Embed Size (px)
Citation preview
A process-populism dimension in the US public?Evidence from two surveys in 2016
Jack Santucci∗
April 9, 2018Original: April 2, 2018
Abstract
This paper finds a two-dimensional structure in American mass opin-ion as of 2016. This structure appears in two different data sets for twodifferent question batteries on who should rule and how. The first data setcomprises 1,000 respondents and includes the famous stealth-democracyquestions. The second comprises 8,000 and includes questions on per-ceptions of a rigged political and economic system. While some of theseissues also help define a traditional, liberal-conservative ideology, othersare entirely orthogonal, suggestive of a separate dimension. Further, ex-ploratory analysis of other items defining this dimension tap both processand policy outputs (e.g., trade, social services, and social tolerance). Inboth data sets, higher scores on this dimension are associated with greaterwealth. Lower scores in the larger set are associated with low educationalattainment and identifying as “other race.” It will be interesting to learnmore about this structure, how it will evolve, and what it ends up meaningfor American party politics.
∗Corresponding email: [email protected]. Views expressed here are solely those ofthe author.
1
1 Introduction
A central concern in democratic politics is how people organize themselves (or areorganized) for collective action. One mode of organization is the political party.Another is the nature of one’s ideology: liberal, conservative, or somewhere inbetween. Most political scientists would agree that, for the past 40-50 years,party and ideology have been losing their separateness as organizing principles:ever fewer liberal Republicans, ever fewer conservative Democrats.
Yet there are reasons to think that 2016 may be different. At the level of thepresidential nominating contest, Donald Trump challenged many conservativepositions (e.g., on the goodness of trade and fiscal restraint) – yet won the Re-publican nomination. In the Democratic camp, Hillary Clinton received criticismfor her stance on “identity politics” and a perception of her closeness to financialelites. It is possible that the intra-party battles were symptoms of a broader shiftin mass ideological attachment, regardless of whether the candidates actuallyheld those stances. According to one new paper, in fact, responses to a repeatset of 31 survey questions had a measurably two-dimensional structure in 2016,which they did not have in 2012.1
Beginning with two sets of survey questions about who should rule and how,this paper tests for the existence of a “process-populism” dimension in Americanpolitics. Using two recent, national samples of public opinion, I measure the idealpoints of 1,000 and 8,000 respondents, respectively. Ideal-point analysis lets ussee how issues are configured.
Two dimensions emerge from each data set. The first is a traditional, liberal-conservative divide, which maps well onto presidential voting. The second di-mension indeed picks up the survey questions on who should rule and how.Exploratory analysis of this dimension also uncovers social intolerance and frus-tration with several policy outputs.
The paper proceeds as follows. Section 2 briefly reviews some of the mostimportant literature on mass political opinion. Section 3 asks what some havesaid about the nature of the second dimension in American politics – should one infact exist. Section 4 discusses the two main data sets: one a team CCES modulewith five “stealth-democracy” questions, the other an 8,000-person survey thatincluded six “rigged-system” questions. Both were done in August 2016. Section5 describes the scaling procedure and states hypotheses about how these itemsshould manifest in ideological space. Section 6 presents results for the question
1. The author and a coauthor, unpublished.
2
batteries of deductive interest. Section 6 investigates the people who define thesecond dimension. Section 7 takes an exploratory look at other issues that definethe dimension of interest. The final part concludes.
2 The study of mass ideology
As social scientists use it, ideology is a construct that describes what politicalissues and attitudes tend to come in packages. People can be more or less ideo-logically constrained, e.g., holding varying blends of the liberal and conservativepositions that we use to measure their belief systems. Thus conceived, the mod-ern study of ideology began in the wake of World War II. This was partly inreaction to the rise of fascism, partly as an outgrowth of early efforts to under-stand campaigns (Lazarsfeld, Berelson, and Gaudet 1944; Berelson, Lazarsfeld,and McPhee 1954; Converse 1964). One of the earliest results was that mostpeople’s beliefs are not very constrained (Campbell et al. 1960).
More recent studies of ideology have sought to disaggregate the public intomore and less politically engaged segments. The basic thrust of this literatureis consistent with what Campbell et al. (1960) found. One finds low constraintwhen one takes either (a) a bird’s-eye view of the entire public or (b) focuses onthe lesser-engaged masses. Those who follow politics most closely, possibly byworking or volunteering in it, tend to be more constrained, in that their attitudeon one issue helps predict their attitude on another (Abramowitz and Saunders2008; Fiorina 2010; Kinder and Kalmoe 2017).
