governance
mechanisms are active versus resolved in different types of project networks.
76
This paper uses one lens to examine one set of internal processes in
campaign organizations. There are other theoretical lenses to apply, other in-
ternal processes to explore, and other questions to ask about teams and talent
in political campaigns. One lens from another area of management scholarship
is team formation in entrepreneurial settings. This literature is rich and grow-
ing [Clough and Vissa, 2018, Lazar et al., 2019, Obstfeld et al., 2020]. And
while I argue political campaigns map more cleanly onto project networks than
entrepreneurial teams, connecting that team formation literature with that of
project networks would help test the former and inform team formation pro-
cesses in the latter. There are other organizational processes yet to explore
in political campaigns as well, including how campaigns engage in strategic
decision-making. In this effort, we might deploy recent organizational liter-
ature on open polities [Weber and Waeger, 2017], strategy-in-practice, and
differences in institutional logics at the campaign and network levels [Louns-
bury and Beckman, 2015].
In large part this chapter serves as an inductive theory-building exer-
cise, identifying empirically testable hypotheses. In the next chapter, I ex-
amine how hiring for individual staff experience and hiring for prior shared
experience (known as team familiarity) correlate with campaign outcomes.
Also unexplored is the demographic and racial makeup of political campaign
staff, which emerged as a central and contested issue in Democratic political
campaign hiring in the late 2010s. There is great variation in campaigns’ use of
staff versus consultants, too, something that could be explored quantitatively
77
through FEC disbursement records.
In addition, this paper focuses narrowly on competitive campaigns for
the U.S. House of Representatives, and behind each of those qualifiers is an-
other set of campaigns worthy of study, including campaigns for the U.S.
Senate; state-level and local campaigns; campaigns where candidates have no
reasonable hope of winning; and political campaigns outside of the United
States (perhaps the most well-studied set of government institutions in the
world).
A few areas of potential study on other actors in the political system
emerged directly from interviewees. One is Republican political consulting
shops: how competition and cooperation between them has evolved over time,
how they go about building their businesses, and a curious practice one con-
sultant called “white-labeling,” in which consulting firms hid their association
with unpopular candidates by routing payments through other friendly firms.
The second set of organizations worth further examination are the DCCC and
NRCC themselves, about which very little has been written since the 1980s,
when scholars found that it was Republicans, rather than Democrats, that
had centralized more authority in their congressional committees [Jacobson,
1985, Conway, 1983, Herrnson, 1986]. Understanding their histories could be
key to explaining the structural differences in the Republican and Democratic
campaign project networks, whose relative power is not what either Xandra
Kayden nor recent work on asymmetry in U.S. politics would predict [Simas,
2018].
78
Finally, candidate support and training organizations (some called “boot
camps”) emerged from my interviews as an important but relatively unknown
player in the Democratic campaign network. These organizations were nu-
merous, relatively new, existed uncomfortably with official party leadership,
and by 2020 were having a meaningful impact in Democratic primary races.19
Understanding these actors, along with other actors that make up what recent
scholars have described as the extended political party, [Fagan et al., 2021, Fa-
gan, 2021] may be as important to understanding Democratic campaigns in the
2020s as understanding political campaign consultants and campaigns staff—
the politically central actors this paper only just begins to describe.
19To name a few: Emily’s List, Arena, Justice Democrats, New Politics, Run For Some-
thing, Emerge America, VoteRunLead, VoteVets, New American Leaders, Running Start,
Wellstone, Analyst Institute, and the National Democratic Training Committee.
79
Chapter 4
The Devil You Know: Team Experience, Team
Familiarity, and Performance in U.S. House
Campaigns from 2012-2016
“We don’t want nobody nobody sent.” - Chicago political operative
Abner Mikva, c. 19701
“Everyone on this race I’ve worked with before, and/or they have
worked with the candidate.” - A 2020 campaign manager
4.1 Abstract
Every two years, candidates for the U.S. House of Representatives make
a series of rapid decisions about whom to hire to operate their campaigns. Yet
most of the guidance they receive is made up of untested, often contradictory
rules of thumb. Which of that advice should candidates follow? This paper
examines two factors that previous research on entrepreneurial team formation
shows contribute to team effectiveness: (1) The average experience level of
personnel, and (2) their history of working together, or team familiarity. For
1Taken from the title of Milton Rakove’s 1979 book on the Daley political machine,
[Rakove, 1979].
80
U.S. House campaigns from 2012-2016, I build measures of both, and use
multilinear regression models to test their correlation with campaign outcomes,
as measured by vote share in the general election and fundraising totals.
For most candidates, hiring more experienced personnel is not associ-
ated with either raising more money or earning more votes, with the excep-
tion of Republican incumbents, for whom having more experienced consultants
may be associated with greater vote share. The strongest and most consistent
finding is the positive association between team familiarity and vote share:
Teams with more experience working together earn more votes. More familiar
Republican teams raise more money, too, though Democratic teams do not.
Together, I interpret these results to mean that in political campaigns, the in-
ability of most campaigns to use experience as a proxy for quality of personnel
leads candidates and their hiring managers to build a team from the people
they already know, which helps Republican candidates raise more money and
all candidates earn more votes.
4.2 Introduction
At the close of 2019, Texas Democrats had a problem. In anticipation
of more competitive races in 2020, Democratic candidates had filed to run for
seats up and down the ballot in record numbers. But 25 years in the political
minority meant those candidates lacked a deep bench of talent to run those
campaigns. The number of successful campaign staff that had not left the
state for bluer pastures had grown so small that the movements of campaign
81
staffers became an important topic of Texas political gossip [Livingston, 2019].
Few electoral contests are studied more closely by academics than those
for the U.S. Congress. A robust body of research on political campaigns has
considered the characteristics of individual politicians, their constituencies,
and the systems that contribute to successful campaigns. Upwards of three
quarters of the variation in electoral outcomes in the U.S. House of Repre-
sentatives can be predicted from just a few factors: Which party holds a
seat, whether the incumbent is running, and how that district voted in the
most recent presidential election [Jacobson, 2015]. The remaining variation in
electoral outcomes is what determines who wins close elections and ultimately
which party controls the legislature. Research has sought to explain this varia-
tion by looking at candidate characteristics, including a candidate’s ideological
positioning relative to the preferences of the district and their opponents [An-
solabehere et al., 2001]; the support candidates receive from extended party
networks [Desmarais et al., 2015]; candidate valence, a dual measure of the
character of candidates and the effectiveness of their campaigns [Stone and
Simas, 2010]; and (at the presidential level) the complexity of information
processing candidates are capable of [Brause et al., 2005].
Largely absent from this conversation, however, is an examination of the
individuals who work to get those politicians elected. Every two years, some
2500 individuals register as candidates for the 435 seats in the House. Serious
House candidates then go on to hire anywhere from 10-150 staff responsible
for raising and spending an average of 1.6 million over the course of 12-15
82
months, an amount that climbs each cycle [Thurber, 2018]. An initial look at
campaign expenditures data from the Federal Election Commission suggests
that in a typical House race 20-30% of those dollars are spent on staff or
consultants. In the 2018 election cycle, for example, out of the 1.63 billion
spent by House candidates, 414 million was spent on personnel. But do
those decisions matter to electoral outcomes? Perhaps election outcomes are
so clearly determined by other factors that hiring one campaign manager over
another contributes little to the final outcome of a campaign. If that is true,
candidates would be wise to focus their attention on other important campaign
decisions.
Lessons from research on other industries and on legislative staff, how-
ever, suggest that staff choices matter a great deal. Entrepreneurs, for exam-
ple, tend to make hiring decisions based on familiarity and homophily [Aldrich
and Kim, 2007, Ruef et al., 2003], seeking out staff who remind them of them-
selves and relying on references from people they know and trust [Granovetter,
1973, McPherson et al., 2001]. But for teams launching new ventures, these
mutually reinforcing tendencies are associated with less revenue, lower profits
and slower hiring growth [Klotz et al., 2014]. Effective entrepreneurial team
building requires hiring for individual functional experience [McGee et al.,
1995], diversity of functional background [Gray, 2017], and a balance between
employees with shared work experience versus those new to the team [McGee
et al., 1995]. This last measure, called team familiarity, has been shown to
have an inverted u-shaped relationship with team performance in industries
83
as diverse as software [Espinosa, 2007] and professional basketball, meaning
that more successful teams combine some shared experience with fresh talent
[Sieweke and Zhao, 2015]. This work on team formation, mostly conducted by
entrepreneurship scholars, stands apart from a separate strand of literature on
hiring (“the supreme problem”) by scholars of human resources rooted in psy-
chological theories [Ployhart, 2012, Ryan and Ployhart, 2014, Ployhart et al.,
2017].
Perhaps because of the predominance of agency theory in political sci-
ence, in which staff are conceptualized straightforwardly as agents carrying
out the wishes of their principals with varying degrees of fidelity, research on
the impact of political staff and teams has been sparse. If staff are only acting
on the direction of the principal, there is little reason to examine their behav-
ior. However, there is some research that suggests this limited view obscures
the influence that political staff can have. Some authors argue that legislative
staff are political professionals with substantial autonomy, particularly with
regard to deciding what information is important for the member to know
and the policy solutions used to implement the member’s legislative agenda
[Romzek and Utter, 1997]. Those that earn the trust of their member are able
to directly influence a member’s goals and may be given broad discretion to
pursue their own policy interests. Legislative staff can also impact a legisla-
tor’s ability to pass bills [Crosson et al., 2018], shape a legislator’s beliefs about
their constituents [Hertel-Fernandez et al., 2019], and bring their boss’s leg-
islative effectiveness in line with that of their former employers [Montgomery
84
and Nyhan, 2017].
Personnel on political campaigns have received substantially less atten-
tion, and that attention has focused primarily on the paid consultants who
work for campaigns. In his book Building a Business of Politics, Adam Shein-
gate traces the development of political consulting into an industry that deter-
mines the strategic directions of campaigns [Sheingate, 2016]. Candidates who
rely on professional political consultants raise more money [Herrnson, 1992]
and are more willing to engage in negative campaigning on issues, but not per-
sonal attacks [Francia and Herrnson, 2007]. The world of campaign consultants
is highly interconnected, with campaigns that share consultants engaging in
similar patterns of risk taking and issue ownership in their campaigns [Nyhan
and Montgomery, 2015].
4.2.1 Hypotheses
If hiring choices matter in most industries, including in legislative of-
fices, it stands to reason they would matter in political campaigns, as well.
This paper addresses the impact of two characteristics of campaign personnel:
(1) Their level of prior campaign experience and (2) the amount of time they
have spent working together on previous campaigns. With regards to team
experience, this paper tests a straightforward hypothesis, which is that having
personnel with prior experience in campaigns will make for a better-performing
campaign:
Hypotheses 1-2: The more experienced a campaign team, the
85
more successful a campaign will be in (1) fundraising and (2) out-
performing their expected vote share.
In addition to their own experience, campaign personnel have experi-
ence working with each other and with the candidate. Given that campaign
hiring is done primarily by word of mouth, and driven by who has worked
with whom on past campaigns, then we should expect to see a high degree of
overlap in history working together on previous campaigns among the person-
nel of each campaign. If political wisdom is right, the more that a team has
worked together on previous campaigns, the better. But if academic literature
on effective teams applies in this context, we should expect to see diminishing,
and eventually negative, returns to having a more familiar team, as they lose
creative and adaptive capacity. This leads to the following hypotheses:
Hypotheses 3-4: The amount of shared campaign experience a
campaign team has will have an inverted u-shape relationship with
a campaign’s ability to (3) fundraise and (4) outperform expected
vote share.
