ArticlePDF Available

The Rise of Partisanship and Super-Cooperators in the U.S. House of Representatives


Abstract and Figures

It is widely reported that partisanship in the United States Congress is at an historic high. Given that individuals are persuaded to follow party lines while having the opportunity and incentives to collaborate with members of the opposite party, our goal is to measure the extent to which legislators tend to form ideological relationships with members of the opposite party. We quantify the level of cooperation, or lack thereof, between Democrat and Republican Party members in the U.S. House of Representatives from 1949–2012. We define a network of over 5 million pairs of representatives, and compare the mutual agreement rates on legislative decisions between two distinct types of pairs: those from the same party and those formed of members from different parties. We find that despite short-term fluctuations, partisanship or non-cooperation in the U.S. Congress has been increasing exponentially for over 60 years with no sign of abating or reversing. Yet, a group of representatives continue to cooperate across party lines despite growing partisanship.
Content may be subject to copyright.
The Rise of Partisanship and Super-
Cooperators in the U.S. House of
Clio Andris
*, David Lee
, Marcus J. Hamilton
, Mauro Martino
, Christian E. Gunning
John Armistead Selden
1 Department of Geography, The Pennsylvania State University, University Park, Pennsylvania, United
States of America, 2 Department of Urban Studies and Planning, Massachusetts Institute of Technology,
Cambridge, Massachusetts, United States of America, 3 Senseable City Lab, Massachusetts Institute of
Technology, Cambridge, Massachusetts, United States of America, 4 Santa Fe Institute, Santa Fe, New
Mexico, United States of America, 5 School of Human Evolution and Social Change, Arizona State
University, Tempe, Arizona, United States of America, 6 IBM Thomas J. Watson Research Center,
Cambridge, Massachusetts, United States of America, 7 Department of Entomology, North Carolina State
University, Raleigh, North Carolina, United States of America, 8 United States Senate Budget Committee,
Washington, District of Columbia, United States of America
It is widely reported that partisanship in the United States Congress is at an histor ic high.
Given that individuals are persuaded to follow party lines while having the opportunity and
incentives to collaborate with members of the opposite party, our goal is to measure the ex-
tent to which legislators tend to form ideological relationships with memb ers of the opposite
party. We quantify the level of cooperation, or lack thereof, between Democrat and Republi-
can Party members in the U.S. House of Representatives from 19492012. We define a
network of over 5 million pairs of representatives, and compare the mutual agreement rates
on legislative decisions between two distinct types of pairs: those from the same party and
those formed of members from different parties. We find that despite short-term fluctuations,
partisanship or non-cooperatio n in the U.S. Congress has been increasing exponentially for
over 60 years with no sign of abating or reversing. Yet, a group of representatives continue
to cooperate across party lines despite growing partisanship.
Americans today are represented by political figures who struggle to cooperate across party
lines at an unprecedented rate, resulting in high profile fiscal and policy battles, government
shutdowns, and an inab ility to resolve problems or enact legislation that guides the nations do-
mestic and foreign policy [1]. Partisanship has been attributed to a number of causes, including
the stratifying wealth distribution of Americans [2]; boundary redistricting [3]; activist activity
at primary elections [4]; changes in Congressional procedural rules [5]; political realignment in
the American South [6]; the shift from electing moderate members to electing partisan
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 1/14
Citation: Andris C, Lee D, Hamilton MJ, Martino M,
Gunning CE, Selden JA (2015) The Rise of
Partisanship and Super-Cooperators in the U.S.
House of Representatives. PLoS ONE 10(4):
e0123507. doi:10.1371/journal.pone.0123507
Academic Editor: Rodrigo Huerta-Quintanilla,
Cinvestav-Merida, MEXICO
Received: March 28, 2014
Accepted: March 4, 2015
Published: April 21, 2015
Copyright: © 2015 Andris et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
Data Availability Statement: Congressional Roll
Call Vote Data originally provided by the Office of the
Clerk of the U.S. House of Representatives.
Accessible through Govtrak at https://www.govtrack.
us/congress/votes. Copyright Civic Impulse, LLC,
(2014) Retrieved in bulk via instructions from https:// Data on
Congressional productivity and approval rate at:
Ornstein N, Mann T, Malbin M, Rugg A (2013) Vital
statistics on Congress. Washington, D.C.: The
Brookings Institution.
members [7] movement by existing members towards ideological poles [8]; and an increasing
political, pervasive media [9].
The individual representatives role in facilitating partisanship is less clear. Party affiliation
significantly shapes a legislators voting record [10], [11], so much that in some cases, a change
in a legislators party affiliation results in an immediate and significant realignment of voting
behavior towards the new party agenda [ 12]. This change is too rapid to be attributable to
contemporaneous changes in constituent ideology, indicating a disconnection between the
representative and his or her constituency. Party leaders also ensure obedience by offering
incentives such as the prospect of assigning a member to a favored committee or promoting
legislation crafted by the member to reach final voting stages, i.e. bringing legislation to
the floor [13]. As many have concluded [1], much is at stake with this type of party-driven
Despite party-level pressures, there are incentives for individual representatives to vote with
members of the opposite party on issues that are specific to a districts geography, such as aging
populations, natural resource management, veterans affairs, or regional concerns [14]. More-
over, regardless of party affiliation, pairwise relationships may form as a result of social interac-
tions such as sponsoring bills, interacting with lobbyists, creating trust networks for
communication, sharing ideas, garnering support for initiatives, negotiating provisions and
sharing ones own sense of ethics and orthopraxy. Vote trading, also known as logrolling, is an-
other incentive for cross-party cooperation [15]. Though difficult to quantify because vote
trading discussions are not public information, these would result in increased inter-party co-
operation on ideological votes.
Given that individuals are persuaded to follow party lines [1013], while having the oppor-
tunity and incentives to collaborate with members of the opposite party [14], [15], our goal is
to measure the extent to which legislators tend to form ideological relationships with members
of the opposite party. Specifically, we uncover cooperation rates between individual members
of Congress, by leveraging a comprehensive dataset of each legislators roll call vote decisions
in agreement or disagreement with each other legislator during a specific Congress. This pro-
cess results in a network of Congressional representatives. Such network structures have been
shown to predict future re-elections, define intra-Congressional communities and describe
temporal dynamics of Congresses [1621].
In studies that model Congressional representatives as nodes in a network, nodes are con-
nected with an edge based on a given similarity between nodes, such as bill co-authorship or
membership on the same committee [1621]. We connect nodes with similar voting records
on individual roll call votes, which represent similarities in ideology. Notably, the network
method differs from prevailing legislator partisanship indexing methods [2224] as the latter
require the subjective quantification of each member on a single (liberal-to-conservative) linear
scale (i.e. dimension). These dimensions are considered valuable because they temporally cor-
relate with instances of landmark time periods and events in U.S. History [24]. Distinctive and
groundbreaking, these dimensions are accepted as standard practice for quantifying polariza-
tion, as they serve as a reliable indicator of the political climate.