Psychometric scaling of policy choice is another way to study ideology. Ratherthan examine attitudes in one-by-one fashion, possibly as predictors of each otheror some outcome (e.g., vote choice), the ideal-point approach projects actors intolow-dimensional space. One advantage here is that such spaces can be studiedtheoretically (Hinich and Munger 1994).
Another benefit of the psychometric approach is that we can study a policyspace empirically, seeing how people and issues together define it. Why botherdoing this? If we accept that ideologies are difficult-to-measure systems of belief,the items we use to tap them will do those jobs imperfectly. The thing we wantto know about is latent. By estimating ideal points, we literally can see howthe various measures are configured. This, in turn, may tell us something aboutpeople’s ideologies. At the very least, it tells us about issues that do not defineparty competition (i.e., do not fall along on the liberal-conservative axis).
The most famous empirical application of ideal-point analysis has been in the
3
study of Congress (Poole and Rosenthal 1997). Other studies increasingly lookat many other actors – too many to note here – including: state legislators (Shorand McCarty 2011; Masket and Shor 2015), Supreme Court justices (Martin andQuinn 2002; Bailey 2007), U.S. Presidents (Bailey and Chang 2001), city coun-cilors (Burnett 2017; Bucchianeri 2017; Santucci 2018), pundits (Noel 2013), andthe members of many foreign assemblies (Jenkins 2000; Rosenthal and Voeten2004; Hix, Noury, and Roland 2006; Dewan and Spirling 2011; Kellermann 2012;Herrmann and Sieberer 2018).
Several studies over the past 20 years have applied ideal-point analysis tothe public as well. The typical approach is to focus on a single dimension (Tau-sanovitch and Warshaw 2013; Hill and Tausanovitch 2015; Jessee 2016; Sideset al. 2018), but others have investigated a second dimension.
3 The case for a second dimension
Say an ideal-point routine turned up a second dimension. What might it repre-sent? Three possibilities are apparent from the literature.
The first possibility taps the familiar dichotomy between social and economicissues. According to several sets of scaling results, economics appear on thefirst dimension, while social issues define the second (Miller and Schofield 2003;Schofield, Miller, and Martin 2003; Miller and Schofield 2008). Treier and Hilly-gus (2009) use the 2000 American National Election Study to single out someof these social issues: capital punishment, abortion, women’s role in society, andgay rights at least.
An everyday interpretation of the second, “social” dimension might be: mostlyrace. This is how many of us teach it in undergraduate classes – and not withoutsome precedent. According to Poole and Rosenthal (1997: 45-56), the seconddimension was most apparent in Congressional politics during what they call the“three-party system of the mid-twentieth century: the period from the late NewDeal into the mid-1970s.” Readers will recall that the Southern faction of theDemocratic Party regularly opposed civil-rights legislation during this period. Byvoting with the Republican Party on those bills, then the Northern Democrats onpurportedly economic bills, Southern Democrats created what appear as threedistinct blocs in a two-dimensional map of the House, with that group at the topand center.
There are reasons to question a purely racial or even social interpretation ofthe (possibly periodic) second dimension in American politics. The first comes
4
from Poole and Rosenthal (1997) themselves, noting that the second dimensionpicked up specifically regional issues and trade policy during the 1890s, i.e., justthose issues that might pit the South and West against what Burnham (1982)called the Northeastern “metropole.” Aldrich and Griffin (2018: 201) go evenfurther: “Slavery is obvious, but the threat of the Populist Party in the Southwas to unite working-class blacks and whites against the middle- and upper-classwhites, and that would be done on economic issues such as bimetallism.” Poole’smost recent interpretation holds that the second dimension picks up conflictbetween each party’s “insider” and “outsider” factions (Poole, Rosenthal, andHare 2015). Finally, it is worth noting that the coalition to pass the Taft-HartleyAct in 1947 over President Truman’s veto – an explicitly economic bill that dealtwith labor unions – was precisely the sort of coalition that might oppose a civil-rights bill: Southern Democrats with the Republicans, the former toward the topof 2-D space. Something at least partly economic must be happening on thesecond, ostensibly social dimension in American politics. And it may be that thisdimension appears only periodically, especially if we just look at Congress.
