Sie sind auf Seite 1von 30


US Government Accountability Office
William Egar<> is a Social Science Analyst at
the US Government Accountability Office, 441 G St. NW, Washington DC, 20314

Congress:The House Minority
Party and the Constructionof the
Ro/I-Call Record
Existing research on congressional panies tends to focus almost exclusively on
the majority party. I argue that the inattention to the House minority party hampers our
unden,ianding of the construction of the roll-call record and, consequently, our
unden,ianding of the sources of polarization in congressional voting. Employing an
original data set of House members' reque;is for recorded votes between 1995 and
20 I 0, I demonstrate that votes demanded by the minority party are disproportionately
divisive and partisan and make Congress appear considerably more polarized based on
commonly used measures. Moreover, minority-requested votes make vulnerable
members of the majority appear more panisan and ideologically extreme.

Estimatesof legislator ideology in the US Congress suggest that

the congressional parties are polarized today to a degree possibly not
seen before, at least since the late nineteenth century.Ideological polar-
izationin Congressdoes not seem to be an artifactof any singlemeasure
of legislatorpreferences(Theriault 2008). Factors such as the ideological
sortingof liberalvoters into the Democratic Party and conservativesinto
the Republican Party (Fiorina, Abrams, and Pope 2011; Levendusky
2009;Rohde 1991),geographicsorting of voters into politicallyhomog-
enous areas (Oppenheimer2005; Theriault 2008), redistricting(Carson
et al. 2007; but see McCarty,Poole, and Rosenthal2009), primaryelec-
tions (Brady, Han, and Pope 2007), gro,ving income inequality
(McCarty,Poole,and Rosenthal2006),and the ideologicalpolarizationof
voters themselves-or at a minimum,party activists(e.g., Abramowitz
201O;Fiorinaet al. 2010)--have all been offeredas explanationsfor why
thispolarizationhas occurred.
Most of the above-listedexplanationsplace their focus on changes
in the electorate.Fewer studies explore the ,vithin-chambersources of
congressional polarization (for exceptions,see Harbridge 2015; Jessee


DOI: !O.l l l l/lsq.l2135
©2016 Washington University in St. Louis
and Theriault 2014; Lee 2009; Roberts and Smith 2003; Theriault
2013). The bulk of research on congressionalpolarizationrelies on esti-
mates of legislator ideology, estimatesthat are based on recorded votes.
Yet not all issues in Congress receive a vote, and those that do are deter-
mined (nonrandornly) by legislators themselves. Rather than being a
random sample of issues, then, members might request a roll-<:allvote in
order to engage in credit claiming (Mayhew 1974), to be certain of the
vote margin on a particularly contentiousissue, or to place opponentson
the record in a politically difficult situation (Roberts and Smith 2003;
Smith 1989), among other reasons.
In the following,I examine how the House minority party uses the
roll-call record to affect the image of its opponents as a matter of elec-
toral strategy. Specifically,I argue that for a variety of reasons, typical
measuresof congressionalpolarizationand measures of partisan conflict
will be inflated as a consequence of minority-partyelectoral incentives.
Using an original data set of House members' requests for recorded
votes, I provide evidence that the minority party strategically selects
divisive issues for recorded votes in order to negatively affect the image
of members of the majorityparty, ,vhich simultaneouslyhas the effect of
making Congress as a whole appear more partisan and ideologically
polarized. Gaining a better understandingof the data-generatingprocess
behind roll-callvoting by examiningrequestsfor recordedvotes can pro-
vide some insight with regard to fluctuations in congressional
polarizationand party conflictover time and can help our understanding
of the House minority party, which is often ignored in scholarlyresearch
on congressionalparties.

Existing Literature: Composition of the Roll-Call Record

Analysis of recorded voting is central to the study of the US

Congress. For instance, estimates of legislator ideal points such as
NOMINATE (Poole and Rosenthal 1997) are used for a wide range of
purposes in legislative studies and are based on recorded votes. Yet a
feature of congressionallife that is often underappreciatedis that not all
legislative items-and indeed, by some measures, relatively few
items-are disposed of by a recordedvote. Much of Congress's work is
completedoff the record. For instance,Clinton and Lapinski (2008) find
that only 11.9% of bills signed into law receive a recorded vote in the
House, with an even smaller percentage (7.9%) receiving a recorded
vote in the Senate; moreover, the bills that do receive recorded votes
tend to be measures of importanceand are disproportionately in certain
issue areas and not others. Clinton and Lapinski suggest that a focus on
recorded voting might serve to exacerbatethe perceptionof congres-
sional conflict.A hypotheticalCongressthat unanimouslypasses 1,000
bills by voice vote, but passes a single bill by a recorded,party-linevote,
would appear highly partisan if one focused on the
roll-call record alone. In other words, fe,v successful measuresactually
receive a recordedvote, and the ones that do are not simply a random
sampleof all issuesthat Congressconsiders.
Asidefrom the fact that not all measuresreceive a recordedvote, a
number of factorscombineto muddy the waterswhen it comes to inter-
preting recordedvotes that do occur. As alreadymentioned,ideal point
estimatesare centralto legislative studies.These estimates(e.g., Clinton,
Jackman,and Rivers 2004; Poole and Rosenthal 1997)are based on an
underlying spatial model of voting that assumes each legislatorvotes
basedon the utilityhe or she wouldreceive from the alternatives offered.
Yet considerableevidence exists that patterns in recorded voting are
affectedby much more than policy preferences.For example,by exam-
ining the within-sessionchange in majoritystatus in the l 07th Senate,
Roberts(2007) demonstratesthat estimatesof the ideologicalrank order
of senators changed, in some cases significantly,following Jim Jef-
fords's defection from the Republican Party. Roberts (2007) explains
that the significantideologicalshift of Tom Daschle (D-SD) and Trent
Lott (R-MS) following the change in majority status can be almost
entirelyexplainedby a unique proceduraldetailof the Senate- majority
leadersfrequentlyvote againstcloturemotionsthat they supportin order
to be able to move a reconsiderationof the vote later on. In other ,vords,
the change in thesemembers' ideal point estimateshad littleto do with a
changein their policyviews. Thoughestimatessuch as NOMINATEare
frequently treated by scholarsas pure estimatesof ideology,Roberts's
work calls attentionto the fact that such measuresare based on roll-call
data, which may be sensitive to factors such as majority status and
Related to the above discussionis partisanpolarization. Although
roll-callmeasuresshow that the congressionalpartiesare more internally
homogenous and distinct from one another than they ,vere 40 years
ago,2 careful attentionto how the record is constructedcan alter infer-
ences about the degree,the timing,or other featuresof how the process
of polarizationhas worked.Notably, the House in the 1970sfor the first
time introducedrecordedvoting in the Committeeof the Whole (Smith
1989).As Robertsand Smith(2003)convincinglydemonstrate,votes on
amendmentsin the Committeeof the Whole tend to be far more partisan
and divisiv e than other types of House votes. The introduction of
recordedvoting on COW amendments, alongwith the ease of electronic
voting, quicklyled to an explosionof amendmentvotes, driven largely
by minority-partyRepublicanswho sought to force members of the
majorityparty to take difficultor embarrassingvotes, ,vhich added a
whole new and divisiveset of votes to the roll-callrecord. As Smith
explains, "[minority Republicans]actively sought ways to challenge
committee products, raise ideologically charged issues, and force
recordedvotes. [Robert]Bauman[R-MD]even entertainedrequestsfor
recorded-voteamendmentsfrom Republicanchallengersto Democratic
incumbentsin order to compelDemocratsto take politicallydangerous
positions"(1989,34). Robertsand Smith (2003)show that partisanleg-
islative strategies,combinedwith adoption of recorded voting in the
Committeeof the Whole, significantlyaffect conclusionsregardingthe
timingand nature of polarizationin the House, suggestinga need to be
sensitiveto how the roll-callrecord is constructedand its implications
for measuresof partisanconflictand polarization.
Scholars have recently raised questions about ho,v much
observed polarization in Congress is due to ,videning differences on
policy substance and how much results from greater party unity on
procedural matters. Theriault (2008) argues that the bulk of contem-
porary polarization is the result of procedural voting. Procedural
votes are thought of by members differently than votes on substan-
tive issues; believed to be more obscure to the public than final
passage or other substantive votes, legislators appear to be more
loyal to their parties when voting on procedural matters (Lee 2009;
Theriault 2008; but see Smith, Ostrander, and Pope 2013). Ideal
point estimates differ considerablywhen based on procedural versus
nonprocedural votes (Jessee and Theriault 2014). The authors sug-
gest that this type of voting behavior tends to inflate measures of
polarization without actually indicating a greater degree of ideologi-
cal distance between the parties. Similarly, Lee (2009) demonstrates
that collective partisan concerns produce a greater level of partisan
"team play," especiallywhen control of national political institutions
is at stake. Fights over policy ultimately become fights over political
power; denying political victories to partisan opponents and
opposing-partypresidents can be electorallybeneficial, incentivizing
heightened partisan conflict in an electorally competitive environ-
ment. Accordingly, Lee (2009) finds that a considerable amount of
polarization and party conflict in the modem Senate occurs when the
Senate takes recorded votes on issues with little to no identifiable
"ideological" content. These previously mentioned studies suggest
the need to be sensitive to nonideological factors that produce
partisan division in roll-call voting.
House Minority-Party Legislative Strategy and its Effect on the
Roll-Call Record