Yet, above methods are best used to gauge the behavior of entire systems, and not well-suit-
ed for discovering interpersonal patterns of agreement forged by pairs of representatives. The
network method is able to sidestep the following conside rations of the current partisans hip
measurement tools [2224]. First, when rating representatives in terms of a chosen vector of
decisions deemed important, the index can be (perhaps incorrectly) manipulated to match cor-
relation with events. The actual vote cocktail used to create the index, as well as how this value
is transformed to a linear value is not clear to the laypersonperhaps nor to the seasoned so-
cial scientist. Also, polarization scale seems to have an arbitrary minimum and maximum that
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 2/14
Funding: This work was supported by the National
Institute of Biomedical Imaging and Bioengineering
(NIBIB), a component of the National Institutes of
Health (NIH) (grant no T32EB009414) (to CEG),; John Templeton Foundation
(grant no 15075) (to CA,
MJH); and National Defense Science and
Engineering Graduate (NDSEG) FellowshipArmy
Research Office (to CA). The funders had no role in
study design, data collection and analysis, decision to
publish, or preparation of the manuscript. IBM
Thomas J. Watson Research Center provided
support in the form of a salary for author MM, but did
not have any additional role in the study design, data
collection and analysis, decision to publish, or
preparation of the manuscript. The specific role of this
author is articulated in the author contributions
Competing Interests: The authors have the
following interests: At the time of the study, David Lee
was a member of Senseable. Clio Andris and Mauro
Martino were past members. Senseable's partners
are a group of corporations, including AT&T, General
Electric, Audi, ENEL, SNCF. Mauro Martino is
employed by IBM Thomas J. Watson Research
Center. There are no patents, products in
development or marketed products to declare. This
does not alter their adherence to all the PLOS ONE
policies on sharing data and materials, as detailed
online in the guide for authors.
depends on the subjective choices of the creators. Secondly, when the difference between two
representatives index values is used describe the ideological distance between a pair of repre-
sentatives, as in [24], false similarities can occur. In this method, the index is centered at zero,
signaling neutrality, and increasingly strong members of one party (the other party) are in-
creasing positive (negative) numbers. However, two moderate members can each have a zero
index, but could actually disagree on every non-procedural issue. Thirdly, indexing methods
are described in whole by aggregate measures, such as mean of members indexes as indicators
of polarization [2224] which obfuscate the role of the individual. Instead, network methods le-
verage raw, disaggregate data on each members voting behavior to uncover how cross-party
pairs form organic relationships in Congress. More drawbacks to traditional index methods,
with a focus on of their inability to detect groups, are astutely described in [21].
In this article, we first examine the decl ine of representatives who agree with representatives
of the opposite political party on proposed legislation, and how this lack of collaborative voting
reflects changes partisanship over the past 60 years (19492012). Our results show how the rel-
ative ease of cross-party cooperation in the late 1960s and early 1970s leads to the decoupling
of the parties and the rise of a select few individuals who drive high rates of cross-party cooper-
ation. We next discuss the correlation between decreased cooperation and decreasing legisla-
tive productivity in the 1990s and 2000s. We finally interpret findings in terms of overall
trends in political climate, multiplicative growth processes, public behavior and the implica-
tions for the U.S. constituency.
Materials and Methods
We use roll call vote data from the U.S. House of Representatives from 1949 (commencement
of the 81st Congress) to 2012 (adjournment of 112nd Congress) (see Table 1) as provided by
the United States Office of the Clerk of the U.S. House of Representatives via Govtrak [26]as
described in [27], in a roll call vote, a representative chooses whether to respond (yay/nay)or
abstain from voting on a bill or motion. Abstentions are relatively rare, and are counted as
nays, as they do not support the legislation. Most abstentions come from members who are
absent or unable to vote on the majority of votes, and have no network connections (i.e. they
are not considered). Substantive roll call votes are proposed actions, bills and legislation regard-
ing topics that produce new laws, such as veterans benefits, the budget and health insurance.
Procedural roll call votes reflect votes on the organization and timing of the agenda [27], such
as motioning to recess. We do not discriminate between these types although the latter are
often unanimous votes and are largely excluded from the data set.
For all B(n,2) possible pairs of representatives in a given Congress, we count the number of
roll call votes where they voted the same way. We tally an agreement when a pair votes either
yay/yay or nay/nay. For example, Congressman A has voted similarly with Congressman
B five times more often than with Congressman C in a session, giving the A-B relationship five
times the weight of A-C. The result is a B(n,2)cell, weighted, undirected graph of pair-wise re-
lationships between representatives. Each pair is classified as either same-party (SP) if they
are members of the same political party, or cross-party (CP) if one representative is Republi-
can and the other Democrat. Independents are rare and are included as CP with all other non-
Independents. Independents are not listed as super-cooperators due to their tendency to be in
a cross-party pair with the majority of Congress. Representative absences are discarded. Agree-
ments are not normalized by total possible votes or any another factor.
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 3/14
There are a tota l of 3,424,343 cross-party (CP) pairs (those comprised of a single Republican
and a single Democrat) and 2,239,357 same-party (SP) Pairs (those comprised of two Demo-
crats or two Republicans) in the 60 years of our study (Table 1).
For each Congress, a threshold value is defined as the crossing point between dueling fre-
quency distributions (i.e. histograms) of CP and SP pair roll call agreements (Fig 1). For
Table 1. Summary Statistics of Congressional Representatives and Voting Records.
Number of Representatives, Starting Year, and Number of
Votes for Each Congress
Average Agreements for Different Types
of Pairs
Cross-Party (CP) Pair Behavior based
on Threshold Value (where Probability
Distributions Meet)
Congress Starting
Democrats Republicans Total
Cross-Party Pairs
Above the
Probability of a
CP pair
Appearing Above
the Threshold
81 1949 269 176 274 90.7 131.0 130.6 124 6383 0.067
82 1951 241 207 180 56.6 80.9 92.3 76 10552 0.106
83 1953 219 221 147 59.4 72.6 91.4 77 6985 0.072
84 1955 236 204 148 64.6 87.9 86.1 80 8427 0.088
85 1957 241 203 193 75.9 101.4 102.5 99 8903 0.091
86 1959 287 159 180 69.9 101.3 103.7 93 6633 0.073
87 1961 273 176 240 93.4 129.0 135.4 125 7548 0.079
88 1963 261 182 231 85.0 123.6 129.4 117 6376 0.067
89 1965 301 142 393 155.9 202.3 216.8 200 7949 0.093
90 1967 251 188 477 211.8 243.8 274.0 257 10029 0.106
91 1969 250 199 443 192.6 214.1 215.1 241 12672 0.127
92 1971 258 187 645 280.5 313.6 336.0 340 11458 0.119
93 1973 248 195 1070 502.1 589.7 590.5 595 12921 0.134
94 1975 294 148 1264 583.5 714.1 732.2 712 9560 0.110
95 1977 293 146 1537 766.4 872.3 934.0 889 10850 0.127
96 1979 280 160 1274 581.1 717.1 769.7 690 11631 0.130
97 1981 246 196 811 395.3 472.2 495.1 482 9830 0.102
98 1983 274 168 905 411.3 578.0 573.2 518 7939 0.086
99 1985 257 182 889 375.0 593.3 566.3 508 5337 0.057
100 1987 263 179 939 409.2 652.3 609.1 563 4807 0.051
101 1989 265 178 904 403.3 609.2 568.2 537 5630 0.060
102 1991 271 170 932 369.3 629.3 593.5 516 3283 0.036
103 1993 261 180 1122 407.1 792.4 794.7 612 1591 0.017
104 1995 207 231 1340 481.2 862.2 1078.1 763 3122 0.033
105 1997 211 232 1187 516.6 813.8 898.3 747 1501 0.015
106 1999 211 225 1214 605.3 903.0 930.6 786 2477 0.026
107 2001 213 226 996 499.4 748.6 782.3 659 1374 0.014
108 2003 208 230 1221 554.0 942.1 992.7 781 455 0.005
109 2005 202 236 1214 533.3 956.0 948.0 766 280 0.003
110 2007 242 205 1876 695.6 1487.3 1376.1 1122 181 0.002
111 2009 261 182 1655 799.4 1336.8 1276.8 1094 1371 0.014
112 2011 200 244 1606 425.3 1137.1 1297.9 838 1508 0.015
Note: These likelihoods can also be dened as expectations as described in [34].