In qualitative interviews (see Chapter 3), a stark distinction emerged
between Democratic and Republican campaign practices. Democrats were
much more likely to rely on full-time staff, whereas Republican candidates
tended to lean much more heavily on outside consultants, occasionally not
having a single individual on their direct payroll. These two groups (full-time
staff and consultants) operate with very different incentive structures: The
86
outcome of a particular race matters much less to consultants, who are of-
ten involved with multiple races across one or multiple states. As a result, I
expect to see a stronger relationship between experience and vote share for
Democratic candidates than Republican ones. In terms of the relationship be-
tween experience and fundraising, I expect this dynamic to be reversed, with
team characteristics and fundraising more strongly correlated in Republican
campaigns. Consultants receive or channel the majority of expenses in a cam-
paign, so they are incentivized to ensure that candidates raise every possible
dollar.
Incumbency, too, should alter these results. Incumbents tend to retain
personnel, which accrue general campaign experience without necessarily ac-
cruing experience in competitive races, given partisan gerrymandering. But
the natural focus of incumbents, who mostly face noncompetitive re-election
campaigns, is on raising funds they can allocate to other representatives rather
than on winning additional vote share. As a result, I expect to see positive re-
lationships between personnel characteristics and fundraising for incumbents,
but no correlation between personnel characteristics and vote share.
4.3 Data
This paper requires construction of a novel data set from multiple
sources, including the Federal Election Commission, Adam Bonica’s DIME
database, and the MIT Elections Lab. I describe its creation in detail here.
87
4.3.1 Dependent Variables
There are two dependent variables used in this paper: The first, pulled
from the MIT Elections dataset, is the share of the vote won by a candidate in a
congressional election. The second is a measure of a campaign’s ability to raise
funds: The share of funds raised by that candidate. I arrive at this measure by
taking the sum of money raised by both candidates in a congressional election,
then subtracting the amount of funds given or loaned to the campaign by each
candidate, then for each candidate dividing the amount of external funds raised
by the total external funds raised in that race.
4.3.2 Constructing Study Variables
In order to examine the impact of campaign personnel, I construct a
database of campaign staff and consultants in federal political campaigns from
the present back to 2003, when the Federal Election Commission began to pub-
lish machine readable campaign disbursement reports. The Federal Election
Commission requires all candidates who raise more than 5,000 to register their
candidacy and report both contributions and expenditures, which it defines as
“A purchase, payment, distribution, loan, advance, deposit or gift of money
or anything of value made for the purpose of influencing a federal election. A
written agreement to make an expenditure is also considered an expenditure”
[FEC, 2020]. Included in those expenditure reports are payments to staff and
consultants paid directly by the campaign committees. This means that for
every paid employee, intern and consultant from each House, Senate and Pres-
88
idential campaign since 2003, these FEC data contain that person’s name and
address, the amount they were paid, and the purpose of that payment.
While organizations such as ProPublica have built interactive interfaces
for exploring the data, the FEC is the only source of bulk data on campaign
expenditures. Turning these 11 million expenditure records into an operable
database involved multiple steps: First, I identify all of the campaign expen-
ditures that might be related to staff and consulting by building a dictionary
of personnel-related terms. Second, I build another dictionary for removing
non-salary staff expenditures, such as food and transportation, payroll pro-
cessing fees, and insurance payments: An individual reimbursed for mileage
might be staff, but an individual paid to cater an event labeled “staff event”
would be excluded. Third, I build decision rules for classifying personnel as
staff or consultants based on a combination of entity type and payment pur-
pose. Lastly, I manually match personnel names, given common (and possibly
intentional) irregularities in the spelling of names so as to track individuals
from one campaign cycle to another. The result is a preliminary dataset of
132,113 campaign staff members and consultants reported to the FEC as hav-
ing received payment in a congressional campaign from 2003 to 2016. To my
knowledge this is the only existing such database.
From that data I build two new measures: team experience and team
familiarity. Previous work on legislative staff experience tends to use total
experience of legislative staff. That is appropriate where the number of staff
in each legislative office is roughly equal, based on congressional budgets. In
89
campaigns, however, the number of personnel varies widely. As a result, in this
paper I measure mean team experience by finding the number of prior election
cycles worked by a particular staffer or consultant, then finding the average
experience of all staff and consultants from a particular campaign. Figure 4.1
shows the distribution of this measure against our two dependent variables,
colored by incumbency status.
This combined measure of staff and consultant experience is a useful one
but also hides variation in staff experience versus consultant experience. While
most political consultancies are small, one-person firms, many (see Chapter 3)
are larger, multiperson firms. Because the FEC does not require consulting
firms to disclose their own staff, this means that a large consulting firm’s
combined level of experience may be much greater than is reflected in this
combined measure of personnel experience. So in addition to this measure of
mean team experience, I build mean experiences measure for staff (excluding
consultants) and consultants (excluding staff). Given that fundraising is often
the responsibility of a specific set of staff or consultants I also build a specific
measure of experience for personnel involved in finance and fundraising.
The second new measure created here is of team familiarity. Related to
the “familiarity” conceptualized by Ruef et al (2003), prior management schol-
arship on teams has defined team familiarity as the amount of prior experience
that team members have with each other (Espinosa et al, 2007; Sieweke and
Zhao, 2015). For this paper, I construct team familiarity in a similar manner,
with one key difference. As with prior research, I first identify the number of
90
prior campaigns that every pair of personnel have worked on together. Then,
rather than taking the average of those measures, I divide that number by the
total number of possible unique connections among the team (N(N1)/2).
I do this because unlike basketball teams, political teams can have a very
large or very small number of team members. Figure 4.2 shows the distri-
bution of this measure against our two dependent variables, again colored by
incumbency status. Table 4.1 includes the descriptive statistics of both new
variables.2
Figure 4.1: Team Experience and...
2Two notes here: First, a handful of candidates in the DIME database won an improb-
able 200% percent of the vote; those observations are removed from the visualizations and
handled as outliers in the analysis. The same goes for extremely high values of team famil-
iarity which in some instances have to do with underlying data problems, and in others may
reveal candidates attempting to hide how much they are paying particular consultants.
91
Figure 4.2: Team Familiarity and...
Table 4.1: Summary Statistics: Team Experience and Familiarity
Statistic N Mean St. Dev. Min Max
Vote Share (%) 4,390 55.4 18.6 0.6 100.0
Total Receipts ( 100Ks) 4,390 12.3 12.3 0.0 220.2
Mean Team Experience 3,762 0.9 0.8 0.0 6.0
Team Familiarity 3,655 0.2 0.4 0.0 5.0
Note: Experience and familiarity measured by number of prior cycles worked.
4.4 Methods and Results
I test both of these independent variables against two dependent vari-
ables: The share of the vote won and the share of funds raised by a congres-
sional candidate. In these models vote share is the percentage (from 0 to 100)
of the two-party vote won by a candidate in the general election. Funds raised
92
is the total raised by a candidate’s election committee in the two years that
include election day.
One feature of this dataset is that its measures of experience and fa-
miliarity are right-censored. The FEC has digitized expenditure data back to
2003. If I were to include 2004 campaigns in this analysis, the experience level
of every campaign team member would be zero. As a result, I limit my analysis
to the 2012, 2014 and 2016 campaign cycles. This is an imperfect approach to
right-censoring. Of the roughly 90,000 personnel paid during these three elec-
tion cycles, 2.6% have at least six congressional cycles of prior experience. As
a result, we cannot know from this data whether those 1800 consultants and
550 staff have six campaign cycles’ worth of experience or twenty. This leaves
open the possibility that the results below underestimate the correlations of
team experience and familiarity with vote totals and fundraising amounts.
Another limitation that flows from having to restrict these models to
three cycles is the generalizability of the results to the period before 2012.
2012-2016 was a period of tremendous tumult on the Republican side of the
American political system, from the rise of the Tea Party after the 2008 elec-
tion through the 2016 election of Donald Trump. 3This change may have
materially impacted the network of Republican campaign personnel during
this period, although in interviews for Chapter 3, veteran Republican con-
sultants reported having adjusted quite well to the Republican Party’s new
3For a lay history of this period in the Republican Party, see Tim Alberta’s American
Carnage, [Alberta, 2019].
93
priorities ideologically and financially.
In order to identify whether this effect might vary based on campaign
type, I then run the same models on a series of candidate subsets: First by
party (Democrats and Republicans), then by candidate status (Incumbents
and Challengers), and finally with all the intersections of those two categories
(Democratic Challengers, Republican Incumbents, etc.). Excluded are can-
didates for open seats, which typically make up 10-12% of races. I exclude
races for open seats based on qualitative observations (see Chapter 3). While
incumbents and challengers consistently operate in distinct ways, candidates
for open seats make up a less coherent category. For example, some of those
candidates have competitive primaries, while others do not; some run in com-
petitive general elections, while others run in non-competitive elections they
are bound to win or lose by wide margins. I interpret these within-group dif-
ferences to mean that candidates for open seats merit separate and focused
study.
4.4.1 Vote Share, Team Experience, and Team Familiarity
To test the association between team experience and vote share, I first
rebuild approximations of prior models of vote share [Canes-Wrone et al.,
2002]. Those models explain a large amount of variation in vote share with a
small number of candidate-specific, electorate-specific and campaign-specific
variables. The candidate-specific measures include ideology, prior legislative
experience (sometimes described as “candidate quality”), share of spending,
94
seniority (specifically whether they are a freshman incumbent). The other
variables include voters’ change in personal income, whether the election year
saw a midterm wave in reaction against an incumbent president, presidential
popularity, and the share of votes won by a candidate’s party in the previous
presidential election.
The models below include a modification of one measure of candidates’
left-right ideology called a DIME scores, which are equivalent to the traditional
roll call vote-based DW-Nominate scores but are able to infer candidate ideol-
ogy based on their position in the network of candidates and donors. I further
modify the DIME scores, which are on a scale of -4 (most liberal) to 4 (most
conservative), to be a measure of ideological extremity on a scale of 0 (most
centrist) to 4 (most extreme). The model also includes prior presidential vote,
modeled as the share of the vote of the candidate’s party’s presidential can-
didate in the previous election in that candidate’s district;4candidate status
(incumbent or challenger); and the difference in between a candidate’s (logged)
spending and the (logged) spending of their general election opponent. In each
of the models I standardize all independent variables.
Table 4.2 includes six of the models run using this set of variables.
Model 1 is a base model. In keeping with political scientists’ push toward
more parsimonious models, it drops variables where they are not statistically
4Rather than interacting this variable by party, I use a modified version of this variable:
for Democrats, I use the measure as is. For Republicans, I modify the variable to be 1 n,
where nis the percent won by the Democratic candidate.
95
significant and do not meaningfully affect the models’ performance [Achen,
2002]. Model 2 adds the team experience variable, which is positively corre-
lated with vote share. Model 3 drops that variable in favor of the consultant
experience variable, which is also positively associated with vote share, and
which accounts for the correlation of the team experience variable. Not shown
here is that the staff experience variable is not associated with vote share.
Model 4 includes the team familiarity variable, on its own not associated with
vote share. Model 5 incorporates both of these variables in the same model. In
this model, the positive association between team familiarity and vote share
washes out the Model 3 association between consultant experience and vote
share. This holds true in Model 6, which excludes outlier observations where
Di>4
n, where Diis Cook’s Distance and nis the number of observations in
the model.