Evidencestronglysuggeststhat the roll-callrecord,more than just

a simple register of legislators' preferenceson policy, also reflects ele-
ments of partisan"team play'' and nonideological fighting(Lee 2009)
and can be used by membersin ways to affectthe appearanceof partisan
opponents(Robertsand Smith2003; Smith 1989). Recognizingthe fact
that the congressionalroll-callrecordis not exogenousbut is insteadpur-
posefully constructed by legislatorsthemselves calls attention to how
members can use the record to affect the public's perceptionof them-
selves,other members,the parties,and Congressas a whole. Exploring
this questioncan help provide insight on why members' voting records
indicatesuch a degreeof polarizationoverthe lastfew decades.
Much of the literatureon congressionalparties deals with how
majority-partyleadership in the House seeks to improve or maintain
their own party's image (Cox and McCubbins 1993, 2005). Far less
attentionhas been dedicatedto the minorityparty- we know little about
how the minorityattemptsto improveits own image or ho,v it attempts
to tarnishthe image of the majority party. Indeed,little has been written
about congressionalminorityparties since Jones (1970),who examines
how politicalconditions(partisancontrol of the presidency,party mar-
gins in Congress,among other factors)allowfor differentminority-party
policy-makingstrategies. More recently, Green (2015) explores how
membersof the minority party work to achievetheir collectiveconcerns
(majority status, policymaking, procedural rights, and success of
Given the fact that the minority party is frequentlyshut out of
decisionmaking in the House due to the majority's controlof the legisla-
tive agenda, and the fact that the main proceduraltools of the House
minority to obstruct and delay proceedingswere taken a,vay with the
passage of Reed's Rules in the late nineteenthcentury,few tools remain
to the minorityparty to remain involved in the legislativeprocess when
opposedby a determinedmajority.One of the few inalienableprocedural
rights left to the minorityis to demandthe yeas and nays on nearly any
matter(subjectto secondingby one-fifthof the chamberin most cases),
a right explicitlymentionedin the Constitution(Article I, Sec. 5). If the
House minorityparty, lacking obstructivetools like the filibusterin the
Senate and often prohibitedfrom offeringamendments,has little influ-
ence in the legislativeprocess, then affectingthe public's perceptionof
the majorityparty in order to affect an electoralchange and becoming
the majorityshould be the minority'sultimategoal. The roll-callrecord,
which the Houseminority doeshave the abilityto affect,is a prime target
for the minority's effortto changepublic perceptions.
Considerableevidence exists that the roll-call record influences
perceptionsof individual members and of Congress. Although they do
not expecttheir constituentsto have extensive knowledgeof theiroverall
voting record,membersof Congressdo worry that a stringof votes that
puts them at odds with their district can be electorallycostly (Fenno
1978). Frequently, though they recognize that most constituents are
unaware of the bulk of their day-to-day Washingtonactivity,members
anticipate how voters would react if they becameinformedon their rep-
resentatives' voting records-via the media, an electoral challenger,
interestgroups,or from some other source (Arnold 1990).Perhapsvali-
dating these members' concerns, recent studies have shown that
memberswho have an ideologically extremevoting recordreceive lower
vote shares and have a lower probabilityof re-election(Canes Wrone,
Brady, and Cogan 2002); along similar lines, Carson et al. (2010)
demonstratethat the same is true for members who have a record of
voting more frequentlywith theirparty.
In addition to the reputationof individualmembers,the roll-call
record affects the public's perception of Congress as an institution.
When respondentsare given open-endedquestions about what makes
them dislike Congress, conflict, bickering,and partisanshipfrequently
appear at the top of the list (Hibbing and Theiss-Morse 1995, 2002).
Consistentwith Hibbing and Theiss-Morse'swork, recent studies have
demonstrated that quarterly measures of congressional approval are
driven in part by the level of partisan conflict in Congress (Ramirez
2009, 2013). By measuring the quarterly proportion of congressional
roll-callvotes where at least 75% of Democratsoppose75% of Republi-
cans, along with other indicatorsof conflictsuch as debt-ceilingvotes
and Senate cloture votes, Ramirez demonstratesthat approval in the
aggregatedoes appearto be negativelyaffectedby partisanconflict.Per-
haps key for this discussionis the finding from Jones (2010) and Jones
and McDermott (2004, 2010) that congressionalapproval electorally
affectsthe partiesdifferently.Specifically,when approvalis high, voters
reward the majorityparty;when approvalis low, candidatesof the con-
gressionalminorityparty tend to benefit.Loweringthe public's appraisal
of Congressought to work to the electoralbenefitof the minority party;
consistentwith Ramirez (2009, 2013), requestingrecorded votes in a
way that makes Congressappearmorepartisanis one way to accomplish
this goal.
Beyondaffectingthe appearanceof Congressas a whole, establish-
ing a record of opposition to the majority party can be electorally
valuableto the minorityparty. Creatinga legislative recordof opposition
to majority-partyproposals places the minority in the most credible
possible position to criticize the majority party in the next election.
Sundquist,though discussingthe relationshipbetweenCongressand an
opposing-partypresident, elaborates this logic of electorallyvaluable
If the president sends a proposal to Capitol Hill or takes a foreign policy ;iand, the
oppo;ition . . . simply mw,i reject it. Otheiwise they are saying the president is a wise
and prudent leader. That would only ;irengthen him and his party for the nex t election
. . . when their whole object is to defeat him when the time arrives. ( 1988, 630, quoted
in Lee20 09)