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 4/14
Fig 1. Probability density functions of same-party and cross-party pairs over time. Probability density functions of the number of roll call vote
agreements between pairs of the same-party (SP) and those pairs of cross-party (CP) pairs. The plots show the steady divergence of CPs and SP agreement
rates over time. Above each distribution is the Congress number (81112), followed by the year the Congress commenced, and the number of total roll call
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 5/14
instance, the 109th Congress threshold value is at 766 agreement s (Table 1, graphically visible
in Fig 1). Although the value itself depends largely on the overall number of roll call votes dur-
ing a given Congress, the threshold signifies the value at which any random pair who exhibits
this number of agreements is equally likely to be a CP or SP pair. A pair found to the right (i.e.
with more vote agreements) is more likely to be of the same party (SP); to the left (i.e. with
fewer agreements), of two different parties (CP) (Fig 1). CP and SP pairs are nearly indistin-
guishable from one another in the 91st Congress , but are unmistakably differe nt today (Fig 1).
To find the individual legislators pairwise agreements over time, we construct a network of
representatives (nodes) connected with edges to other nodes if the pairs vote agreement rate is
above the threshold value for that particular Congress (Fig 2 ). This configuration illustrates the
parting of political parties through time while highlighting each individual. (Interactive visuali-
zations are available in S1 Database.)
votes during the two sessions of each Congress. Pairs with few agreements (below the local minima of a consistently- increasing CP distribution), including
representatives from Washington D.C., Puerto Rico are removed.
Fig 2. Division of Democrat and Republican Party members over time. Each member of the U.S. House
of Representatives from 19492012 is drawn as a single node. Republican (R) representatives are in red and
Democrat (D) representatives are in blue, party affiliation changes are not reflected. Edges are drawn
between members who agree above the Congress threshold value of votes. The threshold value is the
number of agreements where any pair exhibiting this number of agreements is equally likely to comprised of
two members of the same party (e.g. D-D or R-R), or a cross-party pair (e.g. D-R). Each node is sized relative
to its total number of connections; edges are thicker if the pair agrees on more votes. The starting year of
each 2-year Congress is written above the network. The network is drawn using a linear-attraction linear-
repulsion model with Barnes Hut optimization [33].
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 6/14
Cooperator pairs
Cross-party (CP) pairs above the threshold value (Table 1) are distinguished as cooperators.
These cooperators agree on roll call votes more often than a random SP pair. Cooperator prev-
alence has decreased by two orders of magnitude from the 1970s to 2000s. From 1967 to 1979,
Congress often had over 10,000 cooperators (max: 12,921) and was comprised of at least 10%
cooperators (max: 13.4%), i.e. at least 10% of CP pairs agreed on more issues than SP pairs. In
comparison, 20012010 held fewer than 1,500 cooperators (min: 181) with fewer than 1.5%
(min: 0.2%) of CP pairs acting as cooperators (Table 1). Longitudinally, partisanship/non-co-
operation has been increasing at an annual rate of about 5% over the last 60 years. The average
number of disagreements on roll call votes between CP pairs is increasing expon entially (Fig
3A), as illustrated by an exponential growth model in the form of y = c
which exhibits a fit
= 236.22, α =0.05, R
=0.88, p < 0.0001). This curve fits the exponential increase of the
raw number of votes disagreed upon per session. When vote disagreements are normalized by
Fig 3. Congressional cooperation rates over time. Four plots of Congressional non-cooperation through time shown as: (a) Average number of roll call
vote disagreements between cross-party (CP) pairs as a function of time. (b) The number of cooperator pairs (e.g. cross-party (CP) pairs who agree more
often than a random same-party (SP) pair) as a function of time. (c) The number of representatives involved in at least one cooperator pair as a function of
time. (d) The number of appearances each cooperator makes relative to all CPs over time evidences super-cooperators from the late 1990s to the present.
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 7/14
possible roll call votes, the trend shows high disagreement rates in the 1950s and early 1960s
(S1 Fig). Periods of cooperation and non-cooperation align with the findings of [ 24].
Though cooperator pairs are relatively infrequent today (Fig 3B) the pairs that exist are driven
by very few individuals (Fig 3C and 3D). We define a super-cooperator as a legislator who is
found in at least 5% of cooperator pairs during a given Congress. Super-cooperators such as
Rep. Ralph Hall (D-TX) guide 48% of all cooperator pairs (see S1 Table for each of 86 super-co-
operators). Rep. Hall, a senior Democrat from rural North Texas (largest city: Sherman), sin-
glehandedly drove nearly half of all cro ss-aisle partnerships by agreeing on past the threshold
with 220 of the 230 Republicans in the 108th Congress (Table 2). Similarly, Rep. Dan Boren
(D-OK), whose Oklahoma district (largest city: Muskogee) shares a border with Rep. Hall, con-
tributed to 42% of all cooperator pairs in the 109th session, by partnering with 119 different
Republicans (Table 2). Super-cooperators Rep. Dan Boren (D-OK) and Rep. Robert Cramer
(D-AL) together accounted for 71.4% of all cooperator pairs in the 109
Congress. Combined,
seven members accounted for 98.3% of all cooperator pairs in the 110
Congress (Fig 3D and
S1 Table). Amassing cooperation in the hands of very few legislators is a new phenomenon. Be-
fore 1990, the maximum participation for any one legislator in a cooperator pair was less than
5%, and often less than 1%.