Each of the models include election-cycle fixed effects, in order to
account for presidential popularity, personal income growth, and midterm
waves.5. I also ran models including district effects, but those models lacked
power. Also tested was a squared form of the study variables, in order to test
for u-shaped relationships between those variables and vote share.
Table 4.3 includes three models in which I investigate the effect of these
variables by party and incumbency. Model 1 includes a new variable that splits
the sample into four categories: Democratic challengers (the referent category
5In Model 6, I use year dummies rather than fixed effects, in order to extract outliers.
96
in the model), Democratic incumbents, Republican challengers and Republi-
can incumbents. According to this model, in 2012-2016 both incumbents and
Republicans performed better than Democrats or challengers. Model 2 inter-
acts this variable with our two study variables. For Republican incumbents
(but no other group), vote share is positively associated with consultant ex-
perience but not familiarity. This result holds in Model 3, in which outlier
observations are excluded.
97
Table 4.2: Vote Share, Team Experience, and Team Familiarity
Dependent variable:
Vote Share (%)
(1) (2) (3) (4) (5) (6)
Spending Advantage 6.40∗∗∗ 6.31∗∗∗ 6.33∗∗∗ 6.34∗∗∗ 6.22∗∗∗ 4.22∗∗∗
(0.27) (0.29) (0.29) (0.30) (0.31) (0.24)
Prior Pres. Vote 8.09∗∗∗ 8.22∗∗∗ 8.19∗∗∗ 8.25∗∗∗ 8.36∗∗∗ 9.86∗∗∗
(0.31) (0.34) (0.33) (0.35) (0.36) (0.25)
Ideological Extremity 0.99∗∗∗ 1.18∗∗∗ 1.05∗∗∗ 1.15∗∗∗ 1.32∗∗∗ 1.26∗∗∗
(0.22) (0.24) (0.24) (0.25) (0.27) (0.18)
Democratic Incumbent 8.42∗∗∗ 7.05∗∗∗ 8.07∗∗∗ 7.26∗∗∗ 6.87∗∗∗ 6.01∗∗∗
(0.75) (0.80) (0.81) (0.80) (0.84) (0.55)
Republican Challenger 3.48∗∗∗ 3.38∗∗∗ 3.88∗∗∗ 3.41∗∗∗ 3.84∗∗∗ 4.68∗∗∗
(0.69) (0.77) (0.74) (0.80) (0.84) (0.54)
Team Experience 0.49∗∗
(0.24)
Consultant Experience 0.46∗∗ 0.27 0.06
(0.23) (0.26) (0.17)
Team Familiarity 0.36 0.51∗∗ 0.63∗∗∗
(0.22) (0.26) (0.18)
Republican Incumbent 0.69 1.09 0.28 0.99 0.59 0.47
(0.84) (0.93) (0.90) (0.95) (1.00) (0.64)
Observations 1,725 1,445 1,566 1,409 1,286 1,169
R20.80 0.80 0.80 0.79 0.78 0.89
Adjusted R20.80 0.79 0.80 0.79 0.78 0.89
Residual Std. Error 4.89
F Statistic 1,167.12∗∗∗ 799.32∗∗∗ 868.51∗∗∗ 743.44∗∗∗ 580.67∗∗∗ 983.67∗∗∗
Note: Year factors and intercept suppressed.
Referent = Democratic Challenger
98
4.4.2 Fundraising, Team Experience, and Team Familiarity
The amount of money candidates raise and spend is a key—if complicated—
predictor of electoral success, as shown in prior work and in Table 4.2. In light
of this, and in order to further examine the effects of team experience and
team familiarity, the models in Table 4.4 examine the association between
these measures and the total amount of money raised by a campaign.
Prior models on the determinants of fundraising success incorporate
party affiliation and the competitiveness of the race [Herrnson, 1992]; and
opponent receipts, opponent quality, cash on hand and whether or not a can-
didate faced a primary challenger [Krasno et al., 1994]. Model 1 in Table 4.4
is a base model built with a majority of these traditional variables or proxies
for them. I also tested inclusion of three additional control variables: the ideo-
logical extremity measure; the number of distinct donors a campaign receives;
and the dollar contributions of a candidate to their own campaign (which are
not limited by law). Model 2 adds in the mean experience of all personnel fo-
cused on finance and fundraising. Not shown here are models that tested total
team experience, staff experience, and consultant experience, none of which
were positively correlated with funds raised. Model 3 includes the team famil-
iarity variable, not associated with fundraising ability. Model 4 and Model 5
include fundraising personnel experience and team familiarity, with Model 5
excluding outlier observations according to the same Cook’s distance cutoff as
in Table 4.2. In Models 2 and 4, fundraising the experience level of fundraising
personnel is positively associated with fundraising totals, but this effect goes
99
Table 4.3: Vote Share, Team Experience, and Team Familiarity by Party and
Incumbency
Dependent variable:
Vote Share (%)
(1) (2) (3)
Spending Advantage 6.40∗∗∗ 6.21∗∗∗ 4.22∗∗∗
(0.27) (0.31) (0.24)
Prior Pres. Vote 8.09∗∗∗ 8.42∗∗∗ 10.16∗∗∗
(0.31) (0.36) (0.26)
Ideological Extremity 0.99∗∗∗ 1.31∗∗∗ 1.23∗∗∗
(0.22) (0.27) (0.19)
Democratic Incumbent 8.42∗∗∗ 6.88∗∗∗ 5.74∗∗∗
(0.75) (0.97) (0.70)
Republican Challenger 3.48∗∗∗ 3.25∗∗∗ 5.70∗∗∗
(0.69) (1.18) (1.21)
Republican Incumbent 12.59∗∗∗ 11.46∗∗∗ 11.21∗∗∗
(0.62) (0.86) (0.62)
Consultant Experience 0.77 0.73
(0.64) (0.45)
Team Familiarity 1.02 0.81
(1.31) (0.98)
Consultant Experience (Dem Inc) 1.06 0.83
(0.79) (0.56)
Consultant Experience (GOP Ch) 1.33 1.04
(1.04) (0.77)
Consultant Experience (GOP Inc) 1.361.26∗∗
(0.76) (0.53)
Team Familiarity (Dem Inc) 0.04 0.36
(1.37) (1.04)
Team Familiarity (GOP Ch) 2.72 0.70
(1.85) (2.29)
Team Familiarity (GOP Ch) 0.92 0.57
(1.36) (1.02)
Observations 1,725 1,286 1,181
R20.80 0.79 0.89
Adjusted R20.80 0.79 0.89
Residual Std. Error 7.88 5.16
F Statistic 1,167.12∗∗∗ 299.00∗∗∗ 569.05∗∗∗
Note: p<0.1; ∗∗p<0.05; ∗∗∗ p<0.01
100
away once outlier observations are removed in Model 5. And in Model 5, team
familiarity is negatively associated with funds raised.
Table 4.5 includes three models examining the effect of these variables
by party and incumbency. Model 1 includes the new party-incumbency cate-
gorical variable introduced above in Table 4.3, and begins with outlier obser-
vations removed as in Table 4.4 Model 5. According to Model 1, in 2012-2016
incumbents raised more money than challengers. There was no statistically sig-
nificant difference between Republican and Democratic challengers’ fundrais-
ing, but Republican incumbents fared slightly better than Democratic ones in
this period.
Model 2 shows results for interacting team familiarity with party. In
Model 2, team familiarity is associated with fundraising negatively for Democrats
(the reference group) and positively for Republicans. Model 3 shows results
for interacting fundraising team experience with incumbency. In Model 3,
fundraising experience is negatively associated with the amount of funds raised
for incumbents but not for challengers. Models showing other non-significant
interactions of these variables are available in Appendix C.
101
Table 4.4: Fundraising, Team Experience, and Team Familiarity
Dependent variable:
Total Receipts (100Ks)
(1) (2) (3) (4) (5)
Prior Pres. Vote 0.07∗∗∗ 0.12∗∗∗ 0.07∗∗ 0.12∗∗∗ 0.12∗∗∗
(0.03) (0.04) (0.03) (0.04) (0.02)
Democratic Incumbent 0.40∗∗∗ 0.39∗∗∗ 0.38∗∗∗ 0.36∗∗∗ 0.45∗∗∗
(0.07) (0.10) (0.08) (0.10) (0.06)
Republican Challenger 0.03 0.03 0.00 0.03 0.02
(0.07) (0.10) (0.08) (0.11) (0.06)
Opponents’ Receipts 0.21∗∗∗ 0.24∗∗∗ 0.25∗∗∗ 0.21∗∗∗ 0.32∗∗∗
(0.02) (0.03) (0.03) (0.03) (0.02)
Candidate Contributions 0.19∗∗∗ 0.22∗∗∗ 0.18∗∗∗ 0.22∗∗∗ 0.22∗∗∗
(0.02) (0.03) (0.02) (0.03) (0.04)
No. Distinct Donors 0.49∗∗∗ 0.44∗∗∗ 0.54∗∗∗ 0.49∗∗∗ 0.59∗∗∗
(0.02) (0.02) (0.02) (0.03) (0.02)
Fundraisers’ Experience 0.06∗∗ 0.06∗∗ 0.00
(0.03) (0.03) (0.02)
Team Familiarity 0.02 0.04 0.03∗∗
(0.02) (0.03) (0.02)
Republican Incumbent 0.26∗∗∗ 0.210.23∗∗ 0.17 0.10
(0.08) (0.11) (0.10) (0.13) (0.06)
Observations 1,729 1,191 1,411 1,025 976
R20.39 0.34 0.40 0.36 0.56
Adjusted R20.39 0.34 0.39 0.35 0.55
Residual Std. Error 0.41
F Statistic 155.89∗∗∗ 76.88∗∗∗ 115.62∗∗∗ 63.01∗∗∗ 110.72∗∗∗
Note: Year factors and intercept suppressed.
Referent = Democratic Challenger
102
Table 4.5: Funds Raised, Team Experience, and Team Familiarity by Party
and Incumbency
Dependent variable:
Funds Raised (100Ks)
(1) (2) (3)
Democratic Incumbent 0.42∗∗∗
(0.05)
Republican Challenger 0.05
(0.05)
Republican Incumbent 0.56∗∗∗
(0.05)
Republican 0.060.06∗∗
(0.03) (0.03)
Team Familiarity 0.07∗∗∗ 0.03
(0.02) (0.02)
Fundraiser Experience 0.00 0.04
(0.02) (0.03)
Incumbent 0.49∗∗∗ 0.46∗∗∗
(0.05) (0.05)
Republican * Team Familiarity 0.06∗∗
(0.03)
Incumbent * Fundraiser Experience 0.06
(0.03)
Observations 1,679 975 975
R20.46 0.56 0.56
Adjusted R20.46 0.55 0.55
Residual Std. Error 0.41 0.41
F Statistic 206.64∗∗∗ 110.10∗∗∗ 109.78∗∗∗
Note: Control variables suppressed.
103
4.5 Discussion
In this paper, I hypothesize four sets of relationships between two char-
acteristics of campaign personnel (mean experience and team familiarity) and
two measures of campaign performance (funds raised and vote share earned).
4.5.1 Team Experience
First, I hypothesized mean team experience would have a straightfor-
ward relationship with vote share: more experience should develop more skill
at the job, which should translate to more money raised and greater vote share.