Toward the end of their long stint as the minorityin the House, a
growing number of House Republicansclearly saw confrontation,not
compromise,as the most direct route to majority status. Scholars, for
instance, note the increasingly aggressive strategy adopted by Newt
Gingrich(R-GA)and other so-calledYoung Turks of the Conservative
OpportunitySocietyin the 1980s.As some minorityRepublicanssensed
that their growingelectoralstrengthresultingfrom changesin the South
was not being proportionatelytranslatedinto seat gains in the House,
Gingrich and others believed a change in minority-partystrategy was
needed to regain majority status. Those Republicans who favored
confrontingmajority Democrats and making campaign issues out of
legislationinsteadof seekingpolicy compromisefoundthemselvesfrus-
tratedwith colleagueswho were ,villingto work with the Democrats.As
former House member Bob Walker (R-PA) explained, the attitude of
more senior Republicans reflected their experience as the House's
"permanentminority"throughmuch of the late twentiethcentury,during
which they sought compromiseand accommodation.Republicanslike
Walker saw this attitudeas a self-fulfillingprophecy,arguingthat work-
ing with the Democratson policyleft minorityRepublicanswith "no real
case to take to the country"duringcampaignseason(quotedin Connelly
and Pitney 1994, 26).
Many members of the Republican House minority in the late
1980sand early 1990sgarneredthe label "bombthrower" (Connellyand
Pitney 1994). These were the members who sought electoral gain
through'"confrontation', confrontation,confrontation'and by 'polariz-
ing issues' they 'sharpenthe differencesbetweenthe two parties.' Theirs
is an ideologicalpoliticsof combat.The bombthrowersfrequentlyattack
the majority's record, and Democrats accuse them of being anti-
Congress"(Connellyand Pitney 1994, 27). Again, this discussionsug-
gestsmore nonideologicalelectoralincentivesfor the minorityto engage
in conflictwith the majority:to maintainthe most crediblepossibleposi-
tion when attackingthe majority's record. Moreover,partisan unity by
the minorityparty, even in legislativedefeat, serves to deny opportuni-
ties for majority-partymembersto claimbipartisanlegislativesuccesses.
Basedon the precedingdiscussion,membersof the minorityparty
should have a number of partisanelectoralincentivesto affectthe com-
position of the roll-<:allrecord. Consistentwith Canes-Wrone,Brady,
and Cogan (2002) and with Carson et al. (2010), the minority party
should be more likely to request recordedvotes in such a way to make
individualmembersof the majorityparty sufferthe electoralconsequen-
ces of appearingideologicallyout of step with their constituentsor more
frequentlyunited with their party. And consistentwith the above-listed
studies on congressionalapproval, the minority party should be more
likelyto requestvotes in a way that enhancesthe appearanceof partisan
conflict,with the expectationthat voters will retaliateby electorallypun-
ishing the majority party; moreover, doing so makes the majority's
bipartisan appeals less credible. Both motivations-to alter the voting
records of individual members and to alter the public's perceptionof
Congressas an institution~ught to mean that, as a matter of electoral
strategy, the minority party requests votes that are more polarizing in
naturethan votes requestedby the majorityparty.Note that the claim is
not that minority-partyrequests for recorded votes have a polarizing
effect on the outcome of votes; rather, I argue that members of the
minorityparty should be more likely than members of the majorityto
selectdivisiveissuesto place on the recordand that this type of behavior
should have the consequenceof inflatingmeasures of partisanshipand

Data and Methods

Although scholarshave examinedissues relating to the construc-

tion of the roll-callrecord,includingthe differencesbetweenlegislative
items that receive recorded votes and those that do not (Clinton and
Lapinski2008; Schicklerand Pearson 2009), as well as differencesin
members' behaviorpatterns on different types of recorded votes (e.g.,
Jessee and Theriault 2014; Lee 2009; Theriault 2008), no published
studies explore the question of who demands recordedvotes in the US
House, a questionof direct importancewhen it comes to understanding
the partisanbasis for the constructionof the roll-callrecord.3 To address
this question,I create an original data set of requestsfor recordedvotes
in the House of Representativesfrom the 104th Congress (1995-96)
through the 111th Congress (2009-10). For each recorded vote that
Vote Requests by Party, 104th through 111th Congress
Minority Majority Minority Requests as Minority
Congress Requests Requests Percentage of All Roll Calls Share of Chamber

104 748 487 0.61 0.47

105 602 498 0.55 0.48
106 552 587 0.48 0.49
107 529 408 0.56 0.49
108 683 476 0.59 0.47
109 660 502 0.57 0.46
110 1203 604 0.67 0.46
111 853 716 0.54 0.4 1

occurredduring this time period, by searchingthe Congressional Record

I identifiedwhich member requested that the vote be recorded. During
this time period, 10,203 recorded votes were taken. For this analysis, I
exclude 95 recorded votes on which the yeas and nays were taken by
rule, or votes that were demanded by nonvoting delegates. Remaining
are 10,108 recorded votes between 1995 and 2010 that were demanded
by a full member of the House.
I proceed in this section by examining descriptivepatterns in the
data to see how often members of the majority and minority request
votes, whether minority-requested votes are more divisive and partisan
than majority-requested votes, and their effect on aggregatemeasures of
polarization. I then systematicallyanalyze how minority-requestedvotes
affect the records of individual members by modeling members' differ-
ence in party unity and ideology scores based on majority- versus
minority-requested votes, ,vith the expectation that minority-requested
votes should make vulnerable members of the majority party appear
more partisan and ideologicallyextreme.
As shown in Table 1, the majority of House roll-call votes are
demandedby members of the minorityparty: 5,830 out of 10,108(about
57.7%) of roll calls between 1995 and 2010 ,vere demanded by
minority-party members. The percentage of minority requests varies
somewhat by Congress, with a low of 48.5% in the 106th Congress
(1999-2001) and a high of 66.6% in the 110th Congress (2007-09). In
all but one Congress (the 106th), the majority of roll calls come as a
result of minority-party demands. In itself, the fact that most of the
House's roll-call record comes as a result of minority-, not majority-
party requests is an intriguingpreliminaryfinding.
Party Voting on Minority-Party-RequestedVotes, 104th through
111th Congress (1995-20 I 0)
Party Votes Party Votes as Minority R<.-qucsts as
Congress (Minority Request) Percentage of Minority Requests Percentage of Party Votes

104 596 0.80 0.70

105 416 0.69 0.70
106 352 0.64 0.68
107 321 0.6 1 0.79
108 494 0.72 0.84
109 4 81 0.73 0.78
110 884 0.73 0.83
111 608 0.71 0.82

As mentionedabove,the minoritypartyought to have a numberof

incentivesto requestroll-callvotes in a way that enhancespartisancon-
flict. One method used by scholars to determine levels of partisan
conflictin voting is to simplycalculatethe numberof "party votes," or
votes whereat leasthalf of one party opposedat leasthalf of the other.If
partisan conflict makes voters disapprove of Congress (Hibbing and
Theiss-Morse1995,2002; Ramirez2009, 2013), and voters punish the
majority party as a result of low approval (Jones 2010; Jones and
McDermott2004,2010),membersof the minorityparty shouldbe more
likely to request party votes than the majority party. Table 2 breaks
down vote requestsby majority/minorityparty and whetherthe vote in
questionwas a partyvote.5
ResultsfromTable 2 and Table 3 show strikingdifferencesin par-
tisan voting behavioron votes requestedby members of the minority
and majority parties. During the entire time period, approximately
54.1% of all recordedHouse votes were party votes; of these, 77.1%
were requested by the minority party. A full 72% of all minority-
requestedvotes are party votes, comparedto just 29.4% of majority-
requestedvotes. In each Congress,well over half of minority-requested
votespit thepartiesagainsteachother(moreoftenin the 60--80%range),
while fewerthan half of majorityrequestsresultin partyvotes.Minority
requestsconsistentlyaccount for the bulk of party votes taken in the
House. By this measure,the minorityparty, as expected,appearsto be
the mainreasonfor Congress'srecordof partisanshipin recordedvotes.6
Another,vay to measurethe divisivenessof votes is to locate the
ideologicalcut point. In the languageof spatialvoting, the cut point is
the ideologicallocationthat dividesthe supportersfrom the opponents
Party Voting on Majority-Party-RequestedVotes, 104th through
111th Congress (1995-2010)
Party Votes Party Votes as Majority Requests as
Congress (Majority Request) Percentage of Majority Requests Percentage of Party Votes

104 250 0.5 1 0.29

105 175 0.35 0.29
106 162 0.28 0.31
107 83 0 .20 0.20
108 90 0 .19 0.15
109 133 0.26 0.22
110 181 0.30 0.17
111 128 0.18 0.17

of a measure.Votes that divide the parties should have cut points that
fall close to the ideologicalcenter, neatly dividing Democrats from
Republicans.Roberts and Smith (2003), for instance, show that cut
pointson House amendmentvotes fall betweenthe party mediansmuch
more frequentlythan finalpassage or other types of votes. Tables 4 and
5 show the total number and the percentageof votes requestedby the
minorityand majorityparties that produceda cut point in-betweenthe
ideologicallocation of the median Democrat and median Republican
duringeachCongress.By this measure,minority-requested votesremain
far more divisivethan majority-requestedvotes; the totalnumberas well
as the percentageof votes that producea cut point dividingthe partiesis
far greaterfor minorityrequestscomparedto majority requests.In only
one Congress(the 104th)did morethan half of majority-requested votes
produce a cut point between the party medians; the next-closest
Congressis the 105thCongresswith 36.1%. By contrast,more than half
of minority-requestedvotes produce party-dividingcut points in each
Congress,reachinga highof 81.4%in the 104thCongress.
As argued by scholarslike Steve Smith, Jason Roberts, Frances
Lee, and Sean Theriault,measuresof polarizationin Congressbased on
the roll-callrecordare influencedby more than pure ideologicalprefer-
ences and can be affectedby nonideologicalfactors.I argue here that
minority-partyelectoralincentivesmean that the minorityhas an incen-
tive to make Congresslook more partisanby requestingthat instances
wherepartiesdisagreethe most becomea matterof publicrecord.These
motivationscan affectmeasuresof memberideologyand consequently
about inferencesregardingthe degreeof congressionalpolarization.As
Lee explains:
Cut Points on Minority-Requested Votes, 104th through 111th
Congress (1995-2010)
Cut Point between Percentage Cut Point
Total Minority-Party Medians between Medians
Congress Vote Requests (Minority Requests) (Minority Requests)