Most super-cooperators are Democrats who hail from Texas (12 appearances), Mississippi
(7), Alabama (5), Louisiana, Indiana (4 appearances each), Georgia, Kentucky, Oklahoma,
Ohio, Pennsylvania and Virginia (3 each). The 104th Congress (19951996) had the most
super-cooperators (13), all of whom were Democrats, mostly from Southern states. Republican
Table 2. Top Super-Cooperators, Comprising at Least 15% of All Cooperator Pairs in a Specific Congress
Congress Representative Total CP Pairs above Threshold (i.e.
Cooperators) in Congress
Appearances as a Percentage of all
Cooperator Pairs in the Congress
108 Rep. Ralph Hall [D-TX-
455 220 48.351648
109 Rep. Dan Boren [D-OK-
280 119 42.50000
110 Rep. Christopher Smith
181 61 33.701657
113 Rep. Jim Matheson
521 172 33.013436
109 Rep. Robert Cramer
280 81 28.928571
110 Rep. Frank LoBiondo
181 31 17.127072
112 Rep. Jim Matheson
1508 235 15.583554
112 Rep. Dan Boren [D-OK-
1508 235 15.583554
112 Rep. Mike Ross [D-AR-
1508 232 15.384615
108 Rep. Robert Cramer
455 69 15.164835
108 Rep. Kenneth Lucas
455 69 15.164835
107 Rep. Ralph Hall [D-TX-
1374 208 15.138282
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 8/14
super-cooperator appearances are mostly limited to New York (10), New Jersey (5) and Mary-
land (4), largely in suburban areas outside New York City and Washington, D.C. This trend
may be shifting, as preliminary results from the 113th Congress show that the majority of
super-cooperators are Republican representatives from New York and New Jersey.
The few super-cooperators, who hand pick legislation and cooperate with members from
each party, despite threat of alienation from his or her party [28], [29], may be todays hallmark
example of carefully representing a constituency. These super-cooperators may earn powerful
reputations through single-handedly foraging the dwindling ties across divisive parties.
Comparison with prevailing statistical methods
We compare the CP pair cooperation rates, produced by the cooperator method, to the
DW-NOMINATE multi-dimensional scaling methods polarization score, (the difference in
Party means of the first dimension) as well as the overlap, (the ideological overlap between
the Democratic and Republican Parties) [25](S2 Fig).
Congresses where CP pairs cooperate, (i.e. appear above the threshold), namely 19491983,
have a wide cooperator value range and a narrow polarization score domain. These Congresses
fall in the 50% all CP pair appearance probabilities, (6.513.5% of the full range: 0.0213.5%)
but only in 20% of the polarization score range (0.430.57 of 0.431.09), indicating that these
30+ years would be hard to distinguish when defined by the polarization score index (S2A Fig).
The opposite is true for some Congresses between 19932011, which post probabilities of ap-
pearing above the threshold in a relatively narrow range between 0.02% and 2.0%, of the afore-
mentioned range, while the polarization score ranges liberally between 0.73 and 1.09, thus
demarcating these years with more political variability than the cooperator method. In essence,
the cooperator method presented here and the DW-NOMINATE polarization score is more
sensitive to later years, though the values correlate (r
: 0.73). Additionally, the DW-NOMI-
NATE method finds that Congresses commencing in 1951 and 1953 exhibit the least polariza-
tion (indexes. 435 and. 433, respectively), while the cooperator method shows that Congresses
commencing in 1973 and 1979 were the most cooperative, where each representative had a
13% chance of appearing above the threshold with a member of the opposite party.
A comparison between the DW-NOMINATEs overlap statistic exhibits a better correla-
tion with CP-pair probabilities of appearing above the roll call vote agreement threshold, i.e.
being cooperators (r
:. 83) (S2B Fig). Still, however, the cooperator method s 19952011 values
have a sizable range, while the overlap method produces values with few significant digits:
19951999 measured at 0.009, 2011 at 0.007 and 20032011 at 0.000, indicating less visible pre-
cision. These values are hard to differentiate over time, while the cooperator method assigns
more a diverse range of values to Congresses in this range (S2B Fig).
The comparison of the two DW-NOMINATE statistics with the newer cooperator statistics
does not indicate that either result is more correct. The cooperator method can add more di-
mension to the characterization of certain time frames, and the DW-NOMINATE statistics
produce more fidelity in other time frames. Yet, we believe that values produced by the cooper-
ator method are straightforward probabilities that are simple to explain with the following
question: What are the odds that any given representative will be a cooperator? This proba-
bility is simpler, but more transparent than DW-NOMINATE, which require knowledge of
feature space and component analysis to interpret these indexes. Instead, the cooperator meth-
od provides a quick overview that can be used across representative governments and other
voting-bodies worldwide. The DW-NOMINATE should be a complement to the cooperator
method, as it remains beneficial for examining multiple facets of each Congress. For example,
it provides multiple descriptive statistics whereas the cooperator method provides few.
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 9/14
Consensus and public opinion
Not surprisingly, partisanship correlates with failure to introduce and pass legislation. The
number of bills introduced (Fig 4A), bills passed ( Fig 4B), and the percentage of introduced
bills that pass (Fig 4C ) fall exponentially over time, in accordance with a fewer cooperator pairs
[30]. The number of bills introduced seems to be most negatively impacted by non-coopera-
tion. This trend is problematic as increase non-cooperation significantly correlates with a de-
crease in Congressional productivity (Fig 4). Moreover, a decrease in efficiency is also driven
by a significant decrease in the number of bills introduced [30], suggesting that increasing non-
cooperation stifles Congressional motivation to innovate. This gridlock has resulted in hyper-
partisanship and current popular criticism that Congress has recorded its least productive year
in 2013 [31].
Moreover, public opinion of Congress has declined simultaneously from 60% favorable rat-
ing in the 1960s to a 10% favorable rating today [30] also correlating with more bifurcation in
Congress over this time period. We discuss these points further below.
Our analysis shows that Congressional partisanship has been increasing exponentially for over
60 years, and has had negative effects on Congressional productivity. This is particularly appar-
ent in the steady reduction of the number of bills introduced onto the floor, suggesting that the
primary negative effect of increasing partisanship is a loss of Congressional innovation.
But why is this pattern of increasing partisanship emerging so strongly? There are complex
interactions that drive decision-making and pair-wise relationships in the House of Represen-
tatives. Though our data does not support a clear attribution of mechanism other than correla-
tional associations with covariates, we find that polarization is part of a long-term exponential
trend implying that non-cooperation multiplicatively breeds non-cooperation. In other words,
todays partisan atmosphere may not be a product of recent political splinter ing (such as the
Southern Democrats [32] or the Republican Tea Party Group). Alternatively, such groups may
have emerged from a growing shift from cooperation, while simultaneously contributing to the
shift. Therefore, while it is incorrect to say that recent divisive political figures are responsible
for increasing partisanship, they have actively contributed to it because these are the types of
non-cooperative figures and factions that the multiplicative system selects. The exponential in-
crease in non-cooperation shows no indication of slowing, or reversing, and so while Congress
has steadily become more non-cooperative over the latter half of the 20th century, this trend
seems likely to continue into the future.
This increase in non-cooperation leads to an interesting electoral paradox. While U.S. voters
have been selecting increasingly partisan representatives for 40 years, public opinion of the U.
