Why Are There So Many Lawyers in Congress?

Abstract

Scholars have long sought to explain the overrepresentation of lawyers in Congress. This article draws on a wealth of data to explore the causes and consequences of this representational imbalance. While lawyers enter politics at higher rates, self‐selection at best provides a partial explanation. Conditional on running, lawyers win at twice the rate of candidates from other backgrounds. Contrary to prevailing theories in the literature, voters do not reward candidates with backgrounds in law. Rather, lawyers win because of a sizable competitive advantage in early fundraising, owing in large part to their professional networks. This study has important implications for who runs for office, who wins, and the demographic composition of Congress. It also identifies an underexplored mechanism by which the US system of campaign finance sustains deep representational imbalances.

Full text

Why Are There So Many Lawyers in Congress?
Adam Bonica*
October 1, 2019
Abstract. Scholars have long sought to explain the overrepresentation of lawyers in Congress. This paper
draws on a wealth of data to explore the causes and consequences of this representational imbalance. While
lawyers enter politics at higher rates, self-selection at best provides a partial explanation. Conditional on
running, lawyers win at twice the rate of candidates from other backgrounds. Contrary to prevailing theories
in the literature, voters do not reward candidates with backgrounds in law. Rather, lawyers win because of a
sizable competitive advantage in early fundraising, owing in large part to their professional networks. This
study has important implications for who runs for office, who wins, and the demographic composition of
Congress. It also identifies an under-explored mechanism by which the U.S. system of campaign finance
sustains deep representational imbalances.
*Stanford University, 307 Encina Hall West, Stanford CA 94305,bonica@stanford.edu.

In reflecting on the role of lawyers in the early American Republic, Alexis De Tocqueville
famously referred to the legal profession, comprised of the bench and the bar, as the “American
Aristocracy” (Tocqueville, 1840). Nearly two centuries later, lawyers continue to dominate Amer-
ican politics. In addition to staffing an entire branch of government, lawyers are well represented
in elected office. While comprising a mere 0.4 percent of the voting-age population, lawyers ac-
counted for 39 percent of seats in the House and 56 percent of seats in the Senate in the 115th
Congress. The overrepresentation of lawyers vastly exceeds even that of millionaires. Relative to
the average citizen, millionaires are approximately ten times more likely to be elected to Congress.1
Lawyers, by comparison, are nearly 100 times more likely to be elected to Congress.
The implications of concentrating political power in the hands of a single profession were not
lost on Tocqueville. Lawyers as a group have shared interests, incentives, and concerns, which, in
turn, shape their political outlook and understanding of policy. Along these lines, contemporary
legal scholars have argued that the legal profession’s unique relationship with politics has benefited
lawyers both economically and politically (Hadfield, 2000, 2008; Posner, 1993). Such claims are
consistent with evidence that vocational background can influence the attitudes and choices of
legislators (Carnes, 2012, 2013; Matter and Stutzer, 2015). Lawyers have also influenced the
organization and inner workings of Congress. Miller (1995) documents the various ways lawyers
have shaped the rules, procedures, and cultural norms of Congress, often to their advantage.
The overrepresentation of lawyers speaks to what is perhaps the fundamental question in the
study of democratic representation: Why are some segments of society so much better represented
than others? Accounting for the electoral success of lawyers offers insights into what it takes to
run for office successfully, the barriers to entry and how they are overcome, and what makes some
individuals better positioned than others to navigate the electoral process.
There is no shortage of theories on why so many lawyers are elected to Congress. In his book
The High Priests of American Politics, Mark Miller (1995) compiles an impressive compendium of
claims and hypotheses put forth over the years by 94 different scholars. Most of these explanations
1Millionaires occupy a slightly greater share of seats during this period (48 percent) but also draw from a much
larger share of the population (4.5 percent) (Cody, 2014).
1

focus either on (1) how a heightened interest in politics combined with career incentives specific to
the legal profession make lawyers more likely to pursue careers in politics or (2) how certain traits
or skills associated with lawyers might appeal to voters or otherwise make for capable candidates.
However, lacking data to test these claims, scholars have been left to speculate which, if any, hold
weight.
The study leverages a new dataset on the educational and professional backgrounds of thou-
sands of congressional candidates to empirically test several key claims and hypotheses advanced
in the literature. In covering the candidate population more broadly, the dataset enables a sys-
tematic accounting of rates of entry into the candidate pool and electoral success by profession
and educational background. In line with past accounts, I find that even when compared to simi-
larly high-status professions, lawyers are more likely to run for political office. While important,
lawyers running at higher rates is only part of the story. Conditional on running for office, lawyers
enjoy much higher rates of electoral success than candidates from other backgrounds—but not for
the reasons typically offered by scholars. Lawyers are neither held in high-esteem by voters nor do
their skills offer a distinct advantage when campaigning. Rather, their competitive advantage lies
in their strength as early fundraisers.
The early fundraising advantage is sizable. Lawyers running as nonincumbents fundraise at
twice the rate as candidates from other backgrounds during the initial months of their campaigns,
generating crucial resources and momentum heading into the primaries. Their success as early
fundraisers owes in large part to money raised from other lawyers. Combined with evidence that
primary elections are especially sensitive to early fundraising outcomes, this suggests that profes-
sional networks are instrumental to winning elections. It also identifies a key mechanism by which
the U.S. system of campaign finance sustains deep representational imbalances.
2 Lawyer-legislators in Comparative Perspective
Congress has historically been a legislative body dominated by lawyer-legislators.2As a result,
the prevalence of lawyers in Congress is often assumed to be a natural consequence of the special
2Calculations based on prior occupations in the Biographical Directory of the U.S. Congress indicate that during
the 1-115th Congresses lawyer-legislators have averaged 62 percent of seats in the House and 71 percent in the Senate.
2

Figure 1: Legal Professionals as Proportion of National Legislatures and Lawyers Per Capita for
OECD Member States
AUT
BEL
CZE
DNK
FRA
DEU
GRC
HUN
ITA
JPN
NLD
POL
SVK
SVN
ESP
SWE
CHE
TUR
AUS
CAN ISR
NZL
GBR
USA
0
10
20
30
40
50
1 2 3
Number of Lawyers per 1,000 Citizens
Percentage of National Legislators with Background in Law
Nations with commonwealth legal systems are in bold.
Source: Data on lawyers in the U.S. Congress are from the author’s calculations. The seat shares
of lawyer-legislators for other countries are calculated from data on professional backgrounds of
members published by the Inter-Parliamentary Union Chronicle of Parliamentary Elections. Cross-
national estimates of lawyer populations are from Michelson (2013). These are divided by popu-
lation estimations from the World Bank to calculate lawyers per capita.
relationship between law and politics (e.g., Eulau and Sprague, 1964). This, of course, implies
that lawyers should be similarly well-represented in legislatures around the world. Using data on
the professional and educational backgrounds of members of national legislatures collected from
the Inter-Parliamentary Union Chronicle of Parliamentary Elections, I calculated the proportion of
members with backgrounds in law for 25 OECD member nations.
As Figure 1 shows, there is substantial cross-national variation in lawyers as a share of the
population and national legislatures. There is a clear relationship between the numbers of lawyers
per capita and seat shares held by lawyers in national legislatures.3Even so, the U.S. is an outlier,
3There is no reason to assume that the number of lawyers per capita is exogenous to the prevalence of lawyer-
legislators. Several scholars have linked the political overrepresentation of lawyers to government-induced demand
3

with over twice as many lawyer-legislators as predicted by the fitted line. By comparison, lawyer-
legislators account for just 13 percent of the U.K. Parliament. The percentages are similar for
other nations that inherited the English Commonwealth system of law. Canada, New Zealand, and
Australia are at 15, 14, and 13 percent, respectively. The percentages for France, the Netherlands,
Sweden, Denmark, and Japan are much lower, ranging from 2 to 6 percent.
Figure 1 shows that a lawyer-dominated legislature is not an inherent feature of representa-
tive democracies. Neither is it innocuous. Section 11 documents ways lawyers have influenced
important political outcomes in their favor.
3 Existing Accounts
Existing scholarly accounts for the prevalence of lawyer-legislators generally fall into one of two
camps, focusing either on (1) how incentive structures specific to the legal profession increases the
supply of lawyer-candidates or (2) how certain traits associated with lawyers or the legal profession
appeal to voters, thereby increasing demand for lawyer-candidates at the polls.4
3.1 Supply-Side Explanations
One explanation for why lawyers enter politics at higher rates is that career incentives in the legal
profession tend to align with holding political office (e.g., Miller, 1995). More so than in other
professions, public service is an opportunity for career advancement. Evidence for this comes
from the widespread practice among law firms of rewarding associates with a sizable clerkship
bonus for spending an extended period away from the firm to clerk for a judge. Likewise, lawyers
are well-positioned to translate political experience into higher-paying jobs upon leaving office
(Diermeier, Keane, and Merlo, 2005; Polsby, 1990; Friedman, 1985).
At the same time, the costs of running for office may be significantly reduced for lawyers. Cam-
paigning for political office is a full-time job. The legal profession may be more accommodating
than others regarding the leave of absence required to conduct a serious campaign. Others have
gone so far as to argue that lawyers can often treat campaigning as a means of furthering their pro-
for legal services (Crandall, Maheshri, and Winston, 2011; Hadfield, 2008; Posner, 1993).
4See Miller (1995) for an excellent treatment of the explanations offered by scholars to explain the prevalence of
lawyers in U.S. politics.
4

fessional goals (Fowler and McClure, 1990). In particular, it provides an opportunity to advertise
and gain name recognition, as well as opportunities for networking, personal brand building, and
gaining clients. This type of exposure could go a long way in offsetting the costs of unsuccessful
campaigns, thus making it rational for lawyers to run for office even when the odds of success are
low.
Political ambition might also be associated with pursuing a career in law, which is widely seen
as a stepping-stone to a career in politics. This sentiment is perhaps best captured by a quote
from Woodrow Wilson, who wrote that: “the profession I choose was politics; the profession I
entered was the law. I entered one because I thought it would lead to the other.” (Miller 1995, 57).
This pathway to elected office was historically significant during the early republic. Practicing
law had traditionally served as an egalitarian gateway into politics. “Self-taught” lawyers could
practice law as a way to gain their footing before embarking on a career in politics. However, the
establishment of bar associations during the early 20th Century brought organized efforts to restrict
entry into the profession through the introduction of formalized legal education and bar exams. As
entering the legal profession became more costly, both in terms of time and money, it became less
viable an option for those looking to gain experience before running for office. At the same time,
costs incurred during law school have made it harder to leave the profession. Along these lines,
Robinson (2015) suggests the pressures of a contemporary legal career—including increasing time
demands and reduced flexibility—might discourage lawyers from entering politics.
The notion that politically ambitious individuals self-select into the legal profession is con-
sistent with evidence that lawyers run for office at higher rates than other groups. Although this
pathway likely serves to increase the number of lawyers who pursue elected office, it is an entirely
different question as to whether a background in law—or having exhibited political ambition from
a young age—improves a candidate’s chances of winning an election. The evidence presented be-
low suggests that while the legal profession produces a disproportionate share of candidates, being
a trained lawyer does not, in itself, make a candidate more desirable in the eyes of voters. This
suggests that while latent political ambition likely increases the supply of lawyer-candidates, it
fails to explain why lawyers win at higher rates conditional on running.
Another potential driver is selective recruitment by party leaders. Before the widespread adop-
5