104 748 609 0.81

105 602 406 0.67
106 552 297 0.54
107 529 277 0.52
108 683 405 0.59
109 660 423 0.64
110 1203 93 1 0.77
111 853 541 0.63

Cut Points on Majority-RequestedVotes, 104th through 111th
Congress (1995-2010)
Cut Point between Percentage Cut Point
Total Majority-Party Medians between Medians
C-0ngrcss Vote Requests (Majority Requests) (Majority Requests)

104 487 255 0.52

105 498 180 0.36
106 587 165 0.28
107 408 83 0.20
108 476 87 0.18
109 502 134 0.27
110 604 183 0.30
111 716 129 0.18

Any vote that tends to divide the two parties from one another, regardlessof the rea-
son, will be treated by such methodologies as measuring members' ideology . . .
Equating party votes with ideological votes is not warranted if .. . panisan conflict can
have any other sourcebesides members' differing preferences on public policy. (2009,
52- 53, emphasis in original)

The evidence shown so far demonstrates a much higher level of

partisan divisiveness on minority-requested votes, consistentwith previ-
ously discussed minority-party electoral incentives. To examine the
question of how the minority party affects measures of polarization, I
use the W-NOMINATE program developed by Poole et al. (2011) to
compute estimates of legislator ideal points. Though NOMINATE
scoresare typicallycalculatedbasedon all nonunanimousrecordedvotes
in a given Congress,I calculatetwo subsetsof ideal point estimatesfor
legislatorsin each Congress,one based only on votes requestedby the
minorityparty and another based on votes requested by the majority
party. Scoresbased on minority-requestedvotes shouldmake Congress
lookmorepolarizedthan scoresbasedon majorityrequests.
The ideologicaldistributionof Democratsand Republicans,shown
in Figure l, differs considerablyin each Congress when using W-
NOMINATE scores based on majority versus minority requests.
Minority-basedideologyscoresmore frequentlymake the partiesappear
both more polarizedand more homogeneousthan majority-basedideol-
ogy scores,which suggest a much greaterdegree of bipartisanoverlap.
The degreeto which this is true varies from Congressto Congress,but
the pattern is consistentin each. Occasionally,the biggest differences
appearamongmembersof the majorityparty (the l 05th and l 09th Con-
gresses are good examples),while in others the overall ideological
changesappearto operateroughly equally across the parties (as in the
108thCongress).Again, votes requestedby the minorityparty serve to
make Congressappear more partisan and more ideological,consistent
with the argument that doing so is in the minority party's electoral
Next, I examinethe estimatedideologicalrank order of legislators
frommostto leastliberalin eachCongressbasedon differentvote subsets.
Figure2 plots, for each Congress,the ideologicalrank orderof all mem-
bers based on majority-request-based W-NOMINATEscores on the x-
axis,with the minority-request-based scoreson the y-axis.Memberswho
fall precisely on the 45-degreeline exhibit no change in rank order
betweentheirtwoscores;thosewho fallaboveor belowwitnesstheirover-
all liberal-<:onservative
rank alteredbasedon the differentvote subsets.
The results from Figure 2 demonstratethat, in additionto infer-
ences about aggregatelevels of polarizationand ideologicalconflictin
Congress,minority-requested voteshavean impacton conclusionsabout
the ideologicalplace of memberswithin their own party. Estimatesof
members' ideologicalrank order are considerablydifferentwhen based
on minority-versus majority-requestedvotes. An interestingpattern is
that the majorityparty (Republicansin all Congressesbesidesthe 110th
and 111th) consistentlyexhibits a greater degree of variation in rank
orderbetweenthe two subsetsof votes.Figure2 showsa greaterlevelof
dispersionaround the 45-degree line for the majority party in each
Congress,suggestingthatmembersof themajorityparty tendto see more
considerablechangesin their ideologicalplacementon minority-versus
majority-requestedvotes. Though not conclusive, this is perhaps
DistributionofW- NOMINATE Scores Based on Minority Requests
vs. Majority Requests, 104th through 111th Congresses. [Color
figure can be viewed at]
1kth Congfffo• (11H - 1M J} tOSlbeong,-. (tffl -'1..., , ... c I ... C111t-191 )

•• •• •• •
•• •• ••
•• •• •• •• ••

1•- l ••- 1••-
'•• - -
•• -

•• ,. .......,... ..,..........,...
W flOIIMlt ...

1071ftCOnf,ff• (20Clt•20H } IOllh C. 'fl ... CIN>..MN) ,... c, , .. CIIM-111'1)

' '
= :•
= •• • • ' •.
f •- f •-

•• •

,." ..., •• ,... ........ •

•• ,
WN01•1'lt ....
, ••
.. .•• •• ......

110th ~- (20Cl1• 20tt} tttl hCo. ...... ~ltH •

:I ,, :
= '
'' '

'' •• r.
' ''


'• '

·. -
•• D
II. -~----·

• 1.t • t •

-~ ......
... ....

• .,, ., ........
., 11(,al lliolft:..,_
•• ••

suggestive that members of the majorityare strategicallytargeted for

manipulationof theirvotingrecordby membersof the minorityparty.
Aggregateevidence shown so far is consistentwith a minority-
party strategyof heighteningthe appearanceof conflictand partisanship
in Congress for electoral purposes: Compared to majority-requested
votes, minority-requestedvotes are more likely to result in party votes
and to produce a cut point between the parties' median members;
Liberal-Conservative Rank Order of the House Using
W-NOMINATE, Based on Minority-and Majority-Requested
Votes, 104th through 111th Congress. [Color figure can be viewed
104th Congr•n (1995-1997) 105th CongreH (1ff7 - 1ttt ) 10lttl Congress (t ltt-2001 )

0 100100•-t00 f 1001'0--
.. ---
107th Congren C2001
• 200J• 108th Coftg,ut (ZOl>-2005I 1ottb ~.t (Zlff-2111)