This proved true only when looking specifically at the experience levels of con-
sulting firms, and then only by consulting firms working for Republican incum-
bents. Otherwise, controlling for team familiarity washed out any association
between experience and vote share. Having more experienced non-consulting
staff was not associated with greater vote share in any model configuration,
or for any cohort of candidates. On balance, hiring experienced personnel did
not prove to be associated with better performance at the ballot box, though
it is possible that consultant experience is underestimated because the FEC
data does not show the individuals working within consulting firms.
Contrary to expectations, neither team experience, nor staff experi-
ence, nor consultant experience was positively associated with fundraising
totals. The experience of fundraising staff was positively associated with
fundraising totals, but that effect fell below statistical significance once outlier
observations were excluded. As with other measures of consultant experi-
104
ence, it is possible this effect size would be greater given more visibility into
the total experience housed within larger consulting shops. But interviews
for Chapter 3 suggest that fundraising-focused consulting firms are nearly al-
ways small, one-person firms. Again, on balance hiring experienced personnel,
even fundraising-specific personnel, did not prove to be associated with higher
fundraising totals.
In terms of cohort-specific hypotheses, I expected to see stronger rela-
tionships between team experience and fundraising for Republican candidates,
relative to Democratic ones, and for incumbent candidates relative to chal-
lenger ones. My models did not support these hypotheses. In fact, for incum-
bents fundraising team experience was negatively associated with fundraising
totals.
4.5.2 Team Familiarity
Controlling for different types of team experience, team familiarity was
positively associated with vote share. It did not have the u-shaped relationship
I hypothesized, nor did its effect vary by type of candidate. When included
in models with measures of team experience, it also absorbed much of the
association that might have been attributed to team (or in this case consultant)
experience.
One puzzle in the fundraising models is team familiarity’s negative as-
sociation with fundraising totals. Interacting familiarity with party shows this
negative association to be driven by Democratic candidates. For Republicans
105
candidates, in contrast, team familiarity and fundraising are positively associ-
ated.
These results present a number of additional puzzles. First, why is team
experience not associated with campaign performance for most candidates?
Second, why would Republican incumbents be the only group to benefit from
more experienced consultants? Third, why would team familiarity be associ-
ated with higher fundraising totals for Republicans but not for Democrats?
4.5.3 Moving Up the Ladder
Interviews with congressional candidates and personnel suggest a few
possible answers to the question of why more experienced campaign teams do
not seem to earn candidates more money or votes than inexperienced ones.
First, campaign work may be more straightforward and technical compared
with most market settings: there is a single measure of success and a bounded
set of potential activities repeated every two years. In this context, perhaps
executing on a known set of activities requires less experience than in other
industries. Second, as discussed in the Chapter 3, perhaps a fundamental
uncertainty about which campaign activities lead to which outcomes means
that more experience does not lead to greater skills or knowledge. As one
interviewee put it to me, “If you stay in, you move up the ladder, regardless of
if you have the skills or capabilities to do that.” This leaves candidates unable
to use experience as a proxy for quality. It also means campaign experience
does not confer on campaign operatives the ability to identify candidates who
106
will outperform expectations.
Why would consultant experience be associated with greater vote share
only for Republicans incumbents? In Chapter 3, I explored the differences be-
tween Democratic and Republican labor markets, which are almost completely
separate networks. The Democratic labor market operates much more like a
traditional labor market than the Republican one: whereas Democratic candi-
dates hire individual staff, often with support of extended party institutions,
Republican campaigns are often run by consulting shops that place their own
employees as staff on congressional campaigns and, therefore, gather institu-
tional memory and knowledge inside of these consulting shops. This position
also enables them to select high-quality candidates as clients. It stands to
reason, too, that the benefit of this knowledge concentration in private firms
would accrue to incumbent Republicans, rather than challenger candidates,
who are a less guaranteed income stream to the consultants.
On their own, these models cannot specify the direction of causality,
a limitation of much of the research on Congress [Binder, 2020]. Perhaps, as
hypothesized, consultant experience makes incumbent Republican campaigns
more effective. But perhaps high-quality Republican incumbents more effec-
tively select personnel, or experienced Republican consultants gravitate toward
high-quality candidates. Considered in light of results from qualitative inter-
views, however, the results above provide evidence against both of these claims.
First, if Republican candidates are selecting experienced personnel in pursuit
of greater performance, we would expect to see challengers hiring to maxi-
107
mize vote share and (mostly safe) incumbents hiring to maximize fundraising.
Second, if consultants are choosing candidates because of a belief they will
outperform, we should see them choosing to work for challengers who will out-
perform in vote share (and potentially flip a race) and incumbents that will
raise more money.
4.5.4 The Devil You Know
The positive association between team familiarity and vote share is
the clearest and most persistent association in any of the models presented
here. Why would this be, when team experience does not have the same
association? Perhaps the latter actually leads to the former: in a labor market
where experience cannot be used as a signal of personnel quality, building
a team based on prior relationships is a rational strategy. If candidates are
broadly unable to identify high-quality personnel by looking at the experience
level of applicants, they can at least screen personnel based on references from
other operatives they have worked with. And the presence of an ongoing
relationship between two team members is more than just reference checking:
if two staffers were able to demonstrate the capacity to work together before
and want to work together again, that suggests a lower likelihood of distracting
intra-team conflict. Given the temporary nature of campaigns, which are
dominated by uncertainty about which tasks lead to which outcomes, this
may be the effective hiring strategy. Or as one member of the Daley political
machine of the mid 20th century put it, candidates “don’t want nobody nobody
108
sent” [Rakove, 1979].
There are other interpretations that the above models cannot rule out.
Familiar groups of personnel might jointly identify and select high-quality
candidates, or some characteristic of high-quality candidates leads them both
to select more familiar teams and also, separately, to outperform. The former
is possible but would require a level of coordination among personnel not
suggested in the ad hoc hiring patterns revealed in Chapter 3. The latter
stands more to reason: high quality candidates may be hiring personnel with
shared work background not strategically, but because it is the path to hiring
a team that requires the least amount of work. This would allow them to
focus on other, more critical activities. In qualitative interviews, however,
nearly all candidates reported hiring through their networks in this way—not
just successful or high quality candidates. If team familiarity is not a driver of
campaign outcomes, we should expect the most successful candidates to engage
in that behavior less, not more. Assuming neither of these are true, we are
left with the more straightforward explanation hypothesized above: having a
more familiar team leads candidates for U.S. House elections to perform better
at the ballot box.
The third puzzle in these results is why team familiarity would be as-
sociated with greater fundraising totals for Republicans, but not Democrats.
Returning again to the structure of Republican versus Democratic campaign
labor markets, interviews from Chapter 3 suggest a potential answer. Demo-
cratic candidates largely raise money by dialing for dollars from lists of poten-
109
tial donors provided by the party or by candidate support groups. Over time,
the candidate becomes the primary point of contact for those relationships,
making them less dependent on their team for fundraising. Republican can-
didates depended more on external fundraising consultants, who continually
broker candidates’ relationships with donors–one candidate interviewed in that
chapter, in fact, knew all of the major Texas-based fundraising consultants by
name, having hired (and sometimes fired) each of them in prior election cy-
cles. If both of these hold true across the country, I interpret this to mean
Republican candidates benefit by keeping those fundraising professionals on
their team, whereas Democratic candidates suffer the cost over time of having
in-house finance directors who do not continually build relationships with new
donors the way Republican fundraising consultants do.
4.5.5 Conclusions, Contributions, and Future Work
Considered together, the models above show that it is team familiar-
ity, not campaign experience, that is associated with campaign performance,
as measured in votes earned and, for Republican candidates, money raised.
In an opaque labor market without clear markers of individual performance,
candidates cannot use basic signals (such as experience) to get an edge in elec-
tions. Nor are more experienced personnel able to identity campaigns led by
candidates who will outperform expectations. Candidates must instead use
decision-making shortcuts (such as getting personnel to hire people they have
worked with before) in order to get a boost in performance.
110
These results contribute primarily to political science literature on elec-
tions. The majority of empirical work on U.S. elections revolves around ques-
tions of campaign finance, incumbency advantage, ideological alignment, and
electoral outcomes. By focusing on campaign teams, this work expands our
understanding of who participates in elections, and when and how they influ-
ence election outcomes.
In doing so, this paper also contributes to a small but growing literature
on the importance of staff in Congress and the public policy process. Dom-
inant frameworks for understanding the public policy process have explicitly
structural roots. Information processing theory, for example, uses as its moti-
vating imagery earthquakes and punctuated equilibria. This imagery does not
lend itself to asking questions about the motivations and behaviors of indi-
vidual tectonic plates. But the system information processing theory seeks to
explain is full of actors making individual choices. Understanding what drives
those individual choices will give us insight into not just the structure, but the
agency of policy change.
These results are just a first step in an examination of campaign per-
sonnel networks. One potential area of future work is to focus on personnel’s
familiarity with the candidate, rather than with each other. Another line of
work could examine more closely the varied roles of staff versus consultants,
and especially the difference in their shared work histories. A third line of work
could extend beyond the campaigns narrowly defined. In many races, Super
PAC spending outpaces that of individual candidates. These organizations
111
are legally at arm’s length from campaigns but are often run by candidates’
longtime associates.
Future work could also explore a number of campaign personnel and
team characteristics, including the gender composition and local versus na-
tional roots of personnel, and their centrality in the network of political actors.
Familiarity, rather than experience, is the path for most political hiring. This
process further entrenches of the elite social networks that have historically
kept women, people of color, and other groups out of power. Other lines of
future work might include: the causes and consequences of organizational pro-
cess variables such as personnel turnover, mobility, and pay; the incorporation
of candidates’ early fundraising totals; and open seat candidates specifically,
which are more likely to be running for truly competitive seats.
The models above also raise questions about some of the control vari-
ables they use. One is the consistently negative association between ideological
extremity and vote share, which suggests ideological extremity costs candidates
votes but not money. Another potentially puzzling result is the impact of dis-
trict lean—as measured by the prior presidential vote of a candidate’s party in
their district. As predicted, it is positively associated with vote share. But it
is negatively associated with fundraising. After a presidential candidate does
well in a district, it seems, it is easier for the opposing party to raise funds in
the following cycle.
Lastly, in Table C.3 in Appendix C, I show the results of including an
additional variable in the model predicting vote share: the number of unique
112
donors to a campaign. In this model, the number of unique donors is posi-
tively correlated with vote share. This might seem self-evident; high-quality
candidates should be able to raise more funds and win more votes. But this
model controls for spending differential between candidates, so the result hints
at the effectiveness of modern populist campaigning.
For students of democracy, understanding more about campaign per-
sonnel is important in two additional ways. First, campaign personnel not only
affect the outcomes of campaigns, but influence the agendas and effectiveness
of our democratic representatives. Second, volunteering or working on political
campaigns is a primary gateway into political participation. Learning more
about who works on campaigns, and just as importantly who does not, can
help us understand why some groups (such as women and ethnic minorities)
are less well-represented in a representative democracy than others. The aim
of this project is to begin to answer some of the very basic questions about
political campaign personnel in hopes of making campaigns, and the actions
of the policymakers they help elect, more transparent to the citizens they are
elected to represent.