S. Congress has been steadily declining. This decline [30] suggests that voters cast their ballots
on a local basis for increasingly partisan representatives whom they view as best representing
their increasingly partisan concerns, leaving few if any moderate legislators to connect parties
for a more cohesive Congress. Elected representatives are increasingly unable to cooperate at a
national Congressional level but are re-elected at least 90% of the time, reflecting an evasion of
collective responsibility. Voters might believe that highly partisan candidates will tip the scale
in one party s favor. However, based on correlations shown here, a partisan candidate may lack
cooperation needed to pass legislation. More moderate legislators may have a competitive ad-
vantage in negotiating for their partys legislation.
A fundamental reversal of increasing non-cooperation, over time, might require either a
change in local ideological perspectives (resulting in a selective shift to fewer partisan represen-
tatives), or a fundamental change in how the electorate votes (from concerns focused on party
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 10 / 14
Fig 4. Congressional productivity as a function of cooperation rates. Three plots of Congressional
productivity as a function of congressional cooperation show a correlation with: (a) The number of bills
introduced during a session. (b) The number of bills passed. (c) The ratio of bills passed to those introduced.
Solid lines indicate exponential fits. Data from [30].
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 11 / 14
issues to concerns focused on global effectiveness). Certainly current affairs do not seem to di-
vide potential cooperators, as cross-party relationships peaked in arguably the most tumultu-
ous period in recent U.S. history, marked with numerous political assassinations and Vietnam
War and the resignation of President Nixon, as illustrated by others, such as [2325]. It may be
that decreased Congressional social interaction in Washington, D.C., combined with increased
telecommunications and commuting to ones home district, may hamper representatives
ability cooperate.
The United States is comprised of 435 unique Congressional districts, each with distinct
physical geographies, economics, communities, cultures and political ideologies. At one time,
these unique constituencies seemed to be represented by a distinct combination of ideologies
from the Democratic and Republican Party. Formerly, legislators exhibited a mixture of ideals
that resonated across party platforms, allowing each to forge a personal voting fingerprint that
reflected the distinctive perspective of his or her unique district and constituency.
Today, districts may remain as socio-economically and geographically unique as in the past,
yet representatives have all but lost their personal voting records to complement their individu-
alized constituencies. Instead, Americans today are represented by political figures whose ideo-
logical roll call voting record in the U.S. House of Representatives generally resembles one of
only two types: either a Republ ican or a Democrat platform, with very little combination. What
this unprecedented hyper partisanship will yield for the future of United States foreign and do-
mestic policy is yet to be seen. This work was primarily performed at the Santa Fe Institute.
Supporting Information
S1 Fig. CP vote disagreements divided by all roll call votes, over time. This figure normalizes
the number of cross-party pair vote disagreements over the total possible votes in the particular
S2 Fig. Cooperator statistics compared with traditional descriptive statistics. Per Congress,
the probability that a legislator is in a CP pair above the threshold (i.e. a cooperator) correlates
with two DW-NOMINATE statistics: political partisanship and overlap, with different dynam-
ics over time. Data from [24].
S1 Database.
S1 Table. Super-cooperators in cooperator pairs, ordered by percentage of appearances.
The data reported in this paper are from Office of the Clerk of the U.S. House of Representa-
tives, via Thomas Online Library of Congress as accessed through Govtrak. We thank the John
Templeton Foundation, MIT Senseable City Lab, National Defense Science and Engineering
Graduate (NDSEG) Fellowship, and the Rockefeller Institute. Authors acknowledge our four
helpful reviewers and two academic editors, especially Dr. Rodrigo Huerta-Quintanilla. We
would like to thank Thomas Ding and Wei Luo for assistance with data collection; Brian King,
Luis M.A. Bettencourt and Chris Wood for discussions; and Nathan Frey, Leon Andris, and
Deryck Holdsworth for advisory assistance with content editing.
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 12 / 14
Author Contributions
Conceived and designed the experiments: CA MJH. Performed the experiments: CA CEG
MJH. Analyzed the data: CA DL CEG MJH MM JAS. Contributed reagents/ma terials/analysis
tools: DL CA. Wrote the paper: JAS CA DL MJH. Designed Interactive Website: MM.
1. Snowe O. The effect of modern partisanship on legislative effectiveness in the 112th Congress. Harv J
on Legis. 2013; 50: 2140.
2. McCarty N, Poole KT, Rosenthal H. Polarized America: The dance of ideology and unequal riches.
Cambridge, MA: MIT Press; 2006.
3. Carson J, Crespin M, Finocchiaro C, Rohde D. Redistricting and party polarization in the U.S. House of
Representatives. Am Polit Res. 2007; 35: 878904.
4. Rosenstone SJ, Hansen JM. Mobilization, participation, and democracy in America. New York: Mac-
millan; 1993.
5. Theriault S. Party polarization in Congress. New York: Cambridge University; 2008.
6. Roberts J, Smith S. Procedural contexts, party strategy and conditional party voting in the U.S. House
of Representatives 19712000. Am J Pol Sci. 2003; 47: 305317.
7. Jenkins J. Examining the bonding effects of party: A comparative analysis of roll-call voting in the U.S.
and Confederate Houses. Am J Pol Sci. 1999; 43: 11441165.
8. Theriault S. Party polarization in the US Congress: Member replacement and member adaptation.
Party Pol. 2006; 12: 483503.
9. Iyengar S, Hahn KS. Red media blue media: Evidence of ideological selectivity in media use. J Com-
mun. 2009; 59: 1939.
10. Snyder J, Groseclose T. Estimating party influence in Congressional roll-call voting. Am J Pol Sci.
2000; 44: 193211.
11. Cohen GL. Party over policy: The dominating impact of group influence on political beliefs. J Pers Soc
Psychol. 2003; 85: 808822. PMID: 14599246
12. Nokken T. Dynamics of Congressional loyalty: party defection and roll call behavior 19471997. Legis
Stud Quart. 2000; 25: 417444.
13. Fleisher R, Bond J. The shrinking middle in the U.S. Congress. Brit J Pol Sci. 2004; 34: 429451.
14. Lee FE. Geographic politics in the U.S. House of Representatives: Coalition building and distribution of
benefits. Am J Pol Sci. 2003; 47: 714728.
15. Carrubba C, Volden C. Coalitional politics and logrolling in legislative institutions. Am J Pol Sci. 2000;
44: 261267.
16. Porter M, Muchab P, Newman M, Warmbrand C. A network analysis of committees in the U.S. House
of Representatives. Proc Nat Acad Sci USA. 2005; 102: 70577062. PMID: 15897470
17. Fowler J. Connecting the Congress: a study of co-sponsorship networks. Pol Anal. 2006; 14: 456487.
18. Zhang Y, Friend A, Traud A, Porter M, Fowler J, Mucha P. Community structure in Congressional co-
sponsorship networks. Physica A. 2008; 387: 17051712.