. ___ _ ,
.... ---
110th co.. tt(2007 • 200t, tt tth c-.,. .. (20ft-tot1J

0 100200)00"00 I 100JODJOO..O

moreover, ideologyscoresbased solely on recordedvotes requested by

the minorityparty paint a picture of a Congressfar more ideologically
polarizedthan ideologyscoresbasedon majority-requested votes.
What remainsto be shown is whetherthe minorityparty attempts
to target electorallyvulnerable members of the House majorityparty.
Scholars have long documented legislators' attempts-particularly
minority-partyattempts-to put opponentson the record in difficultor
embarrassingpositions (Robertsand Smith 2003; Smith 1989).Claims
that an incumbentmember is out of step with his or her district,or is
overly partisan,are commonplaceduring congressionalcampaigns.For
instance,the National RepublicanCongressionalCommitteeran adver-
tisements in 2014 against John Barrow (D-GA) for voting ''with
[Barack] Obama 85% of the time"8 and against Carol Shea-Porter
(D-NH)for voting ''with Nancy Pelosi's Democrats95% of the time."9
In the same district,the DemocraticCongressionalCampaignCommit-
tee (DCCC)attackedboth prospectiveRepublicanHouse nominees for
being "too extreme" for Ne,v Hampshire's first district.10 Such claims
are based on calculationsof party-unityscores that do not make distinc-
tions between vote types. Again, evidence does suggest that members
pay an electoralprice for ideological extremity(Canes Wrone, Brady,
and Cogan2002)or for partisanunity (Carsoneta!. 2010).
Accordingly,if vote requests by the minority party are driven by
electoralstrategy, its members ought to request votes in a manner that
makes electorally vulnerable members of the majority appear more
extremeand more unifiedwith their party. This is likelyto be an inexact
science;recognizingthat enforcingpartisanunity is an activitythat can
cost its membersat the polls, party leadersdo not expectcompleteparty
unity on every vote (Carson et al. 2010; Lebo, McGlynn, and Koger
2007). Instead, partisan pressure tends to be exerted selectively,exer-
cised only when absolutelyneeded to win on legislation,and leadersof
the majorityparty prefer to allowelectorallyvulnerable membersto cast
votes against the party when it benefitsthe member electorally(King
and Zeckhauser2003; Smith2007). Thus, while the minoritymight try
to force electorallymarginalmembersof the majority party to stick with
their party more often and appearmore ideologically extreme,members
of the majorityand their leadersactivelytry to avoidsuch an outcome.
From the minority'sperspective,makingthe majorityparty appear
more unified and ideologically extreme could ideally be done without
themselvesincurringthe electoralcosts of extremityor partisan unity.
This,however, is unlikelyto be the case.Denyingsupportto the majority
party (i.e., unifyingagainst the majority party) is a key method to spur
greaterunity among membersof the majority,as the majorityis forcedto
exert a greatereffort to keep its rank-and-filein line and still win despite
unified opposition(Lebo, McGlynn,and Koger 2007). So the minority
has a number of considerations: how to increase the appearance of
partisanshipand ideologicalextremityamong members of the majority
party, while at the same time minimizingthe costs of partisanshipand
extremityamong its own members. Anecdotal evidence suggests that
membersof the minorityrecognizethe potentialtrade-offbetweenseek-
ing re-electionand seekingmajoritystatus,and some are more willingto
bear such costs than others.A formermemberof HouseRepublicanlead-
ershipduringtheirlong stintin the minoritywas quotedas saying:
It [the wi llingness to be a "pany activfat'1 fluctuates over time, and it's also to some
extent related to how safe the House member's district is . . . . There's not a perfect cor-
relation there, but it's a lot easier for a guy to be willing to go to war if he's not that
worried about getting re-elected. I would put Newt [Gingrich) right at the top of the list
of those guys who are primarily intere.ied in a majority and a national stmtegy. That's
one of his strengths. That's what he always brings to the fmy. But it's a lot tougher for
a member to get caught up in all of that, if in fact he's got a tough dhiri~i or a re-
election campaign back home. (Quoted in Connelly and Pitney 1994, 25}

Aggregate-levelvariation in party unity and W-NOMINATE

scoresoperatesas anticipated-minority-requestedvotes sho,v more evi-
dence of polarization and partisan conflict. Expectations about how
individual legislators' records differ between minority-and majority-
requestedvotes are less clear.The minorityparty ought to requestvotes
in a way that makes vulnerable majority members look more extreme
and more partisan; but the majority's ''vote options" (King and Zeck-
hauser 2003) and selectiveuse of direct pressure (Smith 2007) might
help their members avoid this outcome,asking their safe members to
take difficult votes while the vulnerablemembers are let off the hook.
Similarly,the minorityshouldtry and do the same, allowingtheirvulner-
able members to vote against the party when necessary. Still, the
minoritymight be more willingto endurethe electoralcosts of party loy-
alty or the appearanceof greaterextremityif it contributesto the goal of
majoritystatus. This is suggested by Connelly and Pitney's discussion
of"party activists"like Newt Gingrich,who sacrificedsome attentionto
their district in pursuit of House majorities.Similarly,Green (2015, 35)
observesthat comparedto the majorityparty,the Houseminorityis often
more concernedwith defeatingmembers of the majoritythan ,vith pro-
tectingits o,vn members.Additionally,Greenarguesthat minority-party
unity in voting can createopportunitiesfor "messagevotes" (2015, 157),
can place the blamefor outcomessolelyon the shouldersof the majority,
and can stand the minorityin a crediblepositionto criticizethe majority
in the future. For all these reasons, members of the minority might be
willing to endure some electoral costs themselvesin order to impose
electoralcosts on membersof the majorityparty.
Whilethe minoritypartyhas the incentiveto imposeelectoralcosts
on the majorityparty by demandingroll-callvotes on difficultissues,an
objectionto this argumentcan be raised:The majorityparty's control
over the legislativeagenda(Cox and McCubbins2005) ought to allow
the party to avoid embarrassingvotes on which the minorityparty can
requestroll calls. Despitethe majorityparty's agendacontrol,however,
thereis stillreasonto believethat the minorityhas some opportunitiesto
demandroll-callvotes in a way that mightharm the majorityat the polls.
First,the majoritymight allow an electorallycostly vote becauseof the
importanceof the policy in questionto the party. Second,the minority
doeshave some opportunitiesto createembarrassingvotesvia use of the
motion to recommit,which House minoritieshave skillfullyused as
political weapons in recent Congresses(e.g., Layton 2007; Ornstein
2010). Third, despite the dearth of open rules in the contemporary
House, based on originallycollectedamendmentsponsorshipdata col-
lected from the CongressionalRecord,membersof the minorityparty
sponsored55% of the amendmentssubject to a roll-callvote between
the 104thand 111th Congresses,suggestingthat the minorityparty is
not completelyshut out of the Houseamendingprocess.Lastly,the votes
allowedby the majorityparty may not alwaysbe individuallycostly,but
they may be costly in the aggregate.Campaignadvertisementslike the
ones mentionedabove that includeaggregatestatisticsabout members'
recordsof partisanloyaltydo not includeany informationaboutthe con-
tent of the votesthemselves.In otherwords,minority-requested voteson
partisanissuessuch as proceduralmotionsor votes on approving special
rulesmay not have much substantivemeaningto voters, but thesevotes
tend to be morepartisanand havethe consequenceof heighteningmem-
bers' aggregate party-unity scores, providing possible fodder for
I now explorehow the party unity and ideologyscoresof individ-
ual legislatorsdiffer betweenmajority-and minority-requestedvotes. I
calculatethe differencein perceivedideologicalextremityfor eachmem-
ber based on the differentvote subsets (a member's W-NOMINATE
score based on minority-requestedvotes, minus his or her W-
NOMINATEscore based on majority-requestedvotes, coded so that
positive values indicate greater ideological extremity on minority-
requestedvotes).I also calculateeachmember's differencein partyunity
on differentvote subsets(party-unityscore on minority-requestedparty
votes,minusthe samefor majority-requested partyvotes;positivevalues
indicatethat a member'svotingrecordon minorityrequestsis morepar-
tisan than his or her record on majorityrequests).We should expect a
numberof patternsbothwithinand acrossparties.Minorityvote requests
Differencein Measured IdeologicalExtremism,Minority-vs.
Majority-RequestedVotes. (a) Difference in Ideological Extremity
of Minority-PartyMembers on Minority vs. Majority Requests, by
Seat Safety. (b) Differencein Ideological Extremity of Marginal
Members on Minority vs. Majority Requests, by Party

-- •"
• •
• •
. '•
• •
I: I= •.
. .•• •~.
. ..

•• •

' ••


•• •I
~ ................
00 U ._. U U .. . .. .

41 ••

., U ,.