113
Chapter 5
Hacks and Wonks: The Effects of Hiring
Campaign Staff into Congressional Offices
from 2005-2016
“Campaign staff don’t usually make good legislative staff. Cam-
paign staff is driven by newness of every campaign, the flow, the
energy...You’re always in a battle. The legislative side is slower,
more methodical. Except for communication side, it doesn’t usu-
ally translate well.” - A 2020 candidate
“It’s as simple as are they more of a people person or are they
someone that would want to sit in an office.” - A 2020 campaign
manager
5.1 Abstract
At the end of a successful electoral campaign, a winning candidate
is rewarded with a new challenge: What happens to the organization they
built to win the campaign, and how should they staff their legislative office?
Some policymakers solve both problems at a stroke by hiring their political
campaign staff onto their legislative teams. This paper examines the effects of
such staff crossover on U.S. House Members’ legislative effectiveness and vote
114
share in subsequent elections. Contrary to lay expectations, I find that from
2005-2016, members of Congress who hired more campaign staff into their
congressional offices were more effective legislators and performed no better
or worse in subsequent campaigns. This paper extends recent scholarship on
legislative staff by examining the campaign staff that cross over into legislative
offices, and their effect on policymakers’ effectiveness and electoral success, two
dimensions of performance studied by political scientists and policy scholars.
5.2 Introduction
When they wake up the day after the election, candidates have a mighty
managerial task ahead of them: Of the people who helped get them elected,
whom will invite to join their legislative office? There is very little guidance
available to newly elected representatives beyond conventional wisdom, which
is to be careful about hiring campaign staff. Political hacks, so the common
wisdom goes, make bad policy wonks [Medvetz, 2010].
And yet some members of Congress regularly ignore this bit of advice.
Instead, they reward campaign staff by hiring them into their new office. From
2005-2016, at least one out of every eight Hill staffers had been paid during the
previous election campaign cycle. Candidates’ willingness to regularly ignore
this standard advice offers us the opportunity to examine whether that con-
ventional wisdom is true. Does hiring campaign staff into their congressional
offices help or hurt legislators?
115
5.3 Theory and Hypotheses
5.3.1 Legislative Staff in the U.S. Congress
Political science research on Congress tends toward one of two levels
of analysis: Some scholars examine Congress as a whole, examining its pro-
ductivity [Cox and Terry, 2008], polarization [Theriault, 2008], and issue focus
[Jones et al., 2019]. Other scholars focus on individual legislators, investigat-
ing their effectiveness [Volden and Wiseman, 2014], susceptibility to financial
influence [Lax et al., 2019], voting patterns [Poole and Rosenthal, 2011], and
attention [Sulkin, 2005].
Legislators do not act alone, however, nor are the institutions in which
they operate their only influence: They are embedded in overlapping webs of
relationships and organizations—committees, parties, and their own offices—
continually pulling at their attention, working to change their minds, and en-
abling or constraining their ability to act. In recent years, political scientists
have begun to spend more time examining the effects of these relationships,
including with campaign donors, staff and consultants; congressional commit-
tees, and constituents [Ringe et al., 2017, Victor et al., 2018, Box-Steffensmeier
et al., 2019].
One set of relationships in which legislators are deeply embedded is also
one they have the greatest hand in constructing: Their own legislative office.
When scholars of Congress have shone the empirical spotlight on legislative
staff, they have shown how staff have an impact on legislators’ effectiveness
and policy positions [Crosson et al., 2018, Hertel-Fernandez et al., 2019]. It is
116
the staff that respond to constituents, draft legislation, schedule the legislator’s
time, interact with media, and help determines the issues the legislator will
focus on—they are critical actors in the creation of public policy [Romzek and
Utter, 1997, Weissert and Weissert, 2000]. And yet only recently have political
scientists begun to demonstrate just how important those staff members are to
democratic representation and legislative effectiveness [Hertel-Fernandez et al.,
2019, McCrain, 2018, Montgomery and Nyhan, 2017].
Historically, political scientists have tended to understand legislative
staff as imperfect extensions of the legislator, whose behavior could be modeled
with a principal-agent approach [Romzek and Utter, 1997, Rosenthal and Bell,
2003]. But as Bell and Rosenthal argue and others have demonstrated, “on a
day-to-day basis, there is nothing that members do in the modern legislature
that is not affected by their staffs” [Montgomery and Nyhan, 2017, Rosen-
thal and Bell, 2003, Rundquist et al., 1992]. A recent wave of scholarship is
taking these actors seriously, finding that Representatives hire staffers in part
to exploit their professional networks [Burgat, 2020], and (at the state level)
greater legislative professionalization leads to less racial discrimination [Land-
grave and Weller, 2020]. Most research on legislative staff, however, presents
policy problems yet to be solved. Congressional staff capacity is on the decline
[LaPira et al., 2020], and one study argues that staff collect and present polit-
ically biased information to their representatives [Furnas, 2019]. Staff are not
all treated equally, either: female staffers are systematically underpaid [Mc-
Crain and Palmer, 2019, Calcagno and Montgomery, 2020], and minority staff
117
are disproportionately placed in constituent service roles [Ziniel, 2020]. This
paper extends that nascent work by examining campaign staff who cross over
into legislative offices, and their effect on policymakers’ effectiveness and elec-
toral success—two dimensions of performance studied by political scientists
and policy scholars.
Introduced by Volden and Wiseman in a 2009 working paper and then
codified in a 2014 book and 2013 article, Legislative Effectiveness Scores are a
measure of a legislator’s ability to introduce substantive legislation and shep-
herd it through the legislative process [Volden and Wiseman, 2009, Volden
et al., 2013, Volden and Wiseman, 2014]. Scholars have since used these scores
to underscore structural differences between the House and Senate [Volden and
Wiseman, 2018]; to measure the importance of committee chairs [Lewallen,
2020]; to quantify the impact of social connections on legislative effectiveness
[Battaglini et al., 2020]; and to demonstrate that legislators anticipating com-
petitive primary elections are less effective legislators [Barber and Schmidt,
2019]. Another line of work is building alternate measures of effectiveness by,
for example, including the progress of ”hitchhiker” bills in updated effective-
ness scores [Casas et al., 2020]. The second set of models in this paper builds
on work predicting legislative electoral outcomes, as introduced and extended
in Chapter 4.
118
5.3.2 Hypotheses
Lay theories of legislative staff suggest campaign staff make bad leg-
islative staff because the roles require different abilities; campaign staff may
struggle to switch into a new task environment. Offices with more politi-
cal staff, then, will be less likely to excel at getting bills pushed through
Congress, measured by Legislative Effectiveness Scores [Volden and Wiseman,
2014]. These campaign crossover staff may also not prioritize passing legis-
lation. Political ”hacks” love the electoral game, lay theory suggests, while
policy ”wonks” relish in the finer details of policy. Offices with more cam-
paign crossover staff may be more likely to focus on constituent services, in
hopes of better supporter turnout in subsequent elections. Or they may focus
on hot-button issues with low probability of becoming legislation but which
have high signal value to partisan voters. Crossover staff, in short, may keep
legislators’ attention on re-election rather than passing policy.
Hypothesis 1: The more campaign staff a legislative office brings
on board, the less effective that legislator will be at pushing bills
through the legislative process.
If these lay theories are right, the result of this shift in attention from
passing legislation to getting re-elected may not make for more effective legis-
lators. But if those same crossover staff keep the legislator focused on electoral
issues, that should make for candidates that are more in touch with their con-
stituents and therefore more competitive in subsequent elections:
119
Hypothesis 2. The more legislators hire campaign staff into their
legislative offices, the more successful they will be in their next
campaign.
5.4 Data
Answering both of these questions requires construction of multiple
novel datasets from a variety of sources, including the Federal Election Com-
mission; the Legistorm.com database of congressional staff histories; Volden
and Wiseman’s legislative effectiveness scores data; and Adam Bonica’s datasets
of congressional fundraising, ideology and electoral outcomes.
5.4.1 Congressional Staff Data
In order to build measures of staff crossover from the campaign trail
to Congress, I purchased data on congressional staff from Legistorm.com, a
private company that collects and provides historical data on members of
Congress and their staff. Using that database, I build measures of total, maxi-
mum and average staff experience for each congressional office in each session.
5.4.2 Building a Measure of Staff Crossover
I then combine this legislative staff database with the campaign staff
database described in the previous chapter (see Section 4.3.2). That allows
me to construct the following two measures: total crossover, a measure of the
number of staff members hired into a legislative office from the campaign trail;
120
and percent crossover, the percentage of staff members in a legislative office
from the campaign. The resulting data set includes 2,604 observations at the
Representative-congress level, spanning from the 109th Congress (2005-2006)
to the 114th (2015-2016); Table 5.1 shows basic descriptive statistics for these
variables.
Table 5.1: Summary Statistics: Staff Experience and Crossover
Statistic N Mean St. Dev. Min Max
Legislative Effectiveness Score 2,604 1.01 1.31 0.00 18.69
Vote Share (%) 2,604 67.43 12.99 19.90 100.00
Crossover Staff (n) 2,604 4.97 4.67 0 32
Crossover Staff (%) 2,604 11.59 10.20 0 87
Total Staff (n) 2,604 43.89 21.88 2 200
Staff Mean Experience 2,604 2.66 1.26 0.06 9.00
Note: Experience measured in years
5.4.3 Legislative Effectiveness
In order to examine the relationship between staff crossover and leg-
islative effectiveness, I combine the crossover data set above with Legislative
Effectiveness Scores (LES) data available from the Center for Effective Law-
making. The dependent variable in these models is their Legislative Effec-
tiveness Score, which is a measure of an individual legislator’s ability to “ad-
vance her agenda items through the legislative process and into law,” tracking
each bill’s progress through Congressional hearings and committees.1In their
1For a detailed description of Legislative Effectiveness Scores, see the CEL’s ”Methodol-
ogy” page, located here: https://thelawmakers.org/methodology.
121
models, Volden and Wiseman explain over half of the variation in legislative
effectiveness with eighteen variables, each of which are included here. The
most important, which the authors build into benchmark scores for each leg-
islator, are seniority, majority party membership, and holding the chair of a
committee or subcommittee.
Figure 5.1 below maps CEL’s legislative effectiveness scores against the
new staff crossover measure, with each point representing a Representative in a
given Congress, colored according to whether they hold an important commit-
tee or subcommittee chair. In this plot, both crossover and LES are logged for
ease of interpretation. The plot demonstrates no obvious relationship between
crossover and legislative effectiveness.
Figure 5.1: Staff Crossover and Legislative Effectiveness
122
5.4.4 Electoral Outcomes
To examine the relationship between staff crossover and electoral out-
comes, I combine the crossover database with data from the FEC and MIT
Elections Lab. The primary outcome variable here is the share of the two-party
vote that a candidate wins in a general election. The primary study variable
here is the amount of former campaign staff an incumbent had in their legisla-
tive office in the run-up to that election. For example, if a candidate wins an
election in 2012, they may then choose to hire campaign staff into their leg-
islative offices for the 2013-2014 Congress. The percentage of that 2013-2014
legislative staff who worked the 2012 election cycle becomes the independent
variable staff crossover. The dependent variable, in this case, becomes that
candidate’s vote share in the 2014 cycle.
Prior work has shown that a significant amount of variation in vote
share can be explained by three factors: candidate status (incumbent, chal-
lenger, or running for an open seat); the share of the two-party vote won in
a district by the previous Democratic presidential candidate; and candidate
party.
Figure 5.2 below plots crossover data against vote share, with Repre-
sentatives colored according to whether they were first term (freshman) rep-
resentatives in the previous cycle. As with legislative effectiveness, there is no
visually obvious relationship between crossover and subsequent vote share; it
is apparent, however, that freshman legislators tend to have more crossover
staff than congressional veterans.