tion of direct primaries during the early 20th Century, the recruitment and nomination of candidates
were largely party affairs. Candidates for Congress were chosen by party leaders—among whom,
lawyers were likely heavily overrepresented—in private meetings and officially nominated at state
party caucuses. The system of “old boy networks” that had dominated many state parties offered
an ideal environment for lawyers to consolidate seat shares to the exclusion of other groups. Con-
temporary party gate-keepers have less control over who runs in the primaries but are key players
in the candidate recruitment process. It is unclear, however, whether recruiters actually prefer
lawyers over other types of candidates. Drawing on a survey of party gate-keepers that asked re-
spondents to rate candidate traits on a scale of (0) “not important,” (1) “somewhat important,” or
(2) “very important,” (Broockman et al., 2014) find that most party leaders deem being a lawyer
as unimportant with an average rating of 0.45. By comparison, party leaders indicated that they
viewed being “able to raise money from friends and associates” more important with an average
rating of 1.33.
3.2 Demand-Side Explanations
Demand-side accounts have focused on identifying personal characteristics associated with lawyers
that voters find attractive or otherwise make for talented politicians. One of the earliest explana-
tions of this sort is known as the high-status argument (Tocqueville, 1840). It holds that as members
of a well-educated, high-status occupation distinct from the traditional aristocracy, lawyers came
to be viewed favorably in the eyes of voters.
Another account known as the American legal culture argument holds the legal profession
played a formative role in shaping the nation’s political culture in its own image, and as a result,
made lawyers uniquely qualified to undertake the business of politics (Scheingold, 1974; Halliday,
1979). In a similar vein, other have argued that lawyers possess “special skills” that lend them-
selves to a career in politics and give them an advantage over other types of professionals (Hain
and Piereson, 1975; Podmore, 1980). Proponents of this idea argue that through legal training,
lawyers acquire vital skills, including the ability to speak, write, argue, and advocate. Although
posited as such, it is unclear why any of these skills would be unique to lawyers.
A notable weakness of demand-side explanations is the unsupported claim that lawyers, re-
6

gardless of the reason, appeal to voters. Evidence from opinion polls casts doubt on the notion
that electorate holds lawyers in high esteem. Public perceptions of lawyers are decidedly negative.
Overwhelmingly majorities view lawyers as dishonest, unethical, and contributing little or nothing
to society (Pew Research Center, 2013; Gallup, 2015). More direct evidence comes from survey
experiments that rely on sophisticated methods to estimate the causal effects of candidate attributes
on voter assessments. Hainmueller, Hopkins, and Yamamoto (2014) use a conjoint analysis design
to investigate how the personal characteristics of candidates affect levels of support. They find
significant effects related to age, religion, and military service but no discernible effect associated
with a career in law. Fong and Grimmer (2016) employ similarly sophisticated methods to esti-
mate treatment effects of candidate traits from a survey experiment that asked respondents to rate
candidates after being shown biographical information. They find that traits associated with legal
experience are penalized by respondents. The results presented in Section 7 are consistent with
these findings.
4 A Framework For Modeling Candidate Entry and Electoral Success
Supply-side and demand-side explanations can both be expressed using a more general model of
candidate entry. The Rational Model of Candidate Entry, which posits that candidates will enter a
race when the expected returns from winning office outweigh the costs of campaigning, provides
the theoretical underpinning for much of the academic literature on candidate entry (Black, 1972;
Rohde, 1979; Jacobson and Kernell, 1983). The most basic formulation weighs the costs and
expected payoffs of running for office,
E[U]=(P⇤B)C(1)
where Pis the probability of winning, Bis the personal benefits of holding elected office, and C
is the personal and financial costs of running. The model generates straightforward comparative
statics. Reduced costs, increased benefits, and improved chances of success are all positively
associated with increased candidate entry.
Supply-side explanations offer a set of arguments for why the career incentives of lawyers
enhance the benefits or reduce the cost of entry; whereas, demand-side explanations offer a set
7

of arguments for why lawyers possess certain competitive advantages that increase the probability
of success conditional on running. Testing their respective predictions has been constrained by
a lack of data on who runs for office. Absent systematic data on the backgrounds of candidates,
one cannot determine whether lawyers enter politics at higher rates than other groups. Likewise,
data on the professional backgrounds of candidates is needed to determine whether lawyers win at
higher rates than other groups.
Here I make use of a newly constructed dataset on the characteristics of congressional candi-
dates from the 2010-2014 election cycles. It covers all 4,966 major party candidates who reached
the $5,000 fundraising threshold set by the Federal Election Commission (FEC) for mandatory
reporting. For each candidate, a team of research assistants collected data on educational back-
ground, including degree earned, degree-granting institution, and year of conferral. These data
were then merged with the DIME data, which includes detailed individual-level data on fundrais-
ing and election outcomes (Bonica, 2016).
Consistent with the literature, status as a lawyer is understood here in terms of membership in
the legal profession. A functional definition comes from James Wilson who defines a professional
as “someone who receives important occupational rewards from a group whose membership is
limited to people who have undergone specialized formal education and have accepted a group de-
fined code of proper conduct.” (Wilson, 1989, pp. 60).5Along these lines, I code profession based
on degree attainment.6,7 One advantage of this approach is that degree attainment represents a set
of unambiguous categorical outcomes.8Moreover, professional training can matter independent
of experiences gained during one’s career. Specialized training of any type is designed to instill
specific tools and approaches to problem-solving. The purpose of law school is to train someone
5Wilson develops this definition in the context of the bureaucracy. Professional background, he argues, is impor-
tant because it exerts an external influence on bureaucrats separate from the organizational incentives of their agencies
and can lead bureaucrats to define their task to reflect their training and the norms and standards of their profession.
These arguments are no less applicable to legislators.
6Degree status is coded based on highest degree attained. The BA category represents the set of candidates for
whom their bachelor’s degree was their terminal degree.
7Carnes (2013) codes the proportion of pre-congressional careers spent in nine occupational categories. This
approach works well given the primary distinction of interest is between legislators with working-class and white-
collar backgrounds. Here the main distinction is between lawyers and non-lawyers.
8This avoids cases where professionals straddle employment categories—for example, a physician in solo practice
is both a professional and a small business owner.
8

to “think like a lawyer,” which carries over to a career in politics (Barton, 2010; Miller, 1995).
To address potential limitations of coding professional backgrounds based on degree attain-
ment, I augment the data set in several ways. First, not everyone who earns a professional degree
pursues a career in the field. However, in practice, nearly all do. I coded professional background
from data on employment history from Project Vote Smart, which covers 83 percent of candidates
included in the sample. Of those with law degrees, only about 1 in 20 had not practiced law. Sec-
ond, law graduates might be more likely to spend time in politics before running for office. I adjust
for this by constructing two variables capture prior political experience. The first controls for prior
experience working as a campaign or legislative staffer. The second controls for having previously
held elected office. Third, as a robustness check, I re-estimated the models with a separate category
included for candidates with law degrees but never practiced law. Lastly, I replicated the results
using an alternative coding scheme based on employment history similar to that of Carnes (2013)
for the subset of candidates with corresponding entries in Project Vote Smart. In both cases, the
results are qualitatively identical.9
5 Who Runs For Office? Who Wins?
The Rational Model of Candidate Entry posits that candidate entry is conditioned on a simple cost-
benefit analysis adjusting for risk. As the costs of running for office fall, so does the threshold for
beliefs about the likelihood of success needed to justify entry. If the costs are sufficiently low and
the benefits sufficiently large, entry might be rational even if the candidate is unlikely to win. At
the other extreme, if running for office would incur significant costs on one’s personal finances or
career, a candidate must be more confident they will succeed for entry to be rational. As such, if
membership in the legal profession makes running for office less costly and holding office more
rewarding, as supply-side accounts suggest, the model predicts that lawyers should be more likely
to run for office but less likely to win when they do.
9A possible concern is that the coding scheme might fail to capture some meaningful variation within professions.
On this point, I note that the same would be true of other coding schemes. Moreover, the coefficients on profession
reported in Tables 7 and 8—and main results reported below—suggests that broad professional categories capture
much of the variation of interest. Although exploring within-profession variation in greater detail is feasible, doing so
remains beyond the scope of this study.
9

Figure 2: Proportion of Degree types of Candidates and Member of Congress Relative to Propor-
tion of Voting-Age Population (2010-2014)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Law Degree
Medical Degree
MBA
PhD
Masters
Bachelors
Some/No College
1/20 1/10 1 10 20 30 40 50 60 70 80 90 100
Relative Odds of Being Candidate/Congressmember
Degree Type
●
●
Candidate
Congressmember
[Source] Author’s calculations. Data on degree attainment are from U.S. Census Bureau, Cur-
rent Population Survey, 2014 Annual Social and Economic Supplement. Categories are based on
highest degree attained.
Figure 2 compares rates of representation in the candidate pool and Congress by degree-type.
To construct the figure, I calculate the shares of candidates and officeholders from each group and
divide them by the corresponding share of the voting-age population (VAP). I then divide these
shares by the respective shares of the voting-age population (VAP) with each type of degree. A
value of one on the x-axis indicates that a group makes up the same share of Congress as they do
the general population. A value of ten indicates that a group of degree-holders are ten times more
likely to serve in Congress than the average citizen.
As the figure shows, lawyers are overrepresented in the candidate population and even more
so among members of Congress. Lawyers are 54 times more likely to be a candidate and 99 times
more likely to serve in Congress. No other group exhibits a similar pattern of representation.
Physicians are the second most overrepresented professional group but, unlike lawyers, account
10

Table 1: Win Rates by Degree type and Incumbency Status
Degree Type Nonincumbents Open Seat Challengers Primary Challengers Incumbents
Law Degree 0.132 0.193 0.108 0.085 0.893
No Law Degree 0.064 0.122 0.052 0.030 0.896
Medical Degree 0.098 0.105 0.096 0.094 0.953
MBA 0.069 0.096 0.060 0.062 0.907
PhD 0.075 0.132 0.055 0.043 0.904
MA 0.084 0.135 0.072 0.053 0.883
BA/BS 0.090 0.155 0.073 0.045 0.896
Some/No College 0.027 0.053 0.022 0.016 0.891
Win rates are defined as the proportion of candidates elected to Congress.
Source: Author’s calculations.
for similar shares of candidates and congressmembers.
This pattern persists when narrowing the sample to nonincumbents. Table 1 breaks down suc-
cess rates by incumbency status and degree type. Lawyers running as nonincumbents win at over
twice the rate of non-lawyers and are nearly three times as likely to win when mounting primary
challenges against same-party incumbents.10 This pattern does not carry over into incumbency,
with lawyers retaining their seats at a slightly lower rate than non-lawyers.
6 Why Elect Lawyers?
I consider two hypotheses consistent with the observation that lawyers enter and win at higher
rates. Each proposes a distinct mechanism that could act to increase the probability of success
conditional on running. Each asserts that lawyers benefit from a distinct type of shared competitive
advantage.
The first hypothesis draws from demand-side accounts that argue that lawyers benefit from the
“high-status” brand of the legal profession and the specialized skills honed by legal training.
H1: Electoral Advantage Hypothesis: Competitive advantages linked to the legal profession
cause lawyers to outperform at the polls, leading to increased rates of electoral success.
10To further complicate matters, the raw win rates are likely to understate the success of lawyer-candidates. Owing
to their high rates of entry, lawyers are far more likely to compete directly against other lawyers. For example, in 2014
the California 33rd Congressional District drew eight candidates, six of whom were lawyers. Since only one candidate
can win, the five others will be unsuccessful. Clustering of this sort can deflate the win rates for lawyers.
11