(a) (b}

should make membersof the majorityparty appear disproportionately

more ideologicallyextreme and partisan, especiallyamong the major-
ity's mostvulnerablemembers.By contrast,the minorityshouldattempt
to protect its most vulnerablemembers,concentratingany increasesin
unity amongthe minority'ssafest members.
As a preliminarydescriptive look at the data, Figures3a and 3b
plot which membersare madeto lookmore or less ideologicallyextreme
based on minority·requestedvotes. First, Figure 3a plots differencesin
minority-party members' apparent ideological extremity based on
minorityversusmajorityrequests.As anticipated,increasesin extremity
on minority·requestedvotes are more commonfor safe minoritymem-
bers (definedas those membersreceivinggreaterthan 55% of the two-
party vote in the previousgeneral election)and less common for mar-
ginalmembers(thosewho received55%or lessof the two•partyvote). 11
Figure 3b shows that, among electorallymarginalmembers,minority
vote requestsincreaseideologicalextremityscores for membersof the
majorityto a greaterextentthan membersof the minority.
Movingbeyondsimplehistograms,I examinethe questionof elec-
toral marginsand minority-partyvote targetingmore systematicallyby
modeling the variables describedabove: members' differencein per-
ceived ideologicalextremityand difference in party unity across the
differentvote subsets.Specifically,the unit of analysisis an incumbent
legislator,and differencein ideologicalextremityand differencein party
unity betweenminority-and majority-requestedvotes serve as depend-
ent variables.For the majorityparty, the dependent variablemight be
consideredthe degree to which a memberis successfully"targeted" by
minority-partyvotes to appear more ideologicallyextremeor partisan;
for the minority, the dependent variable might be considered which
members bear the electoralcosts of targetingmembers of the majority
party. I estimatefixed effects linear regressionmodels that account for
unobservedcharacteristicsof individuallegislatorsthat do not vary over
time, such as party, ideology,region,and so on.12 Fixed effectsmodels
leverage variationwithin units (here, variation among legislatorsfrom
year to year) to estimatethe effects of time-varyingindependentvaria-
bles (Wooldridge2005). Specifically,the time-varyingfactorsfor which
I controlincludethe incumbent's share of the two-partyvote in the most
recent election,his or her districtpartisanship(measuredas the share of
the two-partyvote receivedby the presidentialcandidateof the legisla-
tor's party in the most recent presidential election), his or her
membershipin the majorityor minorityparty, and an indicatorfor the
party identificationof the presidentrelativeto Congress;the latter vari-
able takes on a value of 1 when the presidentis of the same party as the
House majority, and O otherwise. Because minority-requestedvotes
should affect membersof the majorityand minorityparties differently,
interactions between majority/minority status and each of the other
time-varyingvariablesare included.
Votes requestedby the minorityparty should targetmajority-party
memberswho are electorallyvulnerable, suggestinga negativerelation-
ship betweena majoritymember's seat safety and his or her increased
extremity/partisanshipon minorityrequests.Consequently,an interaction
betweenMajorityParty and IncumbentPrevious Vote Share should be
negative.The majorityought to try and shieldits vulnerablemembersby
allowingmembersin weakly partisandistrictsto vote againstthe party,
whileseeking additionalsupport from membersin stronglypartisandis-
tricts(thosewho are more likelyto be able to affordto take toughvotes).
Accordingly,the coefficienton the interactionterm betweenMajority
Party and DistrictPartisanshipshould be positive. Statisticalevidence
suggeststhat presidentsthemselvescan drive up partisanshipand polar-
izationin the legislative process(Lee 2009);anecdotalevidencesuggests
thatparty leadersattemptto build recordsof oppositionnot only to initia-
tives of the majorityparty, but to initiativesof presidentsof the opposing
party, engaging in "strategic disagreement" for electoral putposes
(Gilmour1995;Lee 2009, 2013; Shelley 1985).The main coefficienton
Fixed Effects Regression:Difference in W-NOMINATEScore and
Difference in Party Unity Based on Minority-vs. Majority-Party
Vote Requests
(1) (2)
\V-NOMINATE Dilf. Party-Unity Diff.

Majority Party 0.393** 0.0173

(0.054 1) (0.0203)
Majority-Party President 0. 160•• 0.0568••
(0.0 104) (0.00389)
Incumbent Previous Vote Share 0.00399•• 0.00108••
(0.000894) (0.000335)
District Partisanship - 0.00201 - 0.00237••
(0.00113) (0.000425)
Majority Party • Majority-Party President - 0.0827*• -0 .0s13••
(0.0 15S) (0.00580)
Majorit y Party • Incumbent Previous Vote Share -0 .00728•• - 0.00 18s••
(0.001 I 0) (0.0004 14)
Majority Party • District Partisanship 0.00322•• 0.00231 ••
(0.0009S I) (0.0003S7)
Constant - 0.107 0. 109••
(0.0647) (0.0242)
N 2972 2972
If 0.236 0.200

Note: Standard errors in parentheses; Includes member- level fixed effects.

• p < 0.05, •• p < 0.0 1.

Majority-PartyPresident should be positive (reflectingthe minority's

attemptto build a recordof oppositionto opposing-partypresidents),and
the coefficienton the interactionbetweenMajorityParty and Majority-
Party President should be negative,becausea majority-party president
shouldspur moreunityamongthe minoritypartythan the majority.
Resultsdisplayed in Table 6 confirmexpectations.Minorityvote
requests make electorallyvulnerable members of the majority party
(measuredby their previous share of the two-partyvote) appear more
ideologically extreme and more frequentlyunited with their party on
party votes. As anticipated,the interactiontenn on Majority Party and
District Partisanshipis positiveand significant,thoughinterpretationof
the substantive effectof districtpartisanship(seeFigure4) doesnot sug-
gest that this is becausethe majorityparty makes its membersin strongly
partisandistrictsvote more frequentlywith the party-in fact, members
of the majority see a fairly constant increase in unity on minority
requests,acrossall types of districts.Instead,it reflectsthe fact that the
Effect of District Partisanshipon Difference. in Party Unity
(Minority-vs. Majority-RequestedVotes)

Effect of District Partisanship on Unity Difference , by Party

With 95% Confidence Interval

~ ~ ~ ~ ~ ~ ~ ~ 70 75 ~ 85 00 95 100
Dislrict Partisanship

I---- Minority Patty -- Majority Patty I

largestincreasesin unity amongthe minorityparty occur in weaklypar-

tisan minority districts.In other words, for the minority,members in
weaklypartisandistrictssee the greatestincreasein party unity,consist-
ent with Theriault's(2008) findingthat minorityHouse membersfrom
weaklypartisandistrictsdisproportionately contributeto the modempro-
ceduralpolarizationin the House, supportingtheir party on procedural
mattersmuch more often in the modernCongressthan in the past, while
still occasionallyvotingagainstthe minorityparty on substantiveissues.
When the presidentis of the same party as the House majorityparty,
membersof the minorityparty appear more unified and ideologically
extreme on the votes that they request. This again might reflect the
minority's attemptsto build a record of oppositionto the presidentand
his party (Gilmour1995;Lee 2009; Sundquist1988).
Figures 5 and 6 demonstrate the predicted effect of members'
partisanship, previous vote margins, and partisan relationship with
the president, with other variables held at their means. Figure 5, for
instance, suggests that a member of the majority party winning with
just over 50% of the two-party vote share will be forced to side with
Effect of Seat Safety on Differences in Unity and Extremity, by
Party (Minority vs. Majority Requests) . (a) Seat Safety and
Changes in Party Unity (Minority- vs. Majority-Requested Votes).
(b) Seat Safety and Changes in Ideological Extrem ity (Minority- vs.
Majority-Requested Votes)

Effect of Previous Vote Share on Unity Difference , by Party

With 95% Conridence lnt91Val


50 ~ 50 M ro n MM 00 ~ 100

1-- Minority Pany -- Majonly Pany !


Effect of Previous Vote Share on Difference in Extremism , by Party

With 95% Co nfidence lnt91Vai

50 ~ 50 M 70 7S MM~~ 100
Vote Share

1-- Minority Pany -- Majority Party I

tuect 01 rres1oenua1 rany on u111erencem rany uruty vvrmonty-
vs. Majority-RequestedVotes)

Effect of Majority-PartyPresidenton Unity Difference,by Party

With 95% ConfidenceInterval

0 1

I---- MinontyParty -- Majority Party I

his or her party approximately 9% more often on minority-requested
votes compared to majority-requestedvotes, compared to a party-
unity increase of just 5% for a similarly situated member of the
minority. Figure 6 demonstrates that, on average, a member of the
House minority will vote with his or her party approximately 6%
more often on minority-requested votes when the opposing party
controls the presidency. Overall, vote-requesting behavior by the
minorityparty does appear to be consistentwith an electoral strategy
of increasing ideological extremity and party unity among the most
vulnerable members of the opposing party and increasing their own
unity when the opposing party controls the presidency.