If we set aside the scaling literature for a moment, we get hints of a differentand potentially fruitful interpretation. Fiorina (2009), for one, has has insistedon a qualitative difference between the “political class” and masses. The formerreflects more attitude constraint, while the public, he says, is moderate andpragmatic. Others have shown group-level variation in support for norms andpolicies associated with liberal democracy and limited government (Drutman,Diamond, and Goldman 2017).
The second possibility might begin with Lipset (1959a), who once posited theexistence of two main dimensions in democratic politics: economic liberalism,then one dealing with “democratic norms.” This second dimension, a sort ofpatience-and-tolerance dimension, appears in his earlier, more famous essay asfollows: “A belief in secular reformist gradualism can only be the ideology ofa relatively well-to-do lower class” (Lipset 1959b: 83). Working-class people, heargued in the later essay, do not fundamentally care about patience-and-toleranceissues because economics are their main concern. Whether they manifest (oraccept) positions of tolerance and gradualism would depend on two things: (1)the issue leadership of their respective parties, (2) the success of which wouldvary with perceptions that economic needs are met. “It is probable,” he writes,“that organized social democracy not only supports civil liberties but influences itssupporters in the same direction” (Lipset 1959a: 500). Given the waning forceof welfare-state policies in recent years (Bonica et al. 2013; Page and Gilens2017), it is possible that the now-apparent second dimension picks up attitudes
5
toward the state of liberal-democratic politics. Another way to say this might bepopulism.
Subsequent work in the populism vein has sought to distinguish it from na-tivism. Both ideologies, according to Mudde and Rovira Kaltwasser (2017), growout of frustration with perceptions that a party system has become unresponsive.Both forms accordingly seek to divide society into uncorrupted masses and a cor-rupt elite. Populism and nativism offer two ways to do this. The first combinesits appeal with elements of socialism. The second uses nationalism to definecertain people out of the political community. According to Mudde (2017), bothforms are hostile to political liberalism and require nimble attention from theliberal-democratic establishment.
A populism (and/or nativism) thesis stands in partial contrast to a third pos-sibility: that attitudes toward democratic processes are independent of concretepolicy preferences. The idea here is that, while citizens generally do not desiremore participation, this attitude waxes and wanes with perceptions of corruption(Hibbing and Theiss-Morse 2002; Dyck and Baldassare 2009, 2012).
4 Data
Two question batteries in two new data sets let us test the hypothesis that aprocess-populism dimension cross-cuts the liberal-conservative divide. I will saymore about its substantive content in two exploratory sections at the end of thispaper.
The first data set is the Reed College module of the 2016 Cooperative Con-gressional Election Study (CCES), the relevant part of which was done in August2016. The CCES regularly includes a range of questions about attitudes to-ward social, economic, environmental, and other policies. Under the directionof Paul Gronke and in partnership with Natalie Adona of the Democracy Fund,the Reed module asked several additional questions of 1,000 randomly sampledrespondents in the wider CCES. These extra questions covered attitudes towardvarious levels of government, election administrators, various election reforms,and electoral integrity issues.
The Reed module also asked five “stealth-democracy” questions on attitudestoward who should hold political power and how they should use it (Hibbing andTheiss-Morse 2002). These agree-disagree statements were as follows:
1. Elected officials would help the country more if they would stop talkingand just take action on important problems.
6
2. What people call ‘compromise’ in politics is really just selling out on one’sprinciples.
3. Our government would run better if decisions were left up to successfulbusiness people.
4. Our government would run better if decisions were left up to non-elected,independent experts rather than politicians or the people.
5. When people argue about political issues, it makes me feel uncomfortable.
Within the stealth-democracy battery, two questions seem especially rele-vant: questions 3 and 4 on government by “successful business people” and“non-elected, independent experts.” Table 1 shows that just about a quarterof Americans agreed with those statements, with fairly even proportions in eachpresidential camp.
Clinton (D) Other Trump (R) OverallJust take action 76.3 70.0 80.5 75.9Compromise is selling out 23.4 48.6 53.2 40.0Business run gov’t 7.7 14.3 51.4 25.3Unelected experts 24.9 32.9 33.9 30.6Arguing uncomfortable 31.3 19.7 30.6 33.1
Table 1: Percentage agreeing with each statement, both overall and in eachcamp, CCES-Reed College stealth-democracy battery.
The second data set is the 2016 Views of the Electorate Research (VOTER)survey, commissioned by the Voter Study Group (VSG), and also done by YouGovin August 2016. The VSG is a set of political scientists and public-opinionresearchers affiliated with the Democracy Fund. One big goal of their 8,000-respondent VOTER survey was to better understand the issue clusters that definethe factions in each party.
Within the VSG data, six “rigged-system” questions seem like they may tappopulism and/or frustration in the public. The agree-disagree statements were:
1. Elections today don’t matter; things stay the same no matter who we votein.
2. America is a fair society where everyone has the opportunity to get ahead.
7
3. Our economic system is biased in favor of the wealthiest Americans.
4. You can’t believe much of what you hear from the mainstream media.
5. People like me don’t have any say in what the government does.
6. Elites in this country don’t understand the problems I am facing.
Table 2 presents the percentage of respondents who agreed with each state-ment above, as well as that percentage in each general-election bloc. Several ofthese questions appear to reflect partisanship. Trump supporters overwhelminglyexpressed agreement with the fundamental fairness of American society and mis-trust of the mainstream media, while Clinton supporters overwhelmingly agreedthat the “economic system is biased.” At the same time, though, substantialnumbers in the opposing parties also agreed with these statements. Finally,nearly 83 percent of the sample said that “elites don’t understand.” Each ofthese questions seems like it might help define a second dimension.
Clinton (D) Other Trump (R) OverallElections don’t matter 35.0 53.3 30.1 35.1
America is fair 30.3 52.6 82.8 55.9Economy is biased 95.5 76.3 47.0 72.0
Can’t believe the media 43.7 76.4 93.2 69.2Ppl like me have no say 59.5 66.2 53.7 57.8Elites don’t understand 84.0 84.8 82.5 83.4
Table 2: Percentage agreeing with each statement, both overall and in eachcamp, Voter Study Group rigged-system battery.
Finally, both the Reed and VSG data sets include standard demographicvariables. These can be used to see whether certain kinds of people systematicallyhave certain kinds of scores on the second dimension.
5 Methods and hypotheses
I use ideal-point estimation to project both surveys’ respondents and policy ques-tions into low-dimensional space. All such approaches make the following as-sumptions: (1) every voter or person has an ideal point in policy space, (2)
8
policy alternatives have locations in that space, and (3) a voter chooses thepolicy or candidate that is closest to their ideal point.
Following Hare (2015), I use the non-parametric, optimal-classification (OC)method developed by Keith Poole (2000, 2005). Versus NOMINATE and othercommon approaches (e.g., Bayesian item-response per Jackman (2001)), OCdoes not impose a functional form on the structure of survey responses. Nordoes it rest on any assumptions about the utility function that respondents usewhen answering a question. OC simply finds the best possible ordering of sur-vey respondents and questions, given the number of dimensions supplied, forclassifying answers correctly. In two (or more) dimensions, this model producesrespondents’ ideal points, then a cutting line (or plane) for each survey question.
Preparing the data involved two steps. The first was selecting a set of ques-tions to analyze. In keeping with the micro-foundations of the ideal-point ap-proach, I selected only those questions with some policy component. In all, thisamounted to 97 questions from the Reed-CCES data, then 60 questions from theVSG. Finally, if responses were not binary, I collapsed them to make them so. Arespondent who either “agrees” or “somewhat agrees” is coded as a “yea” or 1.Those who disagree at all are coded as “nays” or 0s. Finally, those who “neitheragree nor disagree” are coded as not voting (9), with missing values preservedthroughout. Collapsing values in this way produces a roll-call-style matrix.2
For each data set, I estimated ideal points in one, two, and three dimensions.Overall, it appears that two dimensions are sufficient to explain answers to mostquestions. Table 3 gives fit statistics for each of the routines. On the left, we seethe percentage of responses correctly classified (PCC). Aggregate proportionalreduction in error (APRE) appears on the right. Both are standard ways to assessthe fit of ideal-point models. PCC is straightforward. APRE weights algorithmperformance by ease of the classification task, penalizing those questions wherevery few people take the minority position (i.e., where classification is easy).Table 3 shows marked classification improvement in both figures when movingfrom one to two dimensions and much less improvement when moving from twoto three dimensions.3
2. For three ranking questions on the CCES, I treated the top-ranked item as a “yea” anditems ranked 2 or lower as “nay.” For two 0-100 thermometer items, I treated ratings greaterthan 50 as “yea,” less than 50