If their status and specialized skills make lawyers uniquely appealing to voters or otherwise
help them compete for votes, we should observe a positive effect on vote shares after controlling
for candidate and contest characteristics.
The second hypothesis considers the role of professional networks in providing financial re-
sources during the early stages of candidacy.
H2: Early Fundraising Advantage Hypothesis: Lawyers have a competitive advantage in early
fundraising. The resulting resource advantage, in turn, increases win rates.
On a practical level, early fundraising plays a fundamental role during the early stages of
candidacy. Straight out of the gates, candidates are expected to raise large sums of money to
get their campaign up and running. Candidates, like anyone else, are subject to budget constraints.
Important decisions regarding hiring, outreach, and how best to allocate time and effort often
depend on a candidate’s ability to fundraise early on. Financial constraints can limit the types of
strategies and talent available to a campaign. They can also be a matter of survival. Bankrupt
campaigns are rarely viable, and a lack of campaign funds is a common reason campaigns falter.
A fundraising advantage is one of the few plausible explanations consistent with lawyers both
running and winning at higher rates. I test this hypothesis using data on itemized contributions
raised during the initial months of candidacy.
7 Are Lawyers Favored at the Polls?
Demand-side explanations offer a set of claims about why lawyers excel as candidates. Despite
evidence that voters do not view a legal background, per se, as a selling point, past scholarship has
claimed that lawyers possess personal traits, such as charisma, that might be difficult to measure
but make for compelling political candidates or capable campaigners. Regardless of the reason,
demand-side explanations all predict that lawyers should outperform candidates from other back-
grounds. I test this prediction using data on vote shares in the primary and general elections.
12

Table 2: Determinants of Vote Shares in General Election Contests (House, 2010-2014): OLS
(1) (2)
Constant 44.52 41.55
(0.59)(0.55)
Law Degree 0.42
(0.44)
Medical Degree 0.95
(1.03)
MBA 0.40
(0.74)
PhD 0.31
(0.89)
Incumbent 8.74 14.03
(0.87)(0.75)
Open Seat 1.00 3.71
(0.94)(0.96)
Candidate Midpoint (CFscore) 1.47
(0.31)
Held Elected Office 0.58
(0.45)
Political Staffer 1.19
(0.60)
ln(Dem. Spending) - ln(Rep. Spending) 1.17
(0.16)
District Pres. Vote Share (Dem) 68.89 77.31
(2.70)(2.54)
2010 5.66 6.32
(0.64)(0.67)
2014 1.97 2.55
(0.64)(0.67)
R20.79 0.77
N. Obs 1,050 1,050
Dependent Variable: Democratic candidate’s share of the two-party vote.
Degree attainment and Held Elected Office and Political Experience are operationalized as signed dummy variables.
Candidate ideology is operationalized as the midpoint between the candidates’ CFscores (Bonica, 2014).
Source: Author’s calculations.
I begin by examining candidate performance in general elections. Table 2 models two-party
vote shares in general elections as a function of degree-type, incumbency status, district parti-
sanship, candidate ideology, campaign spending, whether candidates have previously held elected
office, and prior political experience as a legislative or campaign staffer. Degree-type enters as
signed indicator variables that take on a value of +1 if the Democratic candidate is a degree-
holder, a value of 1 if the Republican candidate is a degree-holder, and a value of 0 if neither or
both candidates are degree-holders.
13

The results provide no evidence that lawyers are advantaged at the polls. The estimated coef-
ficient for Law Degree is small and statistically insignificant, indicating that lawyers perform no
better than candidates from other backgrounds.
General election contests are primarily determined by factors that are beyond a candidate’s
control, such as incumbency status, district partisanship, and national partisan moods, as shown in
Model 2. Primary elections, by comparison, depend more on the individual talents and character-
istics of the candidates.
Modeling primary election outcomes introduces additional complexities. Unlike general elec-
tions, where two candidates compete for vote share, the numbers of candidates competing in pri-
mary contests can vary. When favorable electoral conditions in a district all but assures the party’s
nominee will be victorious in November, primaries often become very crowded. To adjust for this,
I normalize vote shares and fundraising totals relative to contest-level averages. For candidate iin
primary contest j, let vij be the candidate’s total number of primary votes, fij be the candidate’s
fundraising total during the primaries, and njbe the total number of candidates competing in the
primary contest. Adjusted vote shares are calculated such that
\
Vote Share =vij
(Âvj/nj)and adjusted
fundraising shares are calculated such that
\
Fundraising Share =fij
(Âfj/nj). This specification en-
ables House and Senate races to be pooled.
I narrow the sample to nonincumbents in contested primaries where two or more candidates had
filed with the FEC and met the minimum requirements to be designated an active candidate. I also
exclude blanket (or “Top Two”) primaries in Louisiana, Washington, and California. An indicator
variable is included for primary challengers. This leaves a total of 2,596 candidates competing
across 966 primary contests.
One potential concern is that the observed characteristics of lawyers and non-lawyers might dif-
fer in meaningful ways. To adjust for potential imbalances, I use matching to pre-process the data.
Covariate balance is achieved using the genetic matching algorithm from the Matching package in
R (Sekhon, 2011). (See supplemental appendix for balance statistics.)
The results reported in Table 3 are inconsistent with demand-side accounts. Model 1, which
regresses normalized vote shares directly on Law Degree, shows that lawyers, on average, receive
a larger share of votes in primary elections. However, the sign on the coefficient reverses when
14

Table 3: Normalized Vote Shares in Competitive Primary Elections: OLS
Pre-Matching Post-Matching
(1) (2) (3) (4)
Constant 0.859 0.483 1.048 0.437
(0.019)(0.046)(0.042)(0.083)
Law Degree 0.125 0.062 0.065 0.132
(0.039)(0.031)(0.057)(0.043)
\
Fundraising Share 0.512 0.538
(0.014)(0.020)
Primary Challenger 0.091 0.070
(0.040)(0.069)
Competitive Seat 0.048 0.039
(0.037)(0.069)
Safe Seat 0.077 0.057
(0.046)(0.082)
Political Staffer 0.006 0.018
(0.056)(0.079)
Female 0.069 0.091
(0.035)(0.059)
Prev. Held Office 0.156 0.148
(0.028)(0.045)
Republican 0.001 0.018
(0.029)(0.046)
Open Seat 0.044 0.057
(0.030)(0.048)
Senate 0.036 0.001
(0.035)(0.053)
2012 0.008 0.050
(0.031)(0.052)
2014 0.010 0.002
(0.034)(0.057)
R20.004 0.400 0.001 0.442
Num. obs. 2,596 2,596 1,086 1,086
Dependent Variable: Normalized vote shares in primary elections.
[Source] Author’s calculations.
including the full set of controls in Model 2. The post-matching results, reported in Models 3 and
4, similarly indicate the effect of Law Degree is negative, suggesting that, if anything, lawyers
underperform in primary elections after adjusting for relevant covariates.11
11The results are robust to excluding primary challengers. They are also robust to limiting the sample to candidates
running in winnable districts where their party’s presidential nominee won at least 40 percent of the two-party vote.
This helps to rule out that lawyers win more often because they are better able to tell when conditions in a district are
favorable. The results are also robust to controlling for candidate ideology. Lastly, results are similar for alternative
specification with a binary dependent variable corresponding to whether a candidate won their primary contest. See
supplemental appendix for results from these robustness checks.
15

8 Do Lawyers Have an Early Fundraising Advantage?
This section examines how professional background relates to early fundraising. I focus on fundrais-
ing by nonincumbents during the first 90 days of candidacy. Itemized contribution records include
transaction dates, which can be used to track early fundraising during the initial months of a can-
didate’s campaign, both in terms of amounts raised and the sources of funding. The start dates
for campaigns are assigned based on the dates reported in the FEC statement of candidacy filings.
Non-itemized contributions can be tracked using quarterly FEC filings where total amounts raised
from unitemized donors appear as a line item.12
Figure 3: Fundraising from Individual Donors During First 90 Days in House Contests
(logarithmic-scaling)
Nonincumbents
Incumbents
1K 10K 100K 500K 1K 10K 100K 500K
0
25
50
75
100
0
50
100
150
Log($ raised during first 90 days)
count
JD
Other
Observations are at the candidate-cycle level. Sample includes 2010-2014 election cycles.
Source: Author’s calculations.
Figure 3 compares early fundraising for House candidates grouped by incumbency status.
Among nonincumbents, there is a noticeable difference between lawyers and non-lawyers. Lawyers
raised an average of $105,861, more than double the $52,360 raised on average by non-lawyers.
A similar pattern is observed for senate candidates, with lawyers raising an average of $363,291,
compared with an average of $186,937 for non-lawyers. These differences are sizable. Even when
12Since candidates enter the race in a staggered fashion, the period covered by the first reporting deadline usually
does not match up with a candidate’s first 90 days in the race. This is adjusted for by interpolating the total unitemized
amounts reported by a candidate’s campaign in its first two quarterly filings.
16

campaigning as nonincumbents, lawyers fundraise on par with incumbents. The early fundraising
advantage for lawyers appears only to apply to non-incumbents. Among incumbents, the fundrais-
ing distributions for lawyers and non-lawyers are statistically indistinguishable.13
I model early fundraising outcomes in Table 4. Profession enters as a set of indicator vari-
ables.14 As before, I control for prior political experience. I additionally control for entry delays,
measured as the number of days since the start of the cycle that a candidate officially filed with the
FEC. At the contest-level, I control for seat status, district partisanship, the number of candidates
competing in the primary, and median household income.15 A second specification, reported in
columns 2 and 4, instead includes fixed effects for primary contests grouped by party, cycle, and
district.16
The sample covers nonincumbents running for the House and Senate during the 2010-2014
election cycles.17 The unit of observation is a candidate-cycle pair. The basic model specification
is Tobit. Censoring is rare in Table 4 but is much more common in Table 5 which breaks down
fundraising by source.
Table 4 provides strong support for the Early Fundraising Advantage Hypothesis. Even with
controls included, the estimated effect of Law Degree on early fundraising is substantial. During
their first 90 days, lawyers are estimated to raise an additional $44,840 in House elections and an
additional $202,360 in Senate elections.18 To compare, the estimated effects for Law Degree and
Held Elected Office—the standard proxy measure for candidate quality—are of similar magnitude.
The findings are robust to alternative modeling assumptions. The results are similar when
using a log-linear specification. They are also insensitive to varying the early fundraising window.
Whether narrowed to the first 30 days or extended to the first 180 days, the ratio at which lawyer
out-fundraise other candidates scales accordingly. Lastly, the results are robust to controlling for
13At-test confirms that the difference between groups is not statistically significant (t=0.49).
14If a candidate holds more than one professional degree, both indicator variables are set active. The reference
category is a candidate without a professional degree.
15Estimates of median household income for congressional districts are from Census.gov and are measured in
$000’s. For Senate contests, median household income is measured at the state-level.
16This necessitates dropping observations where candidates ran unopposed in the primaries.
17Current or former House members running for Senate seats are excluded.
18Note that the correct interpretation of the Tobit coefficients is the effect on the uncensored latent variable, not the
outcome. Marginal effects for Law Degree are included in the tables as Law Degree (dE[Y]/dx).
17