Discussion and Conclusions

The precedingevidenceis stronglysuggestiveof a strategyby the

minorityparty to make Congressappearmore partisanand polarizedin
the aggregate.On an individuallevel,voting behavioron majority-and
minority-requestedvotes is also suggestiveof a minority-partyattempt
to make electorallyvulnerablemembers of the majority party appear
morepartisanand ideological,as well as to build a recordof opposition
to opposing-partypresidents,while shiftingthe costs of this strategyto
its electorallysafe members.Whether or not the minorityparty is suc-
cessful in electorally harming the majority via its vote-requesting
strategiesis a separatequestion,but anecdotaland statisticalevidenceis
highlysuggestiveof attemptsto do so.
Effortsin recentdecadesby membersof the Houseminorityparty
to sharpendifferencesbetweenthe parties, particularlyby Newt Gingrich
and the ConservativeOpportunitySocietyin the 1980sand 1990s,have
received some scholarly attention (e.g., Connelly and Pitney 1994;
Schickler2001; Smith 1989;Zelizer2004), but the attentionpaid to the
minorityoftencomesin the formof anecdotalaccountsor descriptionsof
the minority's effort to gain publicity outside the legislaturethrough
sharpenedpartisan rhetoric and appearanceson news programs;other
accounts that discuss the House minorityparty within the legislature
often focus on members who took advantageof time for one-minute
speechesto make barbedpartisanattacks.Far less attentionhas beenpaid
to the broaderlegislativeconsequencesof the Gingrich-likeshift in the
behaviorof the Houseminorityparty from a strategyof compromiseto
one of conflict (for counterexamples,see Roberts and Smith 2003;
Theriault 2013). Because of its particular electoral incentives, the
minorityparty in Congress is a potential candidate to help explain the
rise of observed polarization in Congress; if the minority party is
simply more likely to request recorded votes on divisive issues today
than it was in previous decades,then this could help account for rising
levels of polarizationin the House. Data presented here, which begin
in 1995, cannot speak to whether this is indeed the case, but it is
suggestive of a new line of research into the origins of polarization
nonetheless.Moreover,consistentwith others who have raised similar
concerns (e.g., Jessee and Theriault 2014; Lee 2009; Roberts 2007),
this work suggests that caution should be used when using ideology
scores, which assume all conflictin recordedvoting is a consequences
of members' differing preferenceson public policy. More than just a
register of legislators' opinions on issues, the roll-call record is also a
target of strategicmanipulationfor members' partisanelectoralgoals.
Future research should extend this data-collectioneffort on
requestsfor recordedvotesacrossbothtimeand acrosschambers.Doing
so would providea betterunderstandingof how the roll-callrecordhas
been constructedover time.As the partiesare ever morecompetitivefor
controlof nationalpoliticalinstitutions,the premiumplacedon any indi-
vidual factor that might produceelectoralgain for one side or the other
shouldnaturallyincrease(Lee2016).Consequently,manipulationof the
roll-callrecordfor partisanelectoralpurposesmay be highertoday than
during the mid-twentiethcentury when Republicans served as the
"permanentminority"in the House, and to a lesser extent, the Senate
(Mann1988).Exploringwhetherthis is the casecan shedlighton whether
the rise in polarizationhas to do with changingvote-requestingbehavior
by the minority.Regardless,the resultsshownhere call for greateratten-
tion to the minority party in Congress,as even a procedurallyweak
minoritymight have a broaderimpacton polarization,partisanconflict,
theappearanceof Congress,and,ultimately,the outcomeof elections.

WilliamEgar<> is a Social Science Analyst at

the US GovernmentAccountabilityOffice, 441 G St. NW, Washington
DC, 20314.

I would like to thank David Canon, Barry Burden, Byron Shafer, Eleanor
Powell, and David Weimer for their invaluable feedback on earlier versions of this
project. Thanks also to Neil Sandberg, without whom the ne.cessary data collection
would not have been possible. Disclaimer : This anicle is based on work conducted
prior to the author-s employment with the Government Accountability Office
(GAO) and does not necessarily reflect the views or opinions of the GAO or any
individuals within the GAO.
I. But see McCarty, Poole, and Rosenthal (2006, 54-56), which argues that
NOMfNATE is unaffeLiedby changes in congressional procedures.
2. See Lee (20 I 5, 264-67) for a discussion of whether roll-callmeasures are use-
ful for discerning whether the congressional parties have actually become more
3. An anonymous reviewer raised the quei.iionof how often congressional votes
are left unrecorded. Unfortunately, comprehensive data on unrecorded votes have not
been collected. To explore this question ,-ystematically, I searched the Congressional
Record for the lir.,-t session of the 106th Congress (1999) for the phrase "the question is
on," which is typically stated by the Speaker prior to any vote. Using this method (and
excluding any insranceswhere "the question is on" appeared in a speech or other nonvot-
ing situation), I was able to identify 1,413 votes taken during the session, of which 603
were recorded, leaving 810 unrecorded votes, assun1ingthat this search method captures
all votes taken. If the lir.,-t session of the I 06th Congress is not an outlier, tllen it would
be safe to say that a considerablenumber of votes are taken witllout a roll call.
4. Rules ,iipulate that for certain items, the yeas and nays must be taken (e.g.,
certain conferencereports, motions to close proceedings, veto overrides). This makes up
only a small proportion of votes in any given Congress; for tllis analysis, those votes are
5. Data on roll-call votes are made available by David W. Rohde and the
Political Institution and Public Choice Program.
6. There are also differencesin the type of votes requested by each party. For
example, between the I 04th and II Ith Congresses, the minority party requested the
maj ority of votes on amendments,special rules, previous question motions, motions to
recommit,and most proceduralmotions (motionsto adjourn,to rise from the Comn1ittee
of the Whole, to table, etc.). The majorityparty reque;ied a majorityof the votes on final
passageof bills and conferencereports, as well as motionsto suspend the rules.
7. This is not to in1plythat electoral success is all that motivates the minority
party when asking for recorded votes. Members might also request a recorded vote in
order to change the outcomeof a voice vote. As a preliminarymethodto address whether
the desire to clumgevote outcomes is a signillcantdriver of minority-partyvote requests,
I nmdomlysampled 25 recorded votes from each Congressduringthe time period,coded
the initial outcome of the voice vote, the position of the roll-call requester,and the out-
come of the subsequent recorded vote. \Vhen the requester was on the losing side of a
roll-call vote (thus lu1vingan incentiveto change the outcon1e), the outcomechanged for
50% ( 13 out of 26) of maj ority-partyrequests, but only 9% (7 out of 80) of minority-
party requests.This preliminaryevidence shows that overturningthe outcomeof a voice
vote is rarely suocessfulfor the minority, suggesting that changing the vote outcome is
not likely to be the primary driver of minority-partyvote-reque;iingbehavior.
8. httpJ/www.politifaclcom/truth--0-meter/statements/2014 /sep/15/national-repub-
lican-congres~ i onal-conlmittee/nrcc-says-rep-jobn-barrow-<1-ga-voted-barack-obama -/
9. http://www.politifacloom/new-hampshire/statementsl2014 /oct/24/national-
guinta-portraying-them-.1s-too-extrem e/28030436
11. Electoraldata were generously sharedby Gary Jacobson.
12. The ~mall number of party switchers in the data are treated as different