123
Figure 5.2: Staff Crossover and Vote Share
5.5 Methods and Results
For both of these dependent variables, I employ the same method,
building up a base model from prior research on that dependent variable,
adding in staff crossover variables and relevant controls into multiple models,
and then running robustness checks for each.
5.5.1 Staff Crossover and Legislative Effectiveness
The models in Table 5.2 have as their dependent variable the Legislative
Effectiveness Score (LES) of a House representative in a particular congress
(the 115th Congress, 116th Congress, etc.). This score generally ranges from 0-
5. The base model includes the variables that prior research has demonstrated
124
are associated with legislative effectiveness scores, the most important of those
being seniority, leadership, and holding the chair of a committee or important
subcommittee. In line with Volden and Wiseman’s models, these models also
include as a control variable a Member’s Legislative Effectiveness Score from
the previous congress. I standardize these variables, then add fixed two-year
effects. As in Chapter 4, and in keeping with the push by political scientists
toward more parsimonious models, it drops variables included in original LES
models that are not statistically significant and do not meaningfully affect the
models’ performance [Achen, 2002].
The second model adds in new variables, including the study variable:
the percent of a legislator’s staff who were paid campaign staffers in the pre-
vious election cycle. As additional controls, the model also includes variables
that are correlated with staff crossover and might serve as alternate expla-
nations for the correlation between our staff crossover and legislative effec-
tiveness. Those variables include the total number of people on a legislators’
payroll, the average federal legislative experience of that staff, whether that
legislator is a freshman, and (given results from Chapter 4) the prior shared
legislative experience of the congressional team, or team familiarity. Of note
in these models is that for legislative staff, team familiarity is negatively as-
sociated with legislative effectiveness. The more shared history a legislative
team has, the less effective their legislator. The third model removes these
additional controls, and the fourth model removes outlier observations from
the model. Models 2 and 3 suggest that having more crossover staff in a leg-
125
islative office is positively correlated with legislative effectiveness; once outlier
observations are removed in Model 4, however, that effect goes away.
5.5.2 Staff Crossover and Electoral Outcomes
The models in Table 5.3 have as their dependent variable the share
of the two party vote in the general election that occurs every two years for
U.S. House members. The base model is limited to a set of standard control
variables used in the study of congressional elections: a candidate’s spending
advantage, the share of the two-party vote won by the candidate’s party in
the prior presidential election, how ideologically extreme a candidate it (as
measured by roll call votes), and a dichotomous variable for the candidate’s
party.
In line with prior work, these structural factors explain the majority of
variation in electoral performance [Jacobson, 2015]. The second model adds in
the study variable, the percentage of their legislative office were campaign staff
in the previous election cycle. Both models below include fixed election-cycle
effects, and reveal no relationship between the number of campaign staff that
a legislator pulls into their office and their electoral performance in subsequent
elections.
5.6 Discussion and Conclusion
This paper contributes to a nascent academic literature about legisla-
tive staff, which argues that to understand how legislative behavior and perfor-
126
Table 5.2: Staff Crossover and Legislative Effectiveness
Dependent variable:
Legislative Effectiveness Score
panel OLS
linear
(1) (2) (3) (4)
Lagged LES 0.35∗∗∗ 0.35∗∗∗ 0.35∗∗∗ 0.40∗∗∗
(0.02) (0.02) (0.02) (0.02)
Seniority 0.08∗∗∗ 0.12∗∗∗ 0.08∗∗∗ 0.00
(0.02) (0.03) (0.02) (0.02)
Prof. State Lege Experience 0.05∗∗ 0.05∗∗ 0.06∗∗ 0.00
(0.02) (0.02) (0.02) (0.01)
Majority Party Member 0.53∗∗∗ 0.53∗∗∗ 0.53∗∗∗ 0.36∗∗∗
(0.06) (0.06) (0.06) (0.03)
Committee Chair 2.21∗∗∗ 2.20∗∗∗ 2.22∗∗∗ 2.24∗∗∗
(0.10) (0.11) (0.10) (0.09)
Subcommittee Chair 0.29∗∗∗ 0.28∗∗∗ 0.29∗∗∗ 0.29∗∗∗
(0.06) (0.06) (0.06) (0.04)
Total Legislative Staff 0.01
(0.03)
Staff Mean Experience 0.02
(0.03)
Seniority: Freshman 0.07
(0.26)
Percent Crossover Staff 0.06∗∗ 0.06∗∗ 0.02
(0.02) (0.02) (0.01)
Staff Familiarity 0.08∗∗∗ 0.05∗∗∗
(0.03) (0.01)
Constant 0.57∗∗∗
(0.04)
Observations 2,374 2,363 2,374 2,216
R20.40 0.40 0.40 0.49
Adjusted R20.40 0.40 0.40 0.49
Residual Std. Error 0.60
F Statistic 260.93∗∗∗ 144.56∗∗∗ 225.03∗∗∗ 153.96∗∗∗
Note: p<0.1; ∗∗p<0.05; ∗∗∗ p<0.01
All coefficients standardized;
robust standard errors in parentheses.
127
Table 5.3: Staff Crossover and Subsequent Elections
Dependent variable:
Vote Share
(1) (2)
Spending Advantage 1.14∗∗∗ 1.12∗∗∗
(0.04) (0.04)
District Lean 0.46∗∗∗ 0.46∗∗∗
(0.02) (0.02)
Ideological Extremity 3.34∗∗∗ 3.13∗∗∗
(0.55) (0.55)
Party: Republican 0.80∗∗ 0.67∗∗
(0.34) (0.34)
Percent Crossover Staff 1.35
(1.63)
Observations 1,435 1,445
R20.70 0.69
Adjusted R20.70 0.69
F Statistic 830.29∗∗∗ 643.42∗∗∗
Note: p<0.1; ∗∗p<0.05; ∗∗∗ p<0.01
Robust standard errors shown;
Outliers removed.
128
mance it is necessary to understand the teams they build and manage. It tests
for the effects of staffing a legislative office with campaign professionals on
two different measures of political performance: Legislative effectiveness and
electoral competitiveness. I hypothesized that an increase in the percentage of
campaign staff a legislator carries into office will lead them to be less effective
legislators but more likely to win re-election than otherwise similar legislators.
Neither of these hypotheses were supported: An increase in crossover staff is
not associated with a decrease in legislative effectiveness or with an increase
in vote share for incumbent candidates running for re-election.
Why would this be? In terms of legislative effectiveness, both aca-
demic and lay theories suggest that hiring more campaign staff would lead
to lower levels of legislative effectiveness. This would happen in two ways:
First, campaign staff should struggle to switch to the new task environment of
legislating. As one campaign staffer put it in interviews for Chapter 3, work-
ing on the campaign versus in Congress “is about people people versus office
people.” Second, theory suggests that their focus would be not on passing
legislation but getting their bosses re-elected, which depends more on serving
constituents and, in the modern era, being seen as regularly fighting with the
other party. Legislative offices rich in campaign crossover staff, focused on
re-election, would spend more time communicating than legislating.
The results from the legislative effectiveness models, which do not show
that hiring more campaign staff leads to less effective legislators, call these
theories into question. There are three possible explanations. First, campaign
129
staff may be more capable of adapting to their new environment than we might
expect. The legislative task is a heavily relational one, and if campaign staff
are chosen primarily by familiarity (see Chapter 3), then the trustworthiness
for which candidates hire people on the campaign trail might serve them well
in a legislative setting.
Second, campaign and legislative environments may be less distinct
today than they used to be. In 1980, journalist Sidney Blumenthal pub-
lished The Permanent Campaign, giving an enduring name to the increas-
ingly blurry line between presidential governing and campaigning [Blumenthal,
1980]. Blumenthal’s book, according to political commentators Ornstein and
Mann, flowed naturally from the work of political scientist David Mayhew,
who posited Congress as a group of 435 individual “single-minded seekers of
reelection” [Mayhew, 1974, Ornstein, 2000]. Neither Mayhew nor Blumenthal
were arguing that campaign organizations were permanent. They were argu-
ing instead—Mayhew descriptively, Blumenthal disdainfully—that electoral
considerations are constant and continually supersede policymaking. Viewed
forty years later, after the rise of social media in the first two decades of the
21st century, Blumenthal’s work is applicable far beyond just presidential cam-
paigns. As one Member of Congress in 2021 wrote his fellow legislators: ”I
have built my staff around comms [communications] rather than legislation”
[Vesoulis, 2021]. One line of potential future work would examine the positions
into which candidates are placing campaign crossover staff, with the expecta-
tion that the emergence of politics-by-social-media, combined with the need
130
for continuous fundraising, may have made campaigning and legislating more
like one job than two.
Third, staff might be hired from the campaign into the legislative office
because newly elected candidates view them as having successfully completed
an extended job interview during the campaign. This was the hypothesis put
forward by Xandra Kayden in the late 1970s, that new legislators and select
staff they believe most trustworthy and/or competent based on the “testing
ground” of the campaign. [Kayden, 1978].
Together, I interpret these models to mean that by hiring campaign
staff into their legislative offices, congressional representatives can boost their
legislative effectiveness without suffering an electoral penalty. In Chapter 6,
I discuss the potential work suggested by these results, in combination with
those from Chapter 3 and Chapter 4. Three potential avenues for research,
however, bear mention here. First, paper examines only one type of crossover:
from campaign teams to congressional offices. There remains much work to be
done studying movement back and forth from the Hill to the campaign trail. A
second area for deeper focus is suggested by the statistically significant negative
association between legislative team familiarity and legislative effectiveness
(see the second model in Table 5.2). This initial association runs counter to
prior work on team familiarity, including the results in Chapter 4. A third
area of potential research would be to tie staff crossover explicitly to changes
in individual policymakers’ agendas, which in turn influence what issues are
given priority by Congress (see [Sulkin, 2011, Jones and Baumgartner, 2005]).
131
In an interview for Chapter 3 of this dissertation, a 2020 congressional
candidate shared their view on the relationship between campaign hacks and
policy wonks:
Campaign staff don’t usually make good legislative staff. Cam-
paign staff is driven by newness of every campaign, the flow, the
energy... You’re always in a battle. The legislative side is slower,
more methodical. Except for communication side, it doesn’t usu-
ally translate well.
This paper contributes to an recent wave of research about legislative
campaign staff [Hertel-Fernandez et al., 2019, Montgomery and Nyhan, 2017]
by examining this common, lay theory of politics. But this lay theory reflects
a gap in political science research as well, between those scholars who study
Congress and those who study campaigns. This paper builds on a smaller but
growing body of work examining the relationship between campaigning and
legislating [Sulkin, 2011, Hacker and Pierson, 2014]. That work examines how
issues cross back and forth from the campaign trail to the halls of Congress,
adding a new dimension to prior work that focused primarily on the influence
of campaign dollars on legislative outcomes. This dissertation demonstrates
that in addition to examining the flow of ideas and money from the campaign
trail to Congress, scholars should investigate the flow of teams and people,
too.
132
Chapter 6
Conclusion
“You’re only as good as the team you surround yourself with...you
kind of live and die by them.” - A 2020 Congressional candidate
“For far too long, progressives have underinvested in talent,” reads the
tagline of the website for Arena, a training program for progressive political
candidates and staff. Founded in 2018, Arena is a relative newcomer to the
world of training campaign staff. The Republican-funded Leadership Institute,
by contrast, has trained a generation of Republican campaign operatives since
its founding in 1979.