19. Cho W, Fowler J. Legislative success in a small world: Social network analysis and the dynamics of
Congressional legislation. J Pol. 2010; 72: 124135.
20. Porter M, Mucha P, Newman M, Friend A. Community structure in the United States House of Repre-
sentatives. Physica A. 2007; 386: 414438.
21. Waugh AS, Pei L, Fowler JH, Mucha P, Porter MA. Party polarization in Congress: A network science
approach; 2011. Preprint. Available:
. Accessed July 15 2012.
22. Poole KT, Rosenthal H. The polarization of American politics. J Pol. 1984; 24: 10611079.
23. Cox G, Poole KT. On measuring partisanship in roll call voting: the U.S. House of Representatives
18771999. Am J Pol Sci. 2002; 46: 477489.
24. Poole KT, Rosenthal H. Congress: A political-economic history of roll call voting. Oxford: Oxford Uni-
versity Press; 1997. See also:
25. Carroll R, Lewis JB, Lo J, Poole KT, Rosenthal H. Measuring bias and uncertainty in DW-NOMINATE
ideal point estimates via the parametric bootstrap. Polit Anal. 2009; 17: 261278.
26. Office of the Clerk of the U.S. House of Representatives, Roll Call Votes; 2014. Accessed through Gov-
trak by Civic Impulse, LLC. Available:
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 13 / 14
27. Clinton J, Jackman S, Rivers D. The statistical analysis of roll call data. Am Pol Sci Rev. 2004; 98:
28. Cox G, McCubbins M. Setting the agenda: Responsible party government in the U.S. House of Repre-
sentatives. Cambridge, UK: Cambridge University Press; 2005.
29. Harbridge L, Malhotra N. Electoral incentives and partisan conflict in Congress: Evidence from survey
experiments. Am J Pol Sci. 2011; 55: 494510.
30. Ornstein N, Mann T, Malbin M, Rugg A. Vital Statistics on Congress. The Brookings Institution. July
2013. Available:
ornstein. Accessed: 15 November 2013.
31. Viser, M. This Congress Going Down as Least Productive: Hyperpartisan Climate Gums Up Bulk of
Laws. The Boston Globe. December 2013. Available:
12/04/congress-course-make-history-least-productive/kGAVEBskUeqCB0htOUG9GI/story.html. Ac-
cessed 10 December 2013.
32. Theriault S, Rohde D. The Gingrich Senators and party polarization in the U.S. Senate. J Pol. 2011; 73:
33. Barnes J, Hut P. SA hierarchical O(N log N) force calculation algorithm. Nature. 1986; 324: 4.
34. Krackhardt D, Stern RN. Informal networks and organizational crises: An experimental simulation. Soc
Psychol Q. 1988; 15: 123140
The Rise of Partisanship in the U.S. House of Representatives
PLOS ONE | DOI:10.1371/journal.pone.0123507 April 21, 2015 14 / 14
... These challenges have led social network researchers to seek alternate, indirect measurement methods. One common alternative involves inferring an unobserved social network from observed groups such as club membership, event participation, or simply spatial proximity (e.g., Breiger, 1974;Newman, 2004;Mizruchi, 1996;Andris et al., 2015;Daniel et al., 2019). However, little is known about the circumstances under which a social network inferred from observed groups accurately captures the unobserved social network of interest. ...
... . are routinely inferred from lawmakers observed memberships in voting blocs (e.g., Andris et al., 2015), and unobserved social networks are routinely inferred from young childrens' play groups (e.g., Daniel et al., 2019). ...
Full-text available
Collecting social network data directly from network members can be challenging. One alternative involves inferring a social network from individuals' memberships in observed groups, such as teams or clubs. Through a series of simulations, I explore when we can expect such inferences to be accurate. I find that an unobserved network can be inferred with high accuracy under a range of circumstances. In particular, I find that social networks inferred from observed groups are more accurate when (1) the unobserved network has a small world structure, (2) the groups are generated by a shuffling or agglomerative process, (3) a large number of groups are observed, and (4) the observed groups' compositions are tightly coupled to the unobserved network's structure. These findings offer guidance for researchers seeking to indirectly measure a social network of interest through observations of groups.
... While these technological advancements promised faster and wider access to information, their influence on the spread of information has turned out to be more nuanced. Indeed, they have also fostered several pervasive issues, such as polarization, misinformation, and the emergence of echo chambers that could influence public opinion and negatively impact society [1,32,21]. While these divergences could already be observed during the 20th century [32], the introduction of social media networks may have increased the ideological divide among opposite factions [22]. ...
... While these divergences could already be observed during the 20th century [32], the introduction of social media networks may have increased the ideological divide among opposite factions [22]. This radicalization in opinions has been shown to be a clear obstacle to dialogue, consensus, and policy-making [1,32], being also considered as "harmful to democracy and society" [34] and a security risk for the UN [37]. Polarized debate is also a fertile environment for the spread of misinformation that may harm society at different levels [12]. ...
Full-text available
The emergence of new public forums in the shape of online social media has introduced unprecedented challenges to public discourse, including polarization, misinformation, and the emergence of echo chambers. While existing research has extensively studied the behavior of active users within echo chambers, little attention has been given to the hidden audience, also known as lurkers, who passively consume content without actively engaging. This study aims to estimate the share of the hidden audience and investigate their interplay with the echo chamber effect. Using Twitter as a case study, we analyze a polarized political debate to understand the engagement patterns and factors influencing the hidden audience's presence. Our findings reveal a relevant fraction of users that consume content without active interaction, which underscores the importance of considering their presence in online debates. Notably, our results indicate that the engagement of the hidden audience is primarily influenced by factors such as the reliability of media sources mentioned in tweets rather than the ideological stance of the user that produced the content. These findings highlight the need for a comprehensive understanding of the hidden audience's role in online debates and how they may influence public opinion.
... Barrat et al. [4] extended the notion of node degree and called it the strength of a node, which was defined by the sum of the weights of all edges incident in it. Equation (2) presents the calculation of the strength of a node u. ...
... A multilayer representation of temporal networks is employed, and a multilayer modularity maximization is used to detect communities in these networks. How much the legislators tend to form ideological relationships with members of the opposite party is measured by Andris et al. [2]. The authors quantify the level of cooperation or lack thereof between Democrat and Republican party members in the U.S. House from 1949 to 2012. ...