Table 4: Early Fundraising by Nonincumbents from Individuals and PACs during First 90 Days:
Tobit
House Senate
(1) (2) (3) (4)
Constant 23.10 2.43 29.09 2626.35
(9.81)(62.22)(106.00)(348.35)
Law Degree 44.84 39.00 202.36 221.77
(5.88)(6.93)(62.91)(66.00)
Medical Degree 13.25 9.33 64.96 30.90
(12.34)(14.02)(120.11)(132.46)
MBA 20.98 20.82 20.21 96.73
(8.76)(10.69)(95.63)(95.22)
PhD 16.57 35.48 46.13 30.88
(12.82)(16.87)(135.41)(136.01)
Held Elected Office 35.34 49.04 201.77 211.21
(5.00)(6.00)(58.03)(60.46)
Filing Delay (Days) 0.01 0.04 0.03 0.17
(0.02)(0.03)(0.19)(0.23)
Female 11.48 9.96 219.59 220.78
(6.26)(7.48)(78.86)(78.16)
Political Staffer 38.24 37.57 6.15 61.24
(11.00)(12.96)(147.68)(144.03)
N Prim. Opponents14.39 29.70
(9.17)(125.79)
Open Seat 44.13 43.51
(6.06)(64.28)
Primary Challenger 27.88 67.46
(8.75)(92.29)
District Partisanship 15.89 49.38
(3.23)(32.61)
Median Household Income 0.77 1.87
(0.16)(2.38)
Republican 1.72 59.07
(5.53)(63.75)
2010 1.41 121.65
(5.65)(68.80)
2014 6.98 33.79
(6.01)(69.24)
Law Degree (dE[Y]/dx) 30.60 28.76 122.89 138.96
(4.01)(5.11)(38.20)(41.36)
Contest Fixed Effects
AIC 35455 26810 6266 5643
Log Likelihood -17709 -12684 -3115 -2703
N. Censored 195 116 51 46
Num. obs. 3,001 2,189 448 399
Dependent Variable: Total amounts raised from individuals and PACs during the first 90 days (000’s of $).
District Partisanship is the share of two-party presidential vote won by the candidate’s party.
Source: Author’s calculations.
18

candidates ideology. (See supplemental appendix for details.)
9 Professional Networks and Early Fundraising
The advice given to first-time candidates is unambiguous about the need to tap into the one’s
personal networks to raise funds early on (EMILY’s List, 2001). Candidates depend almost ex-
clusively on personal acquaintances to raise funds needed to jump-start their campaigns. An
implication of this is that the ability to fundraise early on depends more on personal connec-
tions than talent or appeal as a candidate. A candidate who is surrounded by affluent friends and
colleagues—especially ones who are seasoned donors or are accustomed to attending fundraising
events—should have little trouble fundraising early on. On the other hand, even the most com-
pelling candidates will struggle to keep pace if their personal networks are devoid of anyone who
fits the typical profile of a political donor.
Perhaps the single most important determinant of a candidate’s personal network is profession.
Lawyers, as a group, are extremely active political donors and tend to have deep pockets. More-
over, the legal industry is well connected with the business community, which can bring wealthy
clients into the fold. This suggests that lawyers’ early fundraising advantage owes in large part to
their professional networks.
Table 5 reports regression results for early fundraising for four professional groups.19 Again,
the models are estimated separately for House and Senate candidates.
The results show that candidates rely disproportionately on other members of their profession
for financial support. Lawyers running for the House and Senate are estimated to raise an additional
$23,330 and $95,540, respectively, from other lawyers during their first 90 days. Physicians enjoy
a similarly sized fundraising boost from other doctors.
Money raised from fellow lawyers accounts for about half of lawyers’ early fundraising pre-
mium. The donations from corporate executives account for much of the remainder. This is con-
sistent with the claim that lawyers’ professional networks might also encompass their clients. It
19Donors are required by the FEC to report their occupation and employer on itemized contributions. This
makes it possible to calculate the amounts raised from each profession. The mapping from self-reported occupa-
tional/employment information onto professional groups relied on an initial set of select terms associated with a given
profession. For example, donors listing their occupation as “lawyer” or “attorney” were coded as legal professionals.
19

Table 5: Early Fundraising from Professional Groups by Nonincumbents: Tobit
House Senate
$000’s
from
Lawyers
$000’s
from
Doctors
$000’s
from
Corp.
Execs
$000’s
from
Aca-
demics
$000’s
from
Lawyers
$000’s
from
Doctors
$000’s
from
Corp.
Execs
$000’s
from
Aca-
demics
Constant -6.59 -5.43 -16.97 -3.72 -55.54 -19.20 -81.47 -34.24
(2.96) (1.28) (3.73) (0.98) (32.21) (8.71) (33.48) (13.38)
Law Degree 23.33 4.44 15.27 3.41 95.54 17.77 62.31 22.81
(1.70) (0.74) (2.17) (0.56) (18.70) (5.11) (19.53) (7.75)
Medical Degree 1.75 22.61 -1.30 5.84 -46.64 20.20 -24.42 5.00
(3.64) (1.47) (4.63) (1.15) (38.57) (9.55) (38.63) (14.75)
MBA 4.32 2.44 9.07 2.08 -6.06 -1.12 17.39 -0.62
(2.63) (1.12) (3.26) (0.87) (28.71) (7.81) (29.42) (11.90)
PhD -3.39 0.63 -2.01 1.67 -0.41 6.26 -19.99 4.55
(3.87) (1.63) (4.85) (1.21) (41.51) (11.02) (42.83) (16.37)
Held Elected Office 10.96 4.25 14.34 3.23 61.81 19.80 73.47 26.67
(1.49) (0.64) (1.87) (0.49) (17.46) (4.74) (18.08) (7.26)
Filing Delay (Days) -0.00 -0.00 -0.00 -0.00 -0.10 -0.02 -0.05 -0.00
(0.01) (0.00) (0.01) (0.00) (0.06) (0.02) (0.06) (0.02)
Female 4.98 1.17 5.37 2.37 76.75 13.71 47.21 31.05
(1.86) (0.80) (2.33) (0.61) (23.17) (6.31) (24.21) (9.35)
Political Staffer 10.55 3.64 12.50 2.97 21.40 -3.56 5.43 10.26
(3.15) (1.35) (3.96) (1.00) (42.15) (11.95) (45.21) (16.84)
N Prim. Opponents1-3.54 0.42 -5.96 0.74 33.49 0.99 29.19 -2.06
(2.80) (1.19) (3.50) (0.92) (37.39) (10.06) (38.73) (15.64)
Open Seat 10.76 3.79 13.89 2.46 14.84 1.96 0.52 7.26
(1.79) (0.76) (2.25) (0.59) (19.20) (5.20) (19.96) (7.89)
Primary Challenger -11.91 -5.57 -12.18 -5.17 -55.49 -1.90 -61.81 -1.86
(2.73) (1.16) (3.36) (0.92) (28.53) (7.51) (28.78) (11.56)
District Partisanship 3.81 2.56 6.65 1.53 14.54 2.67 24.59 7.17
(0.98) (0.42) (1.23) (0.33) (9.82) (2.63) (10.06) (3.98)
Median Household Income 0.23 0.05 0.37 0.07 0.81 -0.27 0.37 0.11
(0.05) (0.02) (0.06) (0.02) (0.71) (0.20) (0.74) (0.30)
Republican -9.52 -0.69 6.49 -5.60 -30.83 -2.37 24.82 -26.42
(1.66) (0.71) (2.10) (0.55) (19.17) (5.17) (20.01) (7.81)
2010 1.07 0.88 3.08 1.06 39.75 7.75 32.03 7.49
(1.71) (0.73) (2.15) (0.58) (20.73) (5.61) (21.43) (8.55)
2014 1.92 0.88 5.15 1.75 14.59 2.70 20.41 2.48
(1.81) (0.77) (2.27) (0.60) (21.07) (5.69) (21.60) (8.68)
Law Degree (dE[Y]/dx) 11.08 2.02 7.45 1.18 44.01 8.14 30.07 7.93
(0.81) (0.34) (1.06) (0.19) (8.61) (2.34) (9.42) (2.69)
AIC 19674 15541 20789 11757 3918 3033 3955 2694
Log Likelihood -9819 -7753 -10377 -5860 -1941 -1498 -1959 -1329
N. Censored 1165 1307 1145 1669 163 179 163 228
Num. obs. 3001 3001 3001 3001 448 448 448 448
Dependent Variable: Total contributions from professional groups during candidate’s first 90 days (in 000’s of $).
Only itemized contributions are included in the totals. Standard errors are in parentheses.
20

also reflects the significant overlap between legal and corporate communities. Most law firms are
primarily set up to provide services for corporate clients. Moreover, a significant percentage of
lawyers are employed by corporations as in-house counsel. Lawyers also account for a surprising
share of corporate executives. As of 2012, 46 CEOs at Fortune 500 firms had law degrees.20
10 Does Early Fundraising Really Explain Electoral Success?
The general finding in the academic literature is that money does matter in elections but only to
a degree (Jacobson, 1983, 1985; Gerber, 1998). However, by focusing on general elections, the
literature has tended to downplay money’s influence on elections. In recent elections, between 80
and 90 percent of congressional races have either been uncontested or in districts that strongly favor
one or the other party, meaning that a strong fundraising performance may improve a candidate’s
vote share in by a few percentage points but will rarely prove decisive.
This same logic does not apply to primary elections. The two factors that make general elec-
tion outcomes so predictable, party and incumbency, generally do not apply to primary contests. In
their place, fundraising is easily the most reliable indicator of success. The bivariate relationship
between fundraising and primary election outcomes is shown in Figure 4, which plots the pre-
dicted probability of winning contested primaries conditional on a candidate’s fundraising share.
It reveals a tight association between fundraising and primary election outcomes.
One challenge in estimating campaign spending effects is that fundraising might be endoge-
nous to the probability of winning. That is, the relationship shown in Figure 4 could be driven by
investment-oriented donors who stand to benefit only if candidates they support end up winning.
There is evidence that endogeneity bias of this sort presents less of a concern for primary elec-
tions than general elections. Bonica (2017) offers evidence of a causal relationship between early
fundraising and primary election outcomes using an instrumental variable design to estimate the
effect of early fundraising on vote shares in primary elections.21
20Curiously, MBAs do not enjoy a fundraising premium of quite the same size from corporate executives. This
might reflect the greater number and variety of MBA programs. Degree-granting institutions with MBA programs
vastly outnumber those that offer legal or medical degrees. Moreover, many MBA programs offer classes online,
which likely curtails alumni-network effects.
21The first instrumental variable specification follows earlier studies in using professional degree-type to instrument
for fundraising (Gerber, 1998; Lau and Pomper, 2002; Diermeier, Keane, and Merlo, 2005). The second instrumental
21

Figure 4: Predicted Probability of Winning Contested Primary Elections Conditional on Normal-
ized Share of Total Fundraising
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
Candidate's Share of Total Fundraising in Primary Contest
Predicted Pr(Win Primary)
The response curve is fit from a bivariate probit model. Shaded areas indicate the 95 percent
confidence intervals.
Source: Author’s calculations.
The informational cues that make general election so predictable are of little use in most pri-
mary contests. District partisanship, the single most important predictor in general elections, does
not discriminate between candidates in partisan primaries. Incumbency status is only informative
in the subset of primary contests where a sitting incumbent faces primary challengers. Meanwhile,
polling data for competitive primaries are scarce, and when available, are subject to difficult to
predict variation in turnout among the primary electorate. This gives investment-minded donors
out to pick winners very little to go on. For these donors, investing early on is an unnecessarily
risky prospect, especially when most other races offer a candidate who is a sure bet. This is con-
sistent with corporations and trade PACs accounting for a tiny fraction of early money raised by
variable specification uses average income in the zip code listed by candidates in their initial FEC filings to instrument
for fundraising.
22

nonincumbents.22
There are also practical reasons for why early fundraising should be instrumental in primary
elections. First, the funds raised early on provide vital resources for candidates to reinvest in
their campaigns. Even the most talented candidates struggle without the ability to hire competent
campaign staffers. Second, campaigns cannot operate at full capacity absent sufficient funding.
Candidates commonly cite fundraising difficulties as a reason for ending their campaigns. Early
fundraising is a strong predictor of whether a candidate drops out before election day (Bonica,
2017; Hassell, 2016).
Estimating Professional Fundraising Effects with Probability of Success Held Constant Ad-
ditional evidence can be had by showing that the early fundraising advantage for lawyers holds
even for candidates contesting seats they are certain to lose. I group nonincumbent House candi-
dates with respect to competitiveness. One group includes candidates running in seats that their
party’s presidential candidate lost by a margin of 20 or more points. The probability of winning
these seats is vanishingly small. During the period of study, not one candidate out of the 647 who
tried succeeded in overcoming a margin of 20 points or more. As such, any contribution made
to these candidates is effectively “wasted” if the objective is to help the candidate get elected. I
further limit the sample to candidates who had not previously held political office to isolate the
effects of professional networks. Focusing on candidates in unwinnable contests should isolate the
effect of professional networks from assessments of candidate viability. The regression results are
reported in Table 6.
Even in contests that are unwinnable, the fundraising advantage for lawyers persists. Given the
minimal likelihood of winning, one can safely rule out that rational beliefs about candidate viability
are driving the fundraising advantage. Far more likely is that the early fundraising advantage stems
from lawyers giving to fellow lawyers for reasons having to do with personal and professional
relationships.
22This does not preclude that less access-oriented donors might still condition on viability to avoid “wasting” their
contributions. Hall and Snyder (2015) find that donors tend to clump their contributions on the candidates who are
the top two vote-getters. This suggests that donors behave strategically by choosing whether to support candidates in
the primaries based on their chances of success. However, this pattern could also be consistent with early fundraising
being essential to establishing a campaign’s viability in the eyes of donors and party leaders.
23

Table 6: Early Fundraising (in 000’s of $) by House Candidates in Unwinnable Contests: Tobit
$000’s $000’s $000’s
from from from
All Donors Lawyers Doctors
Constant 15.35 6.65 2.01
(12.47)(3.71)(2.54)
Law Degree 29.65 13.22 4.43
(7.89)(2.17)(1.54)
Medical Degree 6.89 0.92 14.14
(11.81)(3.42)(2.14)
MBA 16.29 4.32 1.14
(9.61)(2.74)(1.95)
PhD 6.40 3.79 0.04
(15.04)(4.67)(3.00)
Filing Delay (Days) 0.00 0.01 0.00
(0.02)(0.01)(0.00)
Female 6.78 1.01 0.62
(6.96)(2.03)(1.39)
Primary Opponents14.89 1.65 0.43
(2.40)(0.70)(0.48)
Open Seat 26.07 7.46 4.28
(7.30)(2.08)(1.41)
District Partisanship 1.07 0.22 0.21
(0.42)(0.13)(0.09)
Median Household Income 0.26 0.11 0.07
(0.17)(0.05)(0.03)
Republican 5.49 5.38 0.32
(6.96)(2.01)(1.39)
2010 7.94 1.78 0.83
(6.24)(1.86)(1.28)
2014 0.14 0.79 1.46
(6.83)(2.03)(1.38)
Law Degree (dE[Y]/dx) 18.48 4.75 1.47
(4.92)(0.78)(0.51)
AIC 5154 2268 1929
Log Likelihood -2562 -1119 -950
N. Censored 55 278 300
Num. obs. 517 517 517
Dependent Variable: Total contributions raised during candidate’s first
90 days (in 000’s of $).
[Source] Author’s calculations.
11 Why the Overrepresentation of Lawyers Matters
Electing so many lawyers is not without consequence. In this section, I address ways in which
lawyer-legislators have influenced political outcomes.
24

Consequences for Descriptive Representation Contemporary democratic norms hold that leg-
islatures ought to reflect the diversity of the societies they represent. In addition to making
Congress less representative with respect class and occupation, lawyer-legislators, from both par-
ties, are disproportionately white and male.
The lack of diversity in Congress has been, at least in part, inherited from the legal profession.
Michelson (2013) finds that the U.S. legal profession lags behind international standards in gender
diversity. He estimates that as of 2010, women accounted for 32 percent of lawyers in the U.S.
as compared with 48 percent in the U.K. and 50 percent in France. Meanwhile, scholars have
uncovered systematic evidence of gender and racial biases within the legal profession (Sen, 2014;
Gorman, 2005; Phillips, 2005).
Table 7 reports the percentage of seats held by race and gender since 1992, referred to as the
“Year of the Woman” (Dolan, 1998), for lawyers and non-lawyers. During this period, only 9
percent of lawyer-legislators were women, compared to 19 percent of non-lawyers. When broken
down by party, the gender disparity becomes even more striking. Lawyer-legislators, in both par-
ties, are significantly less likely to be women. The differences are especially stark for Democrats.
Women accounted for just 13 percent of seats held by Democrats with law degrees versus 30 per-
cent of seats held by Democrats without law degrees.23
Table 7: Demographics of Members of Congress Weighted by Seat Shares (1993-2014)
% % % African % White
Degree Type Female Latino American Male N
All Law Degree 9 5 8 81 3,036
No Law Degree 19 4 8 72 3,419
Dem Law Degree 13 5 14 71 1,683
No Law Degree 30 8 17 54 1,511
Rep Law Degree 5 2 0 93 1,348
No Law Degree 11 2 1 86 1,896
[Source] Author’s calculations. Congressional Quarterly.
Policy Implications Barton (2010) has written extensively on the lawyer-judge bias in the legal
system. He argues that judges, having spent their formative years training and becoming profes-
23See Table A9 for regression results for multiple professional groups and controls for party and election cycle.
25

sionalized as lawyers, retain personal and professional biases that favor the legal profession. He
documents the numerous ways the courts have acted to promote the interests of lawyers.
Here, I show that these professional biases also carry over to a legislative setting. I adopt an
empirical strategy similar to that used to estimate party influence on roll call voting (McCarty,
Poole, and Rosenthal, 2001; Snyder and Groseclose, 2000; Clinton, Jackman, and Rivers, 2004).
For each roll call vote cast in the House and Senate during the 100-114th Congresses, vote choices
are modeled as a function of legislative ideology (as measured by DW-NOMINATE scores) and
lawyer-specific effects. Specifically, for roll call j, restricted and unrestricted models are fit with a
probit function,
Restricted :Yij ⇠b0+b1dwnomi(2)
Unrestricted :Yij ⇠b0+b1dwnomi+b2lawyeri(3)
Likelihood ratio tests are used to measure improvement in model fit. The likelihood ratio statistic
(LR) captures the relative importance of the lawyer-specific effects in explaining vote choices.
b2has a similar interpretation but provides additional information on the direction of the effect.
Evidence consistent with a lawyer-legislator bias should show that the lawyer-specific effects are
more important on votes that would directly impact the legal profession and align with the interests
of the legal profession.
26

Table 8: Top Congressional Roll Calls Ranked By Improvement in Model Fit
Title Question Description LR b2ABA
104 H.R. 956 Common Sense Product Liability
Legal Reform Act
Passage Enact comprehensive product liability reform; implement “loser
pays” rule in product liability suits.
21.0 -1.01 (0.23) 
107 H.R. 2563 Thomas Amdt. Adopt Limits personal injury claims in medical malpractice. 19.8 -1.14 (0.28) 
100 H.R. 1054 Military Medical Malpractice
Claims
Passage Permit active members of the military to sue the federal govern-
ment for malpractice occurring in U.S. military hospitals.
19.3 0.85 (0.20) +
104 H.R. 988 Attorney Accountability Act of
1995
Passage Enact civil litigation reform; limit attorney fees; sanction attor-
neys for frivolous law suits.
16.5 -1.02 (0.28) 
104 H.R. 956 Cox Amdt. Adopt Eliminate joint and several liability for noneconomic losses in
civil lawsuits involving interstate commerce.
16.2 -0.79 (0.20) 
106 H.R. 833 Conyers Amdt. Adopt An amendment to waive the provisions of title 11 relating to
small business debtors where they result in the loss of 5 or more
jobs.
15.6 -0.79 (0.20) 
107 S. 1052 Craig Amdt. Table Allow beneficiaries to bring personal injury claims against
health insurers for damages resulting from a denial of claim for
coverage.
15.2 -2.85 (1.12) 
107 H.R. 956 Flake Substitute To Smith Amdt. Adopt Amendment to prohibit funding to administer the Cuban Assets
Control Regulations with respect to travel.
14.8 -0.56 (0.15)
113 H.R. 4660 Scott Amdt. Adopt Eliminate all funding to Legal Services Corporation. 14.5 -0.81 (0.22) 
104 H.R. 956 Common Sense Legal Standards
Reform Act
Recommit Limit punitive damages in product liability suits to $1m. 14.5 1.10 (0.31) +
The table lists the top ten roll call votes (out of 28,430) by improvement in model fit associated with including lawyer-specific effects.
The column labeled LR (D) reports the likelihood ratio statistic. The column labeled b2reports the estimated coefficients and standard
errors (in parentheses) of the lawyer-specific effects. The column labeled ABA indicates the implied directionality based on the ABA’s
stated legislative priorities.
Source: Author’s calculations. Congressional roll call data is from Voteview.com.

Table 8 lists the top ten roll calls (out of 28,430 in total) ranked by improvement in model
fit. Nine of the top ten votes directly concern the legal profession. Topping the list is a vote
on the Common Sense Legal Standards Reform Act. This is followed by two votes pertaining to
medical malpractice liability and another vote on the Attorney Accountability Act of 1995. Also
reported in Table 8 are the estimated coefficients for the lawyer-specific effects (b2). An adjacent
column reports the directionality of the official legislative position of the American Bar Associa-
tion (ABA).24 Consistent with expectations of professional bias, the coefficients consistently align
with the ABA. Lawyer-legislators are significantly less likely to support legislation that would cap
awards for damages, limit product or medical liability, or regulate attorney fees. Meanwhile, they
are more likely to support legislation that would remove constraints on filing suit, increase funding
for the Legal Services Corporation, or promote fee-shifting provisions in public interest suits.
The above demonstrates that lawyer-legislators look and vote differently than their peers. This
should further allay concerns that a lawyer-dominated Congress is inconsequential. To the contrary,
it matters a great deal for policies in which lawyers have a direct stake.
12 Concluding Remarks
This study rekindles one of the most enduring questions in the study of American politics: Why
are so many lawyers elected to Congress?
Although the decision to run for office rests with the individual, campaigning is not a solitary
pursuit. To channel a prominent lawyer-turned-politician, “it takes a village” to mount a successful
campaign. The support provided by the legal profession is of two types. First, the structure and
professional norms of the legal profession create incentives that tilt the calculus in favor of run-
ning for political office. Running for office is less costly and more rewarding in terms of career
development in the legal profession than for other professions. Second, the legal community pro-
vides organizational and financial backing to lawyer-candidates. While the incentive structure of
the legal profession does much to encourage its members to enter politics, the electoral success of
lawyers owes in large part to the resource advantage derived from their professional networks. In
24The ABA positions are coded based on a 118-page document published by its governmental affairs office that
details the ABA’s official positions on hundreds of legislative issues and specific bills (American Bar Association,
2016b).
28

particular, the sizable early fundraising advantage has helped sustain the largest and most enduring
representational imbalance in American politics.
This is not without consequence. Scholars and policymakers alike have arguably been far too
dismissive of the outsized role of lawyers in American politics. Congress has inherited many of
the demographic, organizational, and behavioral characteristics of the legal profession. Lawyer-
legislators, in both parties, are far less likely to be women or underrepresented minorities than
legislators drawn from other backgrounds. Congress is less diverse as a result. Lawyers have also
shaped the inner-workings of Congress. Many of the institutional norms and practices, from the
emphasis on proper procedure and processes to the distinctively prosecutorial style of congres-
sional hearings, can be traced back to the legal profession. In many respects, serving in Congress
requires one to think and act like a lawyer. Along these lines, Miller has convincingly argued that
in shaping American political institutions, lawyers have produced a legalistic, procedural-oriented
government dominated by “lawyers’ ways, lawyers’ language, as well as lawyers’ approaches to
problem-solving” (Miller 1995, 162).
Future research might expand on the results presented here on the policy implications of elect-
ing so many lawyers. Despite the legal services market’s estimated size of $437 billion (Legal
Executive Institute, 2016)—which would rank it as the 15th largest state by economic output be-
hind Washington and ahead of Indiana—the industry remains entirely self-regulated. The ABA has
fiercely defended the independence of the legal profession and has successfully lobbied against,
or simply exempted the legal industry from, any legislation that it claims would infringe on the
profession’s self-proclaimed right to self-regulation. As shown in Table 8, the ABA’s lobbying
efforts have benefited enormously from the overrepresentation of lawyers.
Given the importance of the legal system, the industry’s insistence on complete self-regulation
naturally spills over into other matters of public policy. This is seen with respect to tax avoid-
ance. Lawyers have played a central role in the development of what has been termed the “income
defense industry,” which caters to high net-worth individuals looking to minimize their tax lia-
bility through less conventional means. Lawyers have been instrumental in exploiting loopholes
with increasingly complex tax maneuvers. Meanwhile, the ABA has vigorously opposed subject-
ing lawyers to provisions in banking and financial legislation, such as the Bank Secrecy Act and
29

Dodd-Frank, intended to prevent tax evasion and money laundering, improve regulatory compli-
ance, and impose fair accounting standards (American Bar Association, 2016a). It has also likely
contributed to an underprovision of legal services for the poor and middle-class (Hadfield, 2008;
Rhode, 2004, 2015). With lawyers increasingly chasing the most remunerative work for wealthy
individuals and corporations, the market for more affordable legal services has been neglected.
This has resulted in a U.S. legal services sector that is easily the largest and most profitable in the
world but ranks a dismal 94th (out 113 countries) in terms of people’s ability to access and afford
of legal representation (World Justice Project, 2016).
The results presented here suggest a need to fundamentally rethink how and when money mat-
ters in elections. Early fundraising influences who runs for office and who wins. This has biased
the electoral process and representational outcomes in favor of the types of individuals who are best
positioned to fundraise. Even as financial barriers to entry have contributed to representational im-
balances, the candidate population is, by comparison, far more representative of the public than
are members of Congress. An implication of this is that electoral success is primarily determined
by personal connections rather than talent or appeal as a politician. This makes early fundraising a
vital area for future research.
30

References
American Bar Association. 2016a. “American Bar Association Fights To Pre-
serve The Independence of The Legal Profession.” Retrieved October 20, 2016
(http://www.americanbar.org/content/dam/aba/uncategorized/GAO/ABAAccomplishments_
RegofLawyers.authcheckdam.pdf).
American Bar Association. 2016b. “Legislative Policies of the American Bar Association.” Re-
trieved October 21, 2016 (https://www.americanbar.org/content/dam/aba/uncategorized/GAO/
legislativeissueslist.authcheckdam.pdf).
Barton, Benjamin H. 2010. The Lawyer-Judge Bias in the American Legal System. Cambridge,MA:
Cambridge University Press.
Black, Gordon S. 1972. “A Theory of Political Ambition: Career Choices and the Role of Struc-
tural Incentives.” The American Political Science Review 66 (1): 144–59.
Bonica, Adam. 2014. “Mapping the Ideological Marketplace.” American Journal of Political Sci-
ence 58 (2): 367–387.
Bonica, Adam. 2016. “Database on Ideology, Money in Politics, and Elections: Public Ver-
sion 2.0.” [Computer file]. Stanford, CA: Stanford University Libraries [distributor]. (http:
//data.stanford.edu/dime).
Bonica, Adam. 2017. “Early Fundraising, Professional Networks, and Electoral Success.” Election
Law Journal .
Broockman, David, Nicholas Carnes, Melody Crowder-Meyer, and Christopher Skovron. 2014.
“Who’s a Good Candidate? How Party Gatekeepers Evaluate Potential Nominees.” Work-
ing Paper available at http://sites.lsa.umich.edu/cskovron/wp-content/uploads/sites/233/2015/
02/broockman_carnes_crowdermeyer_skovron_whos_a_good_candidate.pdf.
Carnes, Nicholas. 2012. “Does the numerical underrepresentation of the working class in Congress
matter?” Legislative Studies Quarterly 37 (1): 5–34.
31

Carnes, Nicholas. 2013. White-Collar Government: The Hidden Role of Class in Economic Policy
Making. Chicago, IL: University of Chicago Press.
Clinton, Joshua D., Simon Jackman, and Douglas Rivers. 2004. “The Statistical Analysis of Roll
Call Data.” The American Political Science Review 98 (2): 355–70.
Cody, Carly. 2014. “Majority In Congress Are Millionaires.” Retreived Novem-
ber 10, 2016 (http://www.npr.org/sections/itsallpolitics/2014/01/10/261398205/
majority-in-congress-are-millionaires).
Crandall, Robert W, Vikram Maheshri, and Clifford Winston. 2011. First thing we do, let’s dereg-
ulate all the Lawyers. Washington, DC: Brooking Institution Press.
Diermeier, Daniel, Michael Keane, and Antonio Merlo. 2005. “A Political Economy Model of
Congressional Careers.” The American Economic Review 95 (1): 347–373.
Dolan, Kathleen. 1998. “Voting for Women in the" Year of the Woman".” American Journal of
Political Science 42 (1): 272–293.
EMILY’s List. 2001. “Making the Dough Rise: A manual for campaign fundraisers.” Available at
(https://www.ndi.org/files/Making%20the%20Dough%20Rise.pdf).
Eulau, Heinz, and John D Sprague. 1964. Lawyers in Politics: A study in professional convergence.
Indianapolis: Bobbs-Merrill Company.
Fong, Christian, and Justin Grimmer. 2016. "Discovery of Treatments from Text Corpora". In
Proceedings of the Annual Meeting of the Association for Computational Linguistics.
Fowler, Linda L, and Robert D McClure. 1990. Political Ambition: Who decides to run for
Congress. New Haven, CT: Yale University Press.
Friedman, Lawrence M. 1985. Total justice. New York, NY: Russell Sage Foundation.
Gallup. 2015. “Honesty/Ethics in Professions.” Retrieved November 3, 2016 (http://www.gallup.
com/poll/1654/honesty-ethics-professions.aspx).
32

Gerber, Alan. 1998. “Estimating The Effect of Campaign Spending on Senate Election Outcomes
Using Instrumental Variables.” American Political Science Review 92 (02): 401–411.
Gorman, Elizabeth H. 2005. “Gender Stereotypes, Same-Gender Preferences, And Organizational
Variation In The Hiring Of Women: Evidence From Law Firms.” American Sociological Review
70 (4): 702–728.
Hadfield, Gillian K. 2000. “The Price of Law: How the market for lawyers distorts the justice
system.” Michigan Law Review 98 (4): 953–1006.
Hadfield, Gillian K. 2008. “Legal Barriers to Innovation: The growing economic cost of profes-
sional control over corporate legal markets.” Stanford Law Review 60: 101–146.
Hain, Paul L, and James E Piereson. 1975. “Lawyers and Politics Revisited: Structural advantages
of lawyer-politicians.” American Journal of Political Science 19 (1): 41–51.
Hainmueller, Jens, Daniel J. Hopkins, and Teppei Yamamoto. 2014. “Causal Inference in Con-
joint Analysis: Understanding Multi-Dimensional Choices Via Stated Preference Experiments.”
Political Analysis 22 (1): 1–30.
Hall, Andrew B., and James M. Snyder, Jr. 2015. “Information and Wasted Votes: A Study of U.S.
Primary Elections.” Quarterly Journal of Political Science 10 (4): 433–459.
Halliday, Terence Charles. 1979. Parameters of Professional Influence: Policies and Politics of
The Chicago Bar Association, 1945-70 PhD thesis University of Chicago, Dept. of Sociology.
Hassell, Hans JG. 2016. “Party Control of Party Primaries: Party Influence in Nominations for the
US Senate.” The Journal of Politics 78 (1): 75–87.
Jacobson, Gary C. 1983. Money in Congressional Elections. New Haven: Yale University Press.
Jacobson, Gary C. 1985. “Money and Votes Reconsidered: Congressional Elections, 1972-1982.”
Public Choice 47 (1): 7–62.
Jacobson, Gary C., and Samuel Kernell. 1983. Strategy and Choice in Congressional Elections.
New Haven: Lexington.
33

Lau, Richard R, and Gerald M Pomper. 2002. “Effectiveness of Negative Campaigning in US
Senate Elections.” American Journal of Political Science 46 (1): 47–66.
Legal Executive Institute. 2016. “The Size of the US Legal Market.” http://legalexecutiveinstitute.
com/the-size-of-the-us-legal-market-shrinking-piece-of-a-bigger-pie-an-lei-graphic/ [Ac-
cessed Jan 10, 2017].
Matter, Ulrich, and Alois Stutzer. 2015. “The Role of Lawyer-Legislators in Shaping the Law:
Evidence from Voting on Tort Reforms.” Journal of Law and Economics 58 (2): 357–384.
McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2001. “The Hunt for Party Discipline in
Congress.” The American Political Science Review 95 (3): 673–87.
Michelson, Ethan. 2013. “Women in the Legal Profession, 1970-2010: A Study of the Global
Supply of Lawyers.” Indiana Journal of Global Legal Studies 20 (18): 1071–1137.
Miller, Mark C. 1995. The High Priests of American Politics: The role of lawyers in American
political institutions. Knoxville, TN: Univ. of Tennessee Press.
Pew Research Center. 2013. “Public Esteem for Military Still High, Clergy in the Middle,
Lawyers at the Bottom.” Retrieved December 1, 2016 (http://www.pewforum.org/2013/07/11/
public-esteem-for-military-still-high/).
Phillips, Damon J. 2005. “Organizational Genealogies and The Persistence of Gender Inequality:
The case of Silicon Valley law firms.” Administrative Science Quarterly 50 (3): 440–472.
Podmore, David B. 1980. Solicitors and The Wider Community. London: Heinemann Educational
Books.
Polsby, Nelson. 1990. The Social Composition of Congress. In The US Congress and the German
Bundestag: Comparisons of Democratic Processes. Vol. 114 Westview Press Boulder, CO.
Posner, Richard A. 1993. “The Material Basis of Jurisprudence.” Indiana Law Journal 69 (1): 37.
Rhode, Deborah L. 2004. Access to Justice. New York, NY: Oxford University Press.
34

Rhode, Deborah L. 2015. The Trouble With Lawyers. New York, NY: Oxford University Press.
Robinson, Nick. 2015. “The Declining Dominance of Lawyers in U.S. Federal Politics.” HLS Cen-
ter on the Legal Profession Research Paper No. 2015-10. Available at http://ssrn.com/abstract=
2684731.
Rohde, David W. 1979. “Risk-Bearing and Progressive Ambition: The case of members of the
United States House of Representatives.” American Journal of Political Science 23 (1): 1–26.
Scheingold, Stuart. 1974. The Politics of Rights: Lawyers, public policy and social change. New
Haven, CT: Yale University Press.
Sekhon, Jasjeet S. 2011. “Multivariate and Propensity Score Matching Software with Automated
Balance Optimization: The Matching package for R.” Journal of Statistical Software 42 (7):
1–52.
Sen, Maya. 2014. “How Judicial Qualification Ratings May Disadvantage Minority and Female
Candidates.” Journal of Law and Courts 2 (1): 33–65.
Snyder, James M., Jr., and Tim Groseclose. 2000. “Estimating Party Influence in Congressional
Roll-Call Voting.” American Journal of Political Science 44 (2): 193–211.
Tocqueville, Alexis de. 1840. Democracy in America. Vol. 1-2 Saunders and Otley.
Wilson, James. 1989. Bureaucracy: What Government Agencies Do And Why They Do It. Basic
Books.
World Justice Project. 2016. “WJP Rule of Law Index 2016.” [online] Available at: http://data.
worldjusticeproject.org/ [Accessed 5 June. 2017].
35

Supplemental Appendix: For Online Publication
1

A Balance Statistics
Figure A1: Covariate Balance Before and After Matching
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
Normalized Fundraising Share
Prev. Held Office
Open Seat
Senate
Legislative Aide
Competitive Seat
Seat Unlikely
2012
Female
2010
2014
Primary Challenge
Safe Seat
Republican
−0.1 0.0 0.1 0.2 0.3
Mean Differences
Sample
●
●
Unadjusted
Adjusted
Covariate Balance
Note: Law Degree is treatment.
2

B Early Fundraising Models with Logged-Dependent Variable
Table A1: Early Fundraising by Nonincumbents from Individuals and PACs during First 90 Days:
Tobit, Logged-Dependent Variable
House Senate
(1) (2) (3) (4)
Constant 8.37 9.11 6.42 12.10
(0.24) (1.33) (0.74) (2.60)
Law Degree 0.82 0.68 2.01 2.14
(0.14) (0.15) (0.45) (0.49)
Medical Degree 0.27 -0.13 -0.34 0.50
(0.30) (0.30) (0.85) (0.97)
MBA 0.81 0.83 -0.08 0.24
(0.21) (0.23) (0.67) (0.70)
PhD 0.07 -0.34 -0.32 -0.14
(0.31) (0.36) (0.95) (0.99)
Held Elected Office 0.97 1.12 1.49 1.46
(0.12) (0.13) (0.41) (0.45)
Filing Delay (Days into Cycle) 0.00 0.00 0.00 -0.00
(0.00) (0.00) (0.00) (0.00)
Female 0.38 0.29 1.49 1.80
(0.15) (0.16) (0.56) (0.58)
Political Staffer 1.02 0.96 1.92 0.84
(0.27) (0.28) (1.06) (1.08)
N Prim. Opponents1-0.75 0.64
(0.22) (0.89)
Open Seat 0.64 0.12
(0.15) (0.46)
Primary Challenger -0.92 -1.84
(0.21) (0.65)
District Partisanship 0.39 0.42
(0.08) (0.23)
Median Household Income 0.01 -0.01
(0.00) (0.02)
Republican 0.09 1.15
(0.13) (0.45)
2010 0.18 0.59
(0.14) (0.49)
2014 0.07 0.42
(0.15) (0.49)
Law Degree (dE[Y]/dx) 0.82 0.68 1.98 2.13
(0.14) (0.15) (0.44) (0.48)
AIC 14928 11078 2416 2241
Log Likelihood -7446 -4818 -1190 -1002
N. Censored 195 116 51 46
Num. obs. 2,998 2,186 448 399
3

Table A2: Early Fundraising from Professional Networks: Tobit, Logged-Dependent Variable
House Senate
$000’s
from
Lawyers
$000’s
from
Doc-
tors
$000’s
from
Corp.
Execs
$000’s
from
Aca-
demics
$000’s
from
Lawyers
$000’s
from
Doc-
tors
$000’s
from
Corp.
Execs
$000’s
from
Aca-
demics
Constant 3.30 1.32 2.05 -0.04 1.32 -0.15 1.37 -1.29
(0.48) (0.51) (0.51) (0.59) (1.20) (1.22) (1.28) (1.42)
Law Degree 3.30 1.50 1.90 2.19 4.15 2.67 2.73 3.31
(0.28) (0.30) (0.30) (0.34) (0.71) (0.72) (0.76) (0.83)
Medical Degree 1.30 4.56 0.70 3.38 -1.58 2.03 -0.57 1.30
(0.60) (0.61) (0.63) (0.70) (1.41) (1.36) (1.47) (1.56)
MBA 1.25 1.10 1.62 1.53 0.64 0.49 0.99 0.70
(0.43) (0.45) (0.45) (0.52) (1.07) (1.09) (1.14) (1.25)
PhD 0.25 0.85 0.23 1.89 -1.08 -0.46 -1.42 1.36
(0.63) (0.65) (0.66) (0.73) (1.55) (1.56) (1.65) (1.74)
Held Elected Office 2.17 1.92 2.45 2.54 2.54 2.84 3.09 3.26
(0.24) (0.26) (0.26) (0.30) (0.66) (0.67) (0.70) (0.77)
Filing Delay (Days into Cycle) -0.00 -0.00 -0.00 -0.00 -0.00 -0.00 -0.00 -0.00
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Female 0.86 0.63 1.18 0.70 2.42 1.69 1.89 2.91
(0.30) (0.32) (0.32) (0.37) (0.88) (0.90) (0.95) (1.01)
Political Staffer 2.07 1.89 2.02 2.69 2.86 0.81 1.77 3.58
(0.52) (0.55) (0.55) (0.61) (1.63) (1.68) (1.77) (1.83)
N Prim. Opponents1-1.39 -0.40 -1.33 0.25 1.44 1.54 1.57 1.26
(0.45) (0.48) (0.47) (0.55) (1.41) (1.42) (1.50) (1.66)
Open Seat 0.90 1.55 1.29 1.01 0.35 0.31 -0.55 1.02
(0.29) (0.31) (0.31) (0.36) (0.72) (0.73) (0.77) (0.84)
Primary Challenger -3.02 -2.66 -2.63 -3.91 -3.01 -2.13 -3.20 -2.22
(0.44) (0.46) (0.46) (0.55) (1.06) (1.06) (1.11) (1.23)
District Partisanship 0.80 0.90 0.96 1.03 0.28 0.83 1.02 1.07
(0.16) (0.17) (0.17) (0.20) (0.37) (0.37) (0.39) (0.42)
Median Household Income 0.03 0.01 0.03 0.03 -0.00 -0.04 -0.02 -0.02
(0.01) (0.01) (0.01) (0.01) (0.03) (0.03) (0.03) (0.03)
Republican -1.22 -0.40 0.87 -3.20 0.45 0.86 1.37 -1.74
(0.27) (0.28) (0.28) (0.33) (0.72) (0.73) (0.77) (0.83)
2010 0.21 0.06 0.65 0.72 1.24 1.40 1.10 0.40
(0.28) (0.29) (0.29) (0.34) (0.78) (0.79) (0.83) (0.91)
2014 0.17 0.30 0.74 1.01 0.95 1.01 0.90 0.65
(0.30) (0.31) (0.31) (0.36) (0.79) (0.80) (0.84) (0.92)
Law Degree (dE[Y]/dx) 2.40 0.99 1.38 1.09 3.15 1.87 2.05 1.87
(0.20) (0.20) (0.22) (0.17) (0.54) (0.51) (0.57) (0.47)
AIC 13726 13086 14044 10963 2166 2078 2209 1816
Log Likelihood -6845 -6525 -7004 -5463 -1065 -1021 -1086 -890
N. Censored 1165 1307 1145 1669 163 179 163 228
Num. obs. 3000 3001 3000 3001 448 448 448 448
4

C Varying Early Fundraising Window
Table A3: Early Fundraising by Nonincumbents from Individuals and PACs during the First 30,
90, and 180 Days
First 30 Days ($000’s) First 90 Days ($000’s) First 180 Days ($000’s)
Constant -5.12 23.10 50.99
(3.98) (9.81) (18.95)
Law Degree 16.17 44.84 83.17
(2.36) (5.88) (11.70)
Medical Degree 6.88 13.25 18.40
(4.96) (12.34) (24.71)
MBA 7.51 20.98 38.03
(3.52) (8.76) (17.41)
PhD -2.61 -16.57 -3.10
(5.20) (12.82) (25.57)
Held Elected Office 11.08 35.34 83.77
(2.02) (5.00) (9.72)
Filing Delay (Days into Cycle) 0.02 0.01 -0.04
(0.01) (0.02) (0.03)
Female 4.28 11.48 22.87
(2.52) (6.26) (11.49)
Political Staffer 12.09 38.24 75.31
(4.40) (11.00) (21.95)
N Prim. Opponents1-2.60 -4.39 34.33
(3.71) (9.17) (17.55)
Open Seat 16.96 44.13 73.20
(2.44) (6.06) (12.09)
Primary Challenger -11.89 -27.88 -86.06
(3.57) (8.75) (14.87)
District Partisanship 6.15 15.89 27.95
(1.31) (3.23) (6.07)
Median Household Income 0.21 0.77 1.37
(0.07) (0.16) (0.32)
Republican 0.38 -1.72 -3.41
(2.23) (5.53) (11.01)
2010 1.19 1.41 10.36
(2.29) (5.65) (11.08)
2014 4.31 6.98 6.76
(2.43) (6.01) (11.84)
Law Degree (dE[Y]/dx) 9.73 30.60 57.36
(1.42) (4.01) (8.07)
AIC 27006 35455 40280
Log Likelihood -13485 -17709 -20122
N. Censored 545 195 151
N. Obs. 3,001 3,001 3,001
5

D Modeling Success in Primary Elections as a Binary
Outcome
Table A4: Candidate Success in Competitive Primary Elections: Probit, Marginal Effects
Pre-Matching Post-Matching
(1) (2) (3) (4)
Constant 0.41 0.43 0.33 0.44
(0.02)(0.05)(0.04)(0.08)
Law Degree 0.07 0.04 0.01 0.05
(0.02)(0.02)(0.03)(0.03)
Primary Competitors10.66 0.26 0.60 0.04
(0.06)(0.09)(0.10)(0.15)
Fundraising Share 0.69 0.84
(0.04)(0.07)
Raised $100K in first 90 Days 0.12 0.14
(0.03)(0.04)
Primary Challenger 0.16 0.16
(0.02)(0.05)
Competitive Seat 0.07 0.04
(0.03)(0.05)
Safe Seat 0.11 0.12
(0.03)(0.06)
Political Staffer 0.12 0.11
(0.05)(0.06)
Ideological Extremity 0.00 0.02
(0.01)(0.02)
Female 0.06 0.05
(0.03)(0.05)
Held Elected Office 0.16 0.16
(0.02)(0.04)
Republican 0.04 0.02
(0.02)(0.04)
Open Seat 0.04 0.04
(0.02)(0.04)
Senate 0.00 0.01
(0.03)(0.04)
2012 0.03 0.07
(0.02)(0.04)
2014 0.01 0.02
(0.03)(0.04)
AIC 2787.32 1996.70 1264.44 882.38
Log Likelihood 1390.66 981.35 629.22 424.19
Deviance 2781.32 1962.70 1258.44 848.38
Num. obs. 2387 2387 1007 1007
Dependent Variable: Candidate won primary contest.
In coding outcomes, no distinction is made between candidates who withdraw before the primary elections and those who exit
after being defeated. To adjust for primary competition, Primary Competitors1is calculated as 1/Np, where Npis the number
of candidates in primary contest p. In order to normalize fundraising across district-level primary contests, each candidate’s
fundraising total is divided by the total sum raised by all other candidates competing in the primary contest.
6

E Modeling Primary Election Outcomes Separately for
Lawyers and Non-Lawyers
Table A5: Candidate Success in Competitive Primary Elections: Probit, Marginal Effects
Non-lawyers Lawyers
Constant 0.48 0.58
(0.04)(0.09)
Fundraising Share 0.63 0.90
(0.05)(0.09)
Raised 00K in first 90 Days 0.10 0.16
(0.03)(0.05)
Primary Competitors10.29 0.10
(0.10)(0.19)
Primary Challenger 0.17 0.12
(0.03)(0.07)
Competitive Seat 0.02 0.09
(0.03)(0.06)
Safe Seat 0.09 0.00
(0.03)(0.07)
Legislative Staffer 0.10 0.15
(0.06)(0.09)
Ideological Extremity 0.01 0.05
(0.01)(0.03)
Female 0.07 0.00
(0.03)(0.06)
Held Elected Office 0.15 0.15
(0.02)(0.05)
Republican 0.07 0.02
(0.03)(0.05)
Open Seat 0.03 0.07
(0.03)(0.05)
Senate 0.01 0.00
(0.03)(0.06)
2012 0.02 0.06
(0.03)(0.05)
2014 0.02 0.04
(0.03)(0.06)
AIC 1518.85 490.22
Log Likelihood 743.43 229.11
Deviance 1486.85 458.22
Num. obs. 1,812 575
Dependent Variable: Candidate won primary contest.
7

F Early Fundraising Models Controlling for Ideology
Table A6: Early Fundraising by Nonincumbents during First 90 Days: Tobit
House Senate
(1) (2) (3) (4)
Constant 28.83 8.99 21.20 2529.17
(10.30)(62.80)(116.50)(349.02)
Law Degree 40.57 37.85 165.57 217.11
(6.16)(7.39)(68.72)(70.52)
Medical Degree 11.63 16.60 74.57 50.96
(13.03)(15.00)(135.96)(147.70)
MBA 15.81 16.23 51.15 134.74
(9.12)(11.37)(103.06)(102.59)
PhD 20.08 35.66 54.54 45.89
(13.19)(17.60)(148.98)(147.73)
Held Elected Office 27.33 40.31 169.70 212.85
(5.22)(6.34)(62.81)(64.06)
Entry Delay (Days into Cycle) 0.03 0.07 0.02 0.62
(0.02)(0.03)(0.21)(0.27)
Ideological Extremity 17.67 21.53 0.16 78.59
(3.10)(4.02)(38.51)(39.05)
Female 11.66 10.77 201.87 178.29
(6.52)(7.93)(83.25)(80.59)
Former Legislative Aid 33.65 34.86 34.91 129.90
(11.24)(13.31)(156.86)(151.64)
Primary Competitors13.93 74.15
(9.61)(139.26)
Open Seat 43.36 49.91
(6.34)(69.80)
Primary Challenger 27.75 19.75
(9.34)(100.22)
District Partisanship 18.15 56.46
(3.43)(35.15)
Median Household Income 0.81 2.02
(0.17)(2.61)
Republican 1.05 69.82
(5.82)(69.51)
2010 0.12 111.05
(5.92)(75.06)
2014 9.02 32.24
(6.33)(75.51)
Law Degree (dE[Y]/dx) 28.54 28.95 106.69 145.14
(4.33)(5.65)(44.28)(47.14)
Contest Fixed Effects
AIC 33368 25144 5824 5251
Log Likelihood -16665 -11850 -2893 -2505
N. Censored 103 56 19 19
Num. obs. 2750 1998 389 349
Dependent Variable: Total amounts raised from individuals and PACs during the first 90 days (in 000’s of $).
Measures of median household income for congressional districts are from census.gov. For Senate contests, median
household income is measured at the state-level. Standard errors are in parentheses.
8

Table A7: Early Fundraising from Professional Groups by Nonincumbents: Tobit
House Senate
$000’s
from
Lawyers
$000’s
from
Doc-
tors
$000’s
from
Corp.
Execs
$000’s
from
Aca-
demics
$000’s
from
Lawyers
$000’s
from
Doc-
tors
$000’s
from
Corp.
Execs
$000’s
from
Aca-
demics
Constant -3.43 -4.32 -12.27 -2.78 -24.41 -10.06 -50.04 -23.19
(3.01) (1.29) (3.74) (1.00) (33.67) (9.03) (34.75) (13.93)
Law Degree 21.17 3.73 11.98 2.80 77.73 12.51 41.72 15.78
(1.73) (0.75) (2.17) (0.57) (19.47) (5.29) (20.18) (8.07)
Medical Degree 0.57 23.37 -3.19 5.83 -49.05 21.67 -32.38 4.55
(3.72) (1.51) (4.65) (1.17) (40.91) (10.16) (40.44) (15.72)
MBA 2.17 1.72 6.06 1.56 -24.01 -4.59 2.41 -5.82
(2.66) (1.13) (3.26) (0.88) (29.86) (7.99) (30.24) (12.25)
PhD -4.16 0.19 -2.99 1.34 15.12 9.28 -6.45 5.17
(3.88) (1.63) (4.81) (1.22) (43.67) (11.49) (45.18) (17.15)
Held Elected Office 7.80 2.97 9.68 2.38 41.32 14.04 49.10 19.11
(1.51) (0.64) (1.87) (0.50) (18.10) (4.86) (18.57) (7.50)
Filing Delay (Days) 0.00 0.00 0.01 -0.00 -0.09 -0.01 -0.05 0.01
(0.01) (0.00) (0.01) (0.00) (0.06) (0.02) (0.06) (0.03)
Ideological Extremity -8.59 -2.56 -11.96 -1.84 -30.84 -7.07 -48.67 -9.32
(1.02) (0.42) (1.26) (0.35) (13.20) (3.36) (14.17) (5.63)
Female 5.03 0.94 5.41 2.36 74.23 12.78 44.14 29.39
(1.87) (0.80) (2.33) (0.61) (23.69) (6.38) (24.52) (9.57)
Political Staffer 8.99 3.04 10.30 2.44 6.60 -4.40 5.50 9.53
(3.12) (1.34) (3.89) (1.00) (43.37) (11.96) (45.31) (17.02)
Primary Competitors1-3.30 0.28 -6.17 0.66 51.12 5.60 41.06 4.80
(2.84) (1.20) (3.51) (0.94) (39.29) (10.51) (40.57) (16.45)
Open Seat 9.94 3.49 12.82 2.24 18.25 2.56 4.94 8.12
(1.82) (0.77) (2.25) (0.60) (19.96) (5.36) (20.53) (8.20)
Primary Challenger -11.84 -5.22 -11.51 -5.09 -44.89 1.31 -55.55 2.76
(2.82) (1.19) (3.42) (0.95) (29.56) (7.73) (29.78) (11.96)
District Partisanship 4.66 2.85 7.60 1.70 18.46 3.45 30.13 8.68
(1.01) (0.43) (1.25) (0.34) (10.12) (2.69) (10.37) (4.12)
Median Household Income 0.25 0.05 0.38 0.07 0.84 -0.28 0.28 0.12
(0.05) (0.02) (0.06) (0.02) (0.74) (0.20) (0.76) (0.31)
Republican -9.95 -0.71 6.67 -5.89 -37.95 -3.91 30.49 -28.26
(1.69) (0.72) (2.11) (0.56) (19.87) (5.32) (20.71) (8.10)
2010 0.85 0.87 2.74 0.97 34.11 5.43 23.54 4.07
(1.74) (0.74) (2.15) (0.58) (21.54) (5.78) (22.17) (8.88)
2014 2.53 1.16 5.83 1.91 13.91 1.67 21.11 -0.41
(1.85) (0.79) (2.29) (0.61) (21.94) (5.87) (22.37) (9.02)
Law Degree (dE[Y]/dx)10.70 1.81 6.24 1.05 39.94 6.41 22.32 6.17
(0.87) (0.36) (1.13) (0.21) (10.00) (2.71) (10.79) (3.16)
AIC 19161 15111 20333 11505 3822 2950 3856 2639
Log Likelihood -9562 -7537 -10148 -5733 -1892 -1456 -1909 -1300
N. Censored 934 1073 900 1424 107 122 106 170
Num. obs. 2750 2750 2750 2750 389 389 389 389
Dependent Variable: Total contributions from professional groups during candidate’s first 90 days (in 000’s of $).
Measures of median household income for congressional districts are from census.gov and are measured in $000’s. For Senate
contests, median household income is measured at the state-level. Only itemized contributions are included in the totals. Standard
errors are in parentheses.
9

G Women in Congress by Professional Background
10

Table A8: Early Fundraising (in 000’s of $) by House Candidates in Unwinnable Contests: Tobit.
$000’s $000’s $000’s
from from from
All Donors Lawyers Doctors
Constant 16.39 5.56 2.03
(12.95)(3.68)(2.57)
Law Degree 29.11 12.59 3.97
(8.37)(2.19)(1.60)
Medical Degree 2.60 2.95 14.50
(12.39)(3.46)(2.17)
MBA 10.51 2.10 0.07
(9.76)(2.65)(1.93)
PhD 10.10 4.52 0.05
(15.08)(4.55)(2.95)
Filing Delay (Days) 0.02 0.01 0.00
(0.02)(0.01)(0.00)
Ideology Extremity 12.45 7.13 3.22
(4.06)(1.36)(0.92)
Female 7.87 1.68 0.21
(7.27)(2.03)(1.42)
Primary Competitors15.57 1.70 0.45
(2.48)(0.69)(0.48)
Open Seat 25.18 5.54 3.36
(7.86)(2.13)(1.46)
District Partisanship 1.12 0.23 0.20
(0.44)(0.13)(0.09)
Median Household Income 0.17 0.06 0.04
(0.18)(0.05)(0.04)
Republican 5.48 5.67 0.17
(7.38)(2.04)(1.44)
2010 4.66 0.30 0.41
(6.52)(1.86)(1.30)
2014 2.58 1.38 1.89
(7.24)(2.04)(1.41)
Law Degree (dE[Y]/dx) 19.03 4.79 1.42
(5.47)(0.83)(0.57)
AIC 4807 2145 1847
Log Likelihood -2387 -1057 -907
N. Censored 27 228 248
Num. obs. 461 461 461
Dependent Variable: Total contributions raised during candidate’s first
90 days (in 000’s of $).
Standard errors are in parentheses.
11

Table A9: Women in Congress by Professional Background: OLS
All Democrats Republicans
(Intercept) 0.21 0.24 0.08
(0.02) (0.03) (0.02)
Law -0.12 -0.20 -0.05
(0.01) (0.02) (0.01)
Medicine -0.03 -0.03 0.00
(0.03) (0.06) (0.03)
Business -0.06 -0.09 -0.03
(0.01) (0.02) (0.01)
Education 0.01 -0.06 0.10
(0.02) (0.02) (0.02)
Republican -0.13
(0.01)
Independent -0.23
(0.08)
1994 0.02 0.01 0.01
(0.02) (0.03) (0.02)
1996 0.03 0.04 0.01
(0.02) (0.03) (0.02)
1998 0.03 0.04 0.01
(0.02) (0.03) (0.02)
2000 0.04 0.07 0.01
(0.02) (0.03) (0.02)
2002 0.05 0.06 0.03
(0.02) (0.03) (0.02)
2004 0.06 0.08 0.03
(0.02) (0.03) (0.02)
2006 0.07 0.10 0.04
(0.02) (0.03) (0.03)
2008 0.06 0.09 0.03
(0.02) (0.03) (0.03)
2010 0.08 0.12 0.04
(0.02) (0.03) (0.02)
2012 0.09 0.17 0.02
(0.02) (0.03) (0.02)
2014 0.11 0.19 0.03
(0.02) (0.03) (0.02)
R20.06 0.06 0.02
Num. obs. 6508 3204 3285
Dependent Variable: Member is Female
[Source] Historical data on gender is from Congressional Quarterly. Professional background is coded based on entries
in the Biographical Directory of the U.S. Congress.
12