Academics have underinvested in the study of political talent, too. The
time between the 1979 launch of the Leadership Institute and the 2018 found-
ing of Arena mirrors the gap between two eras of academic research into
political staff, one in the 1970s [Kayden, 1973, Rosenbloom, 1973, Wilson,
1974, Fenno, 1978, Kayden, 1978, Stillman, 1978]; and another that gathered
steam in the 2010s, focusing in large part on legislative staff and campaign
consultants [Nyhan and Montgomery, 2015, Sheingate, 2016, Crosson et al.,
2018, McCrain, 2018, Medvic, 2018, Furnas, 2019, Hertel-Fernandez et al.,
2019, McCrain and Palmer, 2019, Burgat, 2020, Calcagno and Montgomery,
133
2020, Landgrave and Weller, 2020, LaPira et al., 2020].1
This dissertation contributes to that second wave of research by ex-
panding its aperture to include political campaign staff. In this conclusion I
summarize that research, including where it fits into others’ work and what
work remains to be done. I end with some observations and recommendations
for policymakers, candidates, personnel, party leaders and regulators.
6.1 Findings and Contributions
In Chapter 3 I argue campaigns are temporary organizations. More
specifically, campaigns are project networks: time bounded organizations oper-
ating on fixed length projects inside of a larger web of permanent organizations
[DeFillippi and Sydow, 2016]. This web includes party structures, political ac-
tion committees, consulting firms and candidate support organizations. Many
of the tensions scholars have found in project networks in other industries are
present in political campaigns, too. One such tension is between the goals of
the campaign and the permanent organizations around them, including the
national party and, to a lesser extent, the consulting firms that serve them.
The other tensions identified by project network scholars have largely
been resolved in this project network, though not always in a way that benefits
candidates. Regarding the tension between individual and collective identity,
1With respect to the few brave souls that published in the intervening period, including
[Binford, 1985, Herrnson, 1992, Rundquist et al., 1992, Romzek and Utter, 1997, Thurber
and Nelson, 2001, Rosenthal and Bell, 2003, Francia and Herrnson, 2007]
134
for example, it is generally understood personnel have their own careers and
firms to worry about. Each campaign is a part of those team members’ on-
going efforts to turn themselves into permanent fixtures of the political cam-
paign network. The tensions between creating and transferring knowledge
and crafting and standardizing practices are also mostly “resolved” in the
campaign network. Function-specific knowledge and practices are transferred
to function-specific consultants, while other knowledge and practices about
campaign management remain mostly tacit and eventually lost as individual
managers and consultants exit the system.
In political campaigns for the U.S. House, there are two project net-
works. At the heart of the Democratic project network are the Democratic
Congressional Campaign Committee and Emily’s List, an independent abor-
tion rights political action committee. Democratic campaigns are mostly run
by campaign managers, who oversee both staff and functional consultants.
The Republican project network, by contrast, revolves around consulting shops
that plan, organize, and execute Republican campaigns.
These two different systems shape the pathways through which politi-
cal campaigns hire differently for Republicans and Democrats, but they share
some commonalities. Candidates’ first hire, for example, matters in both net-
works, but it is more evident on the Republican side that the choice of a
general consultant serves as a signal of the seriousness and viability of a cam-
paign. Campaigns on both sides of the aisle hire almost exclusively through
their existing and extended networks, though Democratic campaigns have be-
135
gun to post jobs publicly. They do this in large part thanks to the absence of
clear signals about the quality of campaign personnel; in an environment with
such great uncertainty about the correlation between effort and outcome, it is
next to impossible to judge the performance of an individual firm or staffer.
Some campaigns go about this network-driven hiring carefully and thought-
fully, others less so. None of the campaigns interviewed here had an explicit,
repeatable hiring process, nor was clear guidance on hiring forthcoming from
party organizations or consulting firms.
One predictable result of this network-driven hiring is a tendency for
campaigns to hire people from similar groups and backgrounds, to the de
facto exclusion of other groups, including women and people of color. By
the late 2010s Democratic campaigns had begun to wrestle with this question
directly, to the point of making diversity and inclusion an explicit goal of some
campaigns.
The tension between these two goals of Democratic campaigns—to win
and to build a diverse bench of candidates and personnel—were just two areas
of goal tension within campaigns. Surprisingly, neither hiring managers nor job
seekers mentioned financial considerations in their list of motivations for hiring
and joining campaigns. Despite the fact that campaign salaries and payments
are publicly available on the Federal Election Commission’s website, salary
was not part of campaign personnel negotiations. Interviewees expressed their
self-interest in terms of competition, career, reputation, and power.
In Chapter 4, I answer quantitatively some of the questions raised in
136
Chapter 3, finding that political teams’ history of working together, called
team familiarity in management research, is positively associated with a can-
didate’s vote share and, for Republican candidates, the amount of money they
are able to raise. Having a more experienced team, however, helps very few
campaigns very little. It is not at all associated with greater fundraising and
only associated with greater vote share when looking specifically at the ex-
perience levels of consultants working on Republican incumbents’ campaigns.
Together, these results paint a picture of a mostly dysfunctional labor market,
in which one of the primary signals campaigns might use to hire—prior expe-
rience running a campaign—doesn’t lead to better performance in fundraising
or at the ballot box. On balance, candidates are better off hiring a team that
has worked together before, than building a team full of veterans without a
history of working together. More than half a century later, successful cam-
paigns still follow the Daley machine advice, “not hiring nobody nobody sent”
[Rakove, 1979].
Chapter 5 follows campaign staff through the election and into Congress,
where each year roughly 8% of legislative staff also worked in the prior elec-
tion as campaign staff. Conventional wisdom is that political hacks make bad
policy wonks, and that elected officials should not hire the former to do the
latter’s job. I find, however, that elected officials who hire more campaign staff
into their legislative office are no less effective as legislators than their peers,
and do no better or worse in their subsequent reelection campaigns. While
candidates may struggle to find competent staff in their campaigns, they are
137
on balance able to use the campaign as a months-long job interview to find
campaign staffers that will not negatively impact their legislative efforts.
6.1.1 Contributions to American Politics
The primary contribution of this dissertation is a straightforward one:
it introduces campaign staff and campaign organizations as worth studying in
their own right. It adds new knowledge about who campaign personnel are,
what motivates their decisions to join campaigns, what motivates candidates
and managers to hire them, the order in which they are hired, the different
organizational structures that they inhabit, and how they impact campaigns.
This knowledge contributes to scholarship on elections, Congress, and public
policy theory.
The majority of empirical work on U.S. elections revolves around ques-
tions of campaign finance, incumbency advantage, ideological alignment, and
electoral outcomes. A smaller corner of the elections literature focuses on cam-
paign effects. This literature is focused mostly on campaign communications
and events, and whether they have an effect on voter behavior. This work
turns the focus inward, to the individuals making the decisions that might
have those effects and the ones responsible for raising and spending the money
that campaign finance scholars follow.
This dissertation, and in particular Chapter 3, also reintroduces orga-
nizational theory to the study of political campaigns. As discussed in Chapter
2, political science scholarship took an institutional turn in the 1970s and
138
1980s, while organizational scholars in management departments carried some
of the same early literature on organizations into the study of their internal
processes and outcomes. As a result, there are some forty years of insight from
organizational theory that might be applied not only to the study of political
campaign teams, but to the study of all types of political organizations.2
The second body of political science work this dissertation contributes
to is the study of individual actors in and around Congress—a small but
growing field of work that has highlighted the role of lobbyists, political cam-
paign consultants, and more recently, legislative staff. This paper examines
the history of those legislative staffers that have a history of working on po-
litical campaigns. A subset of this work specifically examines the networks
that comprise those actors. This paper contributes to that body of work, too,
by demonstrating the way networks of shared work experience shape political
campaign hiring.
6.1.2 Contributions to Public Policy Literature
By contributing to this growing body of work on subsets of actors in
the political process, this dissertation also contributes to theory on the pub-
lic policy process. Many theories of the public policy process have explicitly
structural roots. The motivating imagery of information processing theory, for
example, is earthquakes and punctuated equilibria in evolution. This does not
2In the 1970s some scholars did work on what they called political organizations; but
they were speaking narrowly of voluntary political organizations, such as local, self-organized
partisan clubs [Wilson, 1974].
139
lend itself to asking questions about the motivations and behaviors of individ-
ual tectonic plates or gene pools. But the system that information processing
theory explains is full of actors making individual choices. Understanding what
drives those individual choices will give us insight into not just the structure,
but the agency of policy change.
Another contribution is to literature on the influence of the campaign
trail on legislative behavior. That work has focused historically on the po-
tential influence of campaign donations and only recently on how the issues
discussed in campaigns might carrying over into congressional behavior [Sulkin,
2011]. Chapter 5 of this dissertation is just the beginning of examining who,
in addition to the candidates themselves, might be the agents carrying lessons
from the campaign into congress.
6.1.3 Contributions to Management Theory
Lastly, this paper contributes to management theory on temporary or-
ganizations and project networks. As these scholars point out, an increasing
number of modern organizations are designed to be temporary. Within per-
manent firms, more work is organized temporarily, with internal task forces or
external consultants leading short-term, fixed-end projects. Yet we know little
about how these kinds of teams form, who they are made up of, or how their
formation affects their performance. This paper offers a view into multiple
teams in two project networks—one Democratic, one Republican.
By extending temporary organization literature into politics, this pa-
140
per also contributes a new kind of case to that literature. As with much
organizational scholarship, the focus of temporary organizations has been on
market-oriented firms and industries. When I approached this dissertation, I
first conceptualized political campaigns as an industry. More fitting, however,
might be a more sociological lens of political campaigns as a “field of action,”
a wider category of human behavior that can encompass fields, like this one,
where the activity is partly but not primarily economic.
6.2 Future Research Directions
The number of future research directions suggested by this dissertation
is an embarrassment of riches. I discuss many of them in the conclusions of
Chapter 3, Chapter 4 and Chapter 5; so I include here only a few of the most
promising. The first is continued work on campaign staff. One obvious area
to be explored with existing data is gender pay equity. Others include the
variation across campaigns in reliance on consultants versus full-time staff;
the degree to which personnel are located within or outside of a candidate’s
district or state; and the centrality of staffers in the broader campaign network.
This dissertation is also just one example of how much political science
might benefit from borrowing theory and methods from management scholar-
ship on organizations, strategy and entrepreneurship. Qualitative interviews
for this dissertation surfaced significant variation, for example, in how cam-
paigns conceptualized and conducted strategic decision-making. In addition
to political campaigns, there are a functionally limitless number of other or-
141
ganizations in politics that have yet to receive much attention, including the
DCCC, NRCC and other central party organizations; the candidate support
organizations that have become influential in Democratic campaign networks;
and the consulting firms that dominate Republican ones. Another piece of
qualitative work might explore mobility across political consulting firms, espe-
cially among Republican consultants, tracing the football coach-like lineages of
consultants mentioned by one interviewee and how they contribute to knowl-
edge spillovers across campaigns.
Finally, future work can bring questions of teams and organizations
back into study of the public policy process. Information processing theory, for
example, focuses on policy agendas. Only recently have scholars begun to ask
how and whether campaigns or legislative staff shape those agendas. Asking
about whether campaign staff might, too, could be one next step in studying
the crossover staff introduced in Chapter 5. Staff, just like their bosses, operate
within the constraints of bounded rationality and heuristic-driven decision-
making that are the understudied microfoundations of information processing
theory.