Full-text available
This work proposes a methodology for a sounder assessment of centrality, some of the most important concepts of network science, in the context of voting networks, which can be established in various situations from politics to online surveys. In this regard, the network nodes can represent members of a parliament, and each edge weight aims to be a probabilistic proxy for the alignment between the edge endpoints. In order to achieve such a goal, different methods to quantify the agreement between peers based on their voting records were carefully considered and compared from a theoretical as well as an experimental point of view. The results confirm the usefulness of the ideas herein presented, and which are flexible enough to be employed in any scenario which can be characterized by the probabilistic agreement of its components.KeywordsNode CentralityWeighted NetworksData SummarizationSocial ComputingVoting Networks
... While happening around the world, one country these effects appear to be particularly pronounced is the United States. The left-right divide has increased on the governmental level (Andris et al. 2015) but also in everyday life, affecting where Americans choose to live (Mummolo and Nall 2017;Brown and Enos 2021), how they raise their children (Tyler and Iyengar 2022), how they deal with misinformation (Petersen et al. 2023;González-Bailón et al. 2023), and which daily cultural and material products they consume (Hetherington and Weiler 2018;Rawlings and Childress 2022). In the information space, besides the growing divergence of news media (Jurkowitz et al. 2020;Broockman and Kalla 2022;Muise et al. 2022), polarization and segregation effects have been observed in diverging public narratives about society and significant events (Li et al. 2017;Demszky et al. 2019), online knowledge curation (Yang and Colavizza 2022), as well as behavior on social media (Adamic and Glance 2005;Rathje et al. 2021;Mukerjee et al. 2022;Rasmussen et al. 2022). ...
Full-text available
Language change is influenced by many factors, but often starts from synchronic variation, where multiple linguistic patterns or forms coexist, or where different speech communities use language in increasingly different ways. Besides regional or economic reasons, communities may form and segregate based on political alignment. The latter, referred to as political polarization, is of growing societal concern across the world. Here we map and quantify linguistic divergence across the partisan left-right divide in the United States, using social media data. We develop a general methodology to delineate (social) media users by their political preference, based on which (potentially biased) news media accounts they do and do not follow on a given platform. Our data consists of 1.5M short posts by 10k users (about 20M words) from the social media platform Twitter (now "X"). Delineating this sample involved mining the platform for the lists of followers (n=422M) of 72 large news media accounts. We quantify divergence in topics of conversation and word frequencies, messaging sentiment, and lexical semantics of words and emoji. We find signs of linguistic divergence across all these aspects, especially in topics and themes of conversation, in line with previous research. While US American English remains largely intelligible within its large speech community, our findings point at areas where miscommunication may eventually arise given ongoing polarization and therefore potential linguistic divergence. Our methodology - combining data mining, lexicostatistics, machine learning, large language models and a systematic human annotation approach - is largely language and platform agnostic. In other words, while we focus here on US political divides and US English, the same approach is applicable to other countries, languages, and social media platforms.
... Ideological polarization has been widely documented in the United States. Data on members of Congress show an increasing divide of the political elites along party lines (e.g., Andris et al., 2015). At the level of the public as well, polls show that supporters of the Democratic and Republican parties have grown increasingly distant in their ideological positions since the 1970s; this trend has accelerated since the beginning of the century (e.g., Webster & Abramowitz, 2017). ...
There is mounting evidence in the United States and worldwide that highlights a widespread and deepening “principled dislike” between partisan groups. Stemming from group identity dynamics, such as “affective polarization,” it is likely to be triggered by exposure to intra-elite conflicts, such as campaign negativity and incivility. However, empirical evidence for this effect is scarce, and it rests only on survey data; causal evidence linking campaign attacks and affective polarization is still missing. In this article, we advance the hypothesis that the effects of exposure to mediatized political attacks are likely mediated by how negative such attacks are perceived. To test our expectations, we leverage new evidence from an online experiment with convenience sample of American voters (N = 1,081). Our results suggest that exposure to intra- elites’ political attacks can drive affective polarization, but this unfolds mostly as a function of perceived negativity of those messages, and only for respondents that are ideologically affiliated with the target of the attack. Negativity is in the eye of the beholder, especially when one is being attacked.
The division of a social group into subgroups with opposing opinions, which we refer to as opinion disparity, is a prevalent phenomenon in society. This phenomenon has been modeled by including mechanisms such as opinion homophily, bounded confidence interactions, and social reinforcement mechanisms. In this paper, we study a complementary mechanism for the formation of opinion disparity based on higher-order interactions, i.e., simultaneous interactions between multiple agents. We present an extension of the planted partition model for uniform hypergraphs as a simple model of community structure, and we consider the hypergraph Susceptible-Infected-Susceptible (SIS) model on a hypergraph with two communities where the binary ideology can spread via links (pairwise interactions) and triangles (three-way interactions). We approximate this contagion process with a mean-field model and find that for strong enough community structure, the two communities can hold very different average opinions. We determine the regimes of structural and infectious parameters for which this opinion disparity can exist, and we find that the existence of these disparities is much more sensitive to the triangle community structure than to the link community structure. We show that the existence and type of opinion disparities are extremely sensitive to differences in the sizes of the two communities.
Signed networks and balance theory provide a natural setting for real-world scenarios that show polarization dynamics, positive/negative relationships and political partisanship. For example, they have been proven effective in studying the increasing polarization of the votes in the two chambers of the U.S. Congress from World War II on Andris, Lee, Hamilton, Martino, Gunning & Selden (2015, PLoS ONE, 10, 1–14) and Aref & Neal (2020, Sci. Rep., 10, 1–10). To provide further insights into this particular case study, we propose the application of a pipeline to analyze and visualize a signed graphs configuration based on the exploitation of the corresponding Laplacian matrix spectral properties. The overall methodology is comparable with others based on the frustration index, but it has at least two main advantages: first, it requires a much lower computational cost and second, it allows for a quantitative and visual assessment of how arbitrarily small subgraphs (even single nodes) contribute to the overall balance (or unbalance) of the network. The proposed pipeline allows the exploration of polarization dynamics shown by the U.S. Congress from 1945 to 2020 at different resolution scales. In fact, we are able to spot and point out the influence of some (groups of) congressmen in the overall balance, as well as to observe and explore polarizations evolution of both chambers across the years.
Historical records from democratic processes and negotiation of constitutional texts are a complex type of data to navigate due to the many different elements that are constantly interacting with one another: people, timelines, different proposed documents, changes to such documents, and voting to approve or reject those changes. In particular, voting records can offer various insights about relationships between people of note in that historical context, such as alliances that can form and dissolve over time and people with unusual behavior. In this paper, we present a toolset developed to aid users in exploring relationships in voting records from a particular domain of constitutional conventions. The toolset consists of two elements: a dataset visualizer, which shows the entire timeline of a convention and allows users to investigate relationships at different moments in time via dimensionality reduction, and a person visualizer, which shows details of a given person's activity in that convention to aid in understanding the behavior observed in the dataset visualizer. We discuss our design choices and how each tool in those elements works towards our goals, and how they were perceived in an evaluation conducted with domain experts.