Abr-.unowitz,Alan. 20 I0. The DLvappeari11g

: E11gagedCitizens, Polarization
andAmericanDemocracy.New Haven, CT: Yale UniversityPress.
Arnold, R. Douglas. 1990. The Logic of CongressionalAction. New Haven, CT: Yale
Brady, David \V., Hallrie Han, and Jeremy C. Pope. 2007. "Primary Elections and Can-
didate Ideology: Out of Step with the Primary Electornte?"LegislativeStudies
Quarterly 32: 79- 105.
Canes-\Vrone,Brnndice, David W. Brady, and John F. Cogan. 2002. "Out of Step, Out
of Office: Elector-diAccountability and House Members' Voting." American
PoliticalScienceReview 96: 127-40.
Carson, Jamie L., Michael H. Crespin, Charles J. Finnoochiaro,and David W. Rohde.
2007. "Redistricting and Party Polarization in the U.S. House of Repre-
sentatives." AmericanPoliticsResearch35: 878- 904.
Carson, Jan1ie L., Gregory Koger, Matthew J . Lebo, and Everett Young. 2010. "The
Electoral Co;is of Party Loyalty in Congress." AmericanJournalof PoliticalSci-
ence 54: 598- 616.
Clinton, Joshua, Simon Jacknum, and Douglas Rivers. 2004. "The Statistical Analysis
of Roll Call Data." AmericanPoliticalScienceReview98: 355-70 .
Clinton, Joshua D., and John Lapinski. 2008. "Laws and Roll Calls in the U.S. Con-
gress, 1891- 1994." LegLrlative StudiesQuarterly33: 5 I 1-41.
Connelly, William F., and John J. Pimey. 1994. Congress' Permanent Minority?
Republicansi11the U.S. House. Lanham, MD: Row,m and Littlefield.
Cox, Gary W., and Mathew D. McCubbins. J993. LegirlativeLeviathan: Party Govern-
ment i11the House.Berkeley: University of California Press.
Cox, Gary W., and Mathew D. McCubbins. 2005. Setting the Agenda: Responsible
Party GovernmentiJ1the Houseof Representatives . Cambridge: Cambridge Uni-
versity Press.
Fenno, Richard. I978. Home Style: House Members in Their Dirtricts. New York:
Fiorina, Morris P., Samuel J. Abrams, and Jeremy Pope. 2010. CultureWar? The Myth
of a PolarizedAmerica. New York: Longman.
Gilmour, John B. I 995. StrategicDiragreement : Stalematein AmericanPolitics. Pitts-
burgh: University of Pittsburgh Press.
Green, Matthew N. 20 I 5. UnderdogPolitics: The MinorityParty in the U.S. Houseof
Representatives.New Haven, CT: Yale University Press.
Harbridge, Laurel. 2015. ls Bipartira11ship Dead? Policy Agreement and Agenda-
Setting in the House of Representatives.Can1bridge: Cambridge University
Hibbing, John R., and Elizabeth Theiss-Morse. I 995. Congressas PublicEnemy: Pub-
lic Attitudes toward American Political Institutions. Cambridge: Cambridge
University Press.
Hibbing, John R., and Elizabeth Theiss-Morse. 2002. Stealth Democracy:Americans'
BeliejraboutHow GovernmentShould fYork.Cambridge: Cambridge University
Jessee, Stephen A., and Sean M. Theriault. 2014. "The Two Faces of Congressional
Roll-Call Voting." PartyPolitics20: 836-48 .
Jones, Charles 0. 1970. The ,1,finority
Party in Congress. Boston: Little, Brown.
Jones, David R. 20 I0. "Partisan Polarization and Congressional Accountability in
House Elections." AmericanJournalof PoliticalScience54: 323-37.
Jones, David R., and Monika L. McDem1ott. 2004. "The Responsible Party Govern-
ment Model in House and Senate Elections." American Journal of Political
Science48: 1- 12.
Jones, David R., and Monika L. McDermott. 20 I 0. Americans, Congress,and Demo-
craticResponsiveness.Ann Arbor: University of Michigan Press.
King, David C., and Richard J. Zeckhauser. 2003. "Congress ional Vote Options." Leg-
islativeStudiesQuarterly28: 387-411 .
Layton, Lyndsey. 2007. "House GOP Uses Procedural Tactic to Frustrate Democratic
Majority." WashingtonPost May 19. http: //www.washingtonpo; m/wp-dyn/
content/article/2007/05/ 18/AR200705 I 80 I697.html (Accessed July 21, 20 I 5).
Lebo, Matthew. J., Adam J. McGlynn, and Gregory Koger. 2007. "Strategic Party Gov-
ernment: Party Influence in Congress, 1789- 2000." American Jour11alof
Political Scie11ce5 1: 464-8 I.
Lee, Frances E. 2009. Beyo11dIdeology: Politics, Principles,and PartLvanvhip i11the
U.S. House.Chicago: University of Chicago Press.
Lee, Frances E. 2013. "Pres idents and Party Teams: The Politics of Debt Limits and
Executive Oversight, 200 1- 2013." Preside11tial StudiesQuarterly 43: 775- 91.
Lee, Frances E. 20 15. "How Party Polarization Affects Governance." A1111ual Reviewof
PoliticalScience 18: 261- 82.
Lee, FrdDcesE. 2016. IJ1Secure Majorities: Congressand the PerpetualCampaign.Chi-
cago: University of Chicago Press.
Levendusky, Matthew. 2009. The Partisa11 Sort: How LiberaLvBecame Democratvand
Co1lvervatives BecameRepublicans.Chicago: University of Chicago Press.
Mann, Thomas E. 1988. '"The Permanent Minority in An1erican Politics." Brooki11gs
Review 6: 33- 38 .
Mayhew, David R. 1974. Congress: The ElectoralCon11ectio11 . New Haven, CT: Yale
University Press.
McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2006. PolarizedAmerica: The
Da11ce of Ideologyand UnequalRiches.Cambridge, MA: MIT Press.
McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2009. "Does Gerrymandering
Cause Polarization?" AmericanJournalof PoliticalScience53: 666-80 .
Oppenheimer, Bruce. 2005. "Deep Red and Blue Congressional Di,'lricts: The Causes
and Consequences of Declining Party Competitiveness." In CongressReconsid-
ered, 8th ed, eds . Lawrence Dodd and Bruce Oppenheimer. \Vashinbrton, DC:
CQ Press, pp. 135- 58.
Orn,iein , Norman. 20 10. "The Motion to Recommit, Hijacked by Politics." Roll Call
May 19.
ne;:opinion (Accessed July 2 1, 20 15).
Poole, Keith T., Jeffrey Lewis, James Lo, and Royce Carroll. 201 I. "Scaling Votes with
W-NOM!NATE in R." Jour11al ofStatLvticalSoftware42 : 1- 2 1.
Poole, Keith T., and Howard Rosenthal. 1997. Congress: A Political-Economic History
of Roll Call Voting. Oxford: Oxford University Press
Ramirez, Mark D. 2009. '"The Dynamics of Partisan Conflict on Congressional Appro-
val." AmericanJournalof PoliticalScie11ce 53: 68 1- 94.
Ramirez, Mark D. 20 13. "The Policy Origins of Congressional Approval." Journal of
Politics75: 198-209.
Roberts, Jason M. 2007. "The Statistical Analysis of Roll-Call Data: A Cautionary
Tale." LegislativeStudiesQuarterly 32 : 341-60 .
Roberts, Jason M., and Steven S. Smith. 2003. "Procedural Contexts, Party Strategy,
and Conditiona l Party Voting in the U.S. House of Representatives." America/I
Journalof PoliticalScie11ce 47 : 305- 17.
Rohde, David W. 199 1. Partiesand Leaders in the PostreformHouse. Chicago: Uni-
versity of Chicago Press.
Schickler, Eric. 200 I. Disjoi11tedPluralism:lnslitt;tio11al Innovationa11dthe Develop-
me11tof the U.S. Congress. Princeton: Princeton University Press.

Schickler, Eric, and Kathryn Pearson. 2009. "Agenda Control, Majority Party Power,
and the House Committee on Rules, 1937- 52." LegLvlative
Studies Quarterly34:
Shelley, Mack C. I983. The Permane11t Majority: I11eConservativeCoalitio11 i11the
UnitedStates Co11gress. Tuscaloosa: Universityof Alabania Press.
Smith, Steven S. 1989.Call to Order: Floor Politicsi11the Ho11se and Senate.Washing-
ton, DC: Brookings Institution.
Smith, Steven S. 2007. Party ltif/uencein Congress.Cambridge:Cambridge University
Smith, Steven S., lan O,irander, and Chri,iopher M. Pope. 2013. "Majority Party Power
and Procedural Motions in the U.S. Senate." legislative Studies Quarterly38:
205- 36.
Sundqui,t, James L. I988. "Needed: A Political Theory for the New Era of Coalition
QuarterlyI 03: 6 I 3- 35.
Government in the United States." Politi.calScie11Ce
Theriault, Sean M. 2008. Party Polarizationi11Congress.Cambridge: Can1bridgeUni-
versity Press.
Theriault, Sean M. 2013. The GiJ1grichSenators: The Roots of Partisan Warfarei11
Co11gress. Oxford: Oxford University Press.
Wooldridge, Jeffrey M. 2005. Eco11ometric A11alysLv of Cross-Sectionand Panel Data.
Cambridge, MA: MJT Press.
Zelizer, Julian E. 2004. On CapitolHill: TheStruggleto ReformCo11gress and itv Con-
sequences,1948-2000.Cambridge: Cambridge University Press.