6.3 Practical Observations and Recommendations
The previous sections, written primarily for academics, cover what is
known and remains unknown about political campaign staff and organizations
inside of the academic literature. The rest of this conclusion is written not
for academics but for candidates, political professionals, policymakers, and
142
citizens with an interest in campaign teams and staff.
For candidates and political operatives. It will come as no sur-
prise that candidates and campaigns hire mostly through networks and mostly
based on who has worked with whom on previous campaigns. What may be
less obvious is why that happens, what costs it carries, and how they might
mitigate some of those downsides.
What I discovered in the course of this research were two key drivers
behind the way that campaigns hire. The first is the temporary nature of cam-
paigns; the second is the general uncertainty about which campaign activities
lead to what campaign outcomes. The impermanence of campaigns, combined
with their pace, turns hiring processes into quick, gut-driven affairs in which
managers choose from among people readily at hand. The uncertainty of cam-
paigns means that it is difficult to tell signal from noise in hiring; how is a
hiring manager to know whether a campaign was successful at raising funds
because of the finance director or because of the candidate? Together, these
factors lead campaigns to hire through the dense networks created by shared
work experience on prior campaigns.
This makes some strategic sense but carries a great cost. First, it means
campaigns can’t identify and learn from their hiring mistakes in a structured
way from campaign to campaign. They rely instead on untested rules of thumb,
informed not by testing but from candidates and consultants that attribute
their success to their own skill, rather than to chance. Second, it means cam-
paigns move too fast to avoid the mistakes network-driven hiring are known to
143
lead to. By hiring people who think similarly, they build teams with fewer di-
verse perspectives available in making difficult decisions. Third, hiring quickly
means campaigns seldom understand the motivations of their employees, and
lose the chance to place them in positions where they might be most effective.
As interviews for Chapter 3 revealed, Democratic campaigns now face
another set of issues created by this pattern of hiring. Because unstructured
hiring processes lead managers to hire who they know and who they are com-
fortable with, it means women and people of color are underrepresented among
campaign staff. And though not shown here, it’s likely the case that the infor-
mality of hiring in political campaigns leads to disparities in pay. A series of
recent sexual harassment scandals has highlighted what female staffers already
knew: the world of politics and campaigns is a minefield of potential abuse of
staff, who are only just beginning to organize themselves into unions in some
Democratic campaigns.
Another effect of the way political campaigns hire is that political op-
eratives, as shown in Chapter 4, do not stay political operatives for long. As
one former staffer shared with me, there is no development pipeline in political
campaigns. Only when this staffer left politics for work at a consulting firm
did they have a boss who asked them about their short-term and long-term
career goals.
For party leaders. There are way party leaders might address some
of these issues. The first is to build up more human capital resources for cam-
paigns. Democratic organizations already offer resume banks and reviewing to
144
campaigns they support. This could be expanded to include training on hiring
processes. Those resume banks can also be expanded into databases search-
able by both candidates and personnel, so candidates could better understand
the work histories of the people they are hiring. Staffers could use it to con-
duct background research on their potential employers, and avoid toxic work
environments. Such a database, perhaps alongside human capital consulting
services, could be operated by a party, a candidate support organization, or a
consulting firm operating either as a for-profit, a non-profit, or a worker-owned
cooperative.
As the Democratic support organization Arena has discovered, both
parties have underinvested in training campaign staff. This is beginning to
change. Still most of the training provided to campaign staff is technical
training focused on fundraising or communications or field operations. Pre-
cious little training exists on building or managing teams, or how campaign
staff might build a career or a consulting firm out of their experience.
For Democratic leaders. Part of the reason Democratic candidates
struggle to find competent staff, while being inundated with offers from con-
sultants as soon as they file is that the on-again, off-again lifestyle of political
campaigns leaves staffers with no professional development path and no stable
employment. Democrats can learn something from Republicans here, where
campaign staff are often absorbed in the off season into consulting firms that
also serve other clients, giving them some stability between campaigns. One
path forward for Democratic leaders might be to build more organizing roles
145
for experienced staffers between campaign cycles.
For Republican policymakers and political operatives. What
little history exists on the National Republican Campaign Committee sug-
gests that its influence began to wane in the 1990s, around the same time
Newt Gingrich moved to reform the Republican Party. In the gap (or rather,
the market) left by the decay of those institutions, for-profit Republican con-
sulting firms have become the repositories of knowledge and practices and de
facto gatekeepers of Republican primaries. One result of this privatization of
Republican campaigns was the absence of an institutional party structure that
anti-Trump Republicans could use to stop the rise of Donald Trump in 2015
and 2016. Today, Republicans are defined primarily as being either pro- or
anti-Trump. I argue the only way out of this situation is to renew some of the
party institutions that decayed after the 1980s, so as to become independent
pillars of power within the Republican Party. I believe these institutions can
be reconstructed so as to re-factionalize and channel popular voices within the
Republican Party, rather than to shut them out.
For policymakers and activists focused on government trans-
parency. Some good news: the Federal Election Commission has a tremen-
dous amount of publicly available information on political campaign staff and
consultants in expenditure reports campaigns are required to submit. But like
federal data on lobbyists, it is not always easy to work with. And there is, to
my knowledge, no data on legislative staff available to the public online. I had
to pay a private, for-profit company for this data about our public institutions.
146
There are other some quick transparency wins to be had here. Legisla-
tive staff data should be available to citizens, for example, not just to paying
lobbyists. Data on campaign staff, consultants, lobbyists, and legislative staff
should be connected, searchable and downloadable in multiple common data
formats. Campaign personnel, including individuals working at campaign con-
sulting firms, should have unique identifiers, as do lobbyists, so that we have a
more perfect idea of who is shaping our government. If this were the case, we
could more easily track a legislative staffer moving to managed an allegedly
unaffiliated Super PAC or taking time off from working in a legislative office
to “volunteer” on a a boss’s campaign.
For citizens and students of democracy, I offer a small note of
hope. Political campaigning, the path by which policymakers rise to power,
is a network without clear, fixed boundaries. Incumbents can be challenged,
unknown candidates can raise money and elections can be run by smart people
without a tremendous amount of political experience. The uncertainty and
porous borders of political campaigning offer multiple entry points for citizens
to shape government. The public square may be full of problems, as American
philospher John Dewey wrote, but it is full of possibility, too.
147
Appendices
148
Appendix A
Candidate and Personnel Interview Template
Below is the template I used in interviews for Chapter 3.
1. Meta
(a) Date:
(b) Respondent:
(c) Campaign:
(d) Title:
(e) Location:
2. Introduction (5 minutes)
(a) Thank you for being open to talk,–I know you’re busy. What work
am I interrupting today?
(b) So I have some official language that I need to get through, to make
sure I’m following all of the university ethical guidelines.
(c) We are researchers at the University of Texas working on a research
team that is exploring how political campaign teams form, how they
process information, and how they make decisions.
149
(d) The goal of our project is to understand the different organizational
forms that campaigns take: Are they more like startups, for exam-
ple, or movie crews, or disaster response teams?
(e) The goal of this interview is to get a sense of how your campaign
in particular operates (or operated). I will keep all identities confi-
dential, and any findings from the study will be reported in a way
that preserves the confidentiality of private information; each case
study I write will be anonymized, with identifying details scram-
bled or replaced. Additional information on the study and data
management was in the email I sent to you on [. . . ].
(f) I usually transcribe the interviews, which allows me to focus on
the interview and lets me listen to the interview again for details
I might have missed and not have to pause mid-interview to make
sure I get perfect notes. Do you have a preference for recording or
not? May I audio record our conversation using a digital recorder.
[BEGIN RECORDING]
3. Section One - Interviewee Background (10 minutes)
(a) For non-candidates
i. Before I’d like to learn a little bit more about you—can you
share a little bit about how you got involved in public service
and politics?
ii. And how did you get involved in this particular campaign?
150
iii. What was it about this campaign that was particularly attrac-
tive to you?
iv. What other work were you considering at the time? What
made you choose this?
v. Why do you think the campaign picked you? Who picked you?
(b) For candidates
i. I’d like to learn a little bit more about you—can you share
a little bit about how you got involved in public service and
politics?
ii. When did you begin thinking about and planning for this race?
iii. (If not incumbents) What other roles (other races, even other
jobs) were you considering at the time? What made you choose
this one?
iv. As you think back to the decision to run, who did you speak
with before you were certain about running?
4. Section Two - Hiring (5-30 minutes)
(a) I’d like to ask some similar questions about some of the people you
hired. Based on data from the FEC, it looks like you all are likely
to end up with north of a dozen paid or consultants on the team.
Of those, which are the ones you would consider having been key
hires?
(b) Which of these did you have a role in bringing on board?
151
(c) If you were to divide the campaign into different periods, so that I
could ask some questions about hiring in each period, how would
you do that?
(d) Period 1
i. Who came on board during this period?
ii. (Pick one)
iii. How did they come to the campaign? How did you find them/-
did they find you?
iv. What other candidates did you consider?
v. How did you choose?
vi. What do you think made them want to come work for you?
vii. Did it work?
(e) Period 2 (same)
(f) Period 3 (same)
(g) We’ve talked so far about people that you remember, and therefore
people that were probably important. Can you talk a bit about
how that might compare to the hiring process for more junior staff
(Point to list and pick out a staffer or two)?
5. Section Three - Planning (10 minutes)
(a) The result of all that hiring (both staff and consultants) is a reason-
ably complex organization, operating in an uncertain environment,
152
some of which you were able to plan for and some of which you
could respond to. I have a few questions about how you did both.
(b) At what points in the campaign did you engage in active planning
or strategizing, and what did that look like?
(c) Period 1
(d) Period 2
(e) (For each of those times) Who was in the room? Whose input and
views were they responsible for representing?
(f) What frameworks or “mental models” did you use?
6. Section Four - Information Processing (15 minutes)
(a) Some things can’t be planned for; they have to be reacted to. What
things over the course of the campaign happened that you would
consider a surprise?
(b) Who in the campaign learned about this first? Who did they tell
next, and so on?
(c) What happened next? Was there one meeting, or many?
(d) Who was a part of deciding what the campaign’s response would
be?
7. Section Five - Wrap Up (10 minutes)
153
(a) Excellent, this is great stuff. That’s about it for today. As we were
talking, were there things that you felt like you didn’t get a chance
to add? Or questions I should have asked?
(b) I mentioned that my hope is to build anonymized case studies of
these campaigns, interviewing 3-4 people. Would you be willing to
forward along a note to other people in the campaign? Ideally that
would be the Finance Director, Campaign Manager, key consultant,
candidate.
(c) What would it take for you to be comfortable with that?
(d) Thank you for the time—it’s appreciated and important and will
help me get a better grip on how campaigns operate. My hope is
that at the end of all of this, I will also have learned a few things
that will be useful to you. I’m happy to share what I have learned
so far, too, if you have other questions.
154
Appendix B
Building a Database of Campaign Personnel
By law, all campaigns for federal office must submit expenditure reports
to the Federal Election Commission. The FEC has those records collected and
digitized from 2003-present—a total of some eleven million expenditures. Each
of those expenditures contains multiple features, including the NAME of the
recipient, the PURPOSE of the expenditure, and the entity type ENTITY TP
of the recipient. I use the code below to construct a campaign personnel
database from those expenditure reports. I operate in the R environment, and
use the following packages in this dissertation:
l ib ra ry ( t id yv er se ) # F or e ve r yt h in g d at a - wr a ng l in g
library(janitor) # fo