In a time of declining support for democracy and intensifying rivalry between democracies and autocracies, understanding how nondemocratic nations portray U.S. elections is vital. And yet, despite the enormous attention U.S. presidential elections attract around the world, the manner in which international media makes sense of U.S. campaigns remains unclear, with only a limited number of comparative studies conducted and even fewer looking at non-Western, nondemocratic nations. Furthermore, current comparative frameworks remain biased toward Western conceptualizations of media and their role in democratic countries, with nondemocratic or transitional democracies used to support theoretical models developed elsewhere. Thus, this study offers strategic media narratives as an alternative means to understand transnational similarities and differences in election reporting emerging from four non-Western, nondemocratic nations by comparing their coverage of the 2020 and 2016 U.S. presidential campaigns. Results show substantial shifts in the nature of coverage, albeit with some similarities between campaigns. Trump remained negatively discussed in both elections, but with reporting in 2020 associating his leadership character to his policies. Whereas Clinton was negatively covered in 2016, Biden was neutrally discussed in 2020 with focus on his character and policies drawn in contrast to Trump. Both political parties were negatively covered, with Chinese, Russian, and Iranian narratives associating Republicans and Democrats as pursing confrontational relations with each nation. Most importantly, discussions of U.S. democracy were substantially more frequent and negative in 2020 compared to 2016. Taken together, the study contributes theoretically and empirically to the study of comparative election research by theorizing the role of narratives in international campaign coverage, addressing the gap in research into nondemocratic media reporting of international elections, and provides one of the few cross-time comparisons enabling insight into the drivers of how and why coverage of U.S. elections change.
Full-text available
This book examines the shape, composition, and practices of the United States political media landscape. It explores the roots of the current epistemic crisis in political communication with a focus on the remarkable 2016 U.S. president election culminating in the victory of Donald Trump and the first year of his presidency. The authors present a detailed map of the American political media landscape based on the analysis of millions of stories and social media posts, revealing a highly polarized and asymmetric media ecosystem. Detailed case studies track the emergence and propagation of disinformation in the American public sphere that took advantage of structural weaknesses in the media institutions across the political spectrum. This book describes how the conservative faction led by Steve Bannon and funded by Robert Mercer was able to inject opposition research into the mainstream media agenda that left an unsubstantiated but indelible stain of corruption on the Clinton campaign. The authors also document how Fox News deflects negative coverage of President Trump and has promoted a series of exaggerated and fabricated counter narratives to defend the president against the damaging news coming out of the Mueller investigation. Based on an analysis of the actors that sought to influence political public discourse, this book argues that the current problems of media and democracy are not the result of Russian interference, behavioral microtargeting and algorithms on social media, political clickbait, hackers, sockpuppets, or trolls, but of asymmetric media structures decades in the making. The crisis is political, not technological.
This article argues that scholars need to consider the structure of House representation to better understand distributive politics. Because House districts (unlike states) are not administrative units in the federal system, House members cannot effectively claim credit for most grant-in-aid funds. Instead, their best credit-claiming opportunities lie in earmarked projects, a small fraction of federal grant dollars. As a consequence, I expect to find: (1) political factors have a much greater effect on the distribution of earmarked projects than on federal funds generally; and (2) project grants are a better support-building tool for coalition leaders than allocations to states. I test this argument with a study of the 1998 reauthorization of surface transportation programs and find strong support for both hypotheses.
In this article, we provide a critical review of the evidence and arguments about party polarization in the House of Representatives during the late 20th century. We show that inferences about party polarization are significantly affected by voting reform in the early 1970s. We observe that a decomposed roll-call record alters our view of the timing of changes in party polarization and therefore requires that we reconsider explanations of the trend. We revisit explanations of party polarization and establish a strong case for placing substantial emphasis on party strategies in explanations of party polarization in floor behavior during the 1980s and 1990s.
This article examines the recent phenomenon of extreme partisanship in the United States Senate. Throughout history, the Senate has been a legislative body dedicated to debating and resolving the nation's most pressing issues. However, in recent years, paralyzing partisanship in Washington has severely impeded the Senate's work. Several Senate procedures, including the rules surrounding filibusters, cloture, and filling the amendment tree, have exacerbated this problem. In this article, Senator Snowe describes the effects of extreme partisanship on the Senate and offers her thoughts about how future Congresses should avoid such setbacks going forward.
Scholars of the U.S. House disagree over the importance of political parties in organizing the legislative process. On the one hand, non-partisan theories stress how congressional organization serves members’ non-partisan goals. On the other hand, partisan theories argue that the House is organized to serve the collective interests of the majority party. This book advances our partisan theory and presents a series of empirical tests of that theory’s predictions (pitted against others). It considers why procedural cartels form, arguing that agenda power is naturally subject to cartelization in busy legislatures. It argues that the majority party has cartelized agenda power in the U.S. House since the adoption of Reed's rules in 1890. The evidence demonstrates that the majority party seizes agenda control at nearly every stage of the legislative process in order to prevent bills that the party dislikes from reaching the floor.
We propose a method of assessing party influence, based on a spatial model, Our method provides the first test of whether observed values of the widely-used Rice index of party dissimilarity are consistent with a "partyless" null model. It also avoids problems that beset previous estimators. Substantively, we find evidence of influence in all but one Congress since 1877. Moreover, our indicator of party pressure is systematically higher for the sorts of roll calls that party theorists believe are more pressured-procedural, organizational, and label-defining votes. Our results refute the widespread notion that parties in the House have typically had negligible influence on roll-call voting behavior. They also document important changes In party influence associated with the packing of the Rules Committee in 1961 and the procedural reforms of 1973.
This paper examines whether political parties influence Congressional roll-call voting. Rather than focusing on contemporary evidence, my approach is historical: analyzing voting behavior in the U.S. and Confederate Houses during the Civil War. The U.S. and Confederate cases provide a unique opportunity for a comparative analysis because the two legislative systems were nearly identical in all facets, except that a strong two-party system was in place in the U.S. while a party system did not exist in the Confederacy. Thus, using vote-scaling techniques developed by Poole and Rosenthal (1985, 1991, 1997), I examine how roll-call voting in a party system (the U.S. House) differs from roll-call voting in a similar nonparty system (the Confederate House). My results indicate that voting in the U.S. House was considerably more predictable than voting in the Confederate House. Moreover, from additional tests, I conclude that these voting differences were due not to differences in the structure of preferences, but rather to the existence (or nonexistence) of political parties. In the U.S. House, party had a significant, independent effect on vote choice, after controlling for members' personal preferences. No such effect existed in the party-less Confederate House.
A novel method is described of directly calculating the force on N in the gravitational N-body problem that grows only as N log N. The technique uses a tree-structured hierarchical subdivision of space into cubic cells, each of which is recursively divided into eight subcells whenever more than one particle is found to occupy the same cell. This tree is constructed anew at every time step, avoiding ambiguity and tangling. Advantages over potential-solving codes include accurate local interactions, freedom from geometrical assumptions and restrictions, and applicability to a wide class of systems, including planetary, stellar, galactic, and cosmological ones. Advantages over previous hierarchical tree-codes include simplicity and the possibility of rigorous analysis of error.
DW-NOMINATE scores for the U.S. Congress are widely used measures of legislators' ideological locations over time. These scores have been used in a large number of studies in political science and closely related fields. In this paper, we extend the work of Lewis and Poole (2004) on the parametric bootstrap to DW-NOMINATE and obtain standard errors for the legislator ideal points. These standard errors are in the range of 1%-4% of the range of DW-NOMINATE coordinates. © The Author 2009. Published by Oxford University Press on behalf of the Society for Political Methodology. All rights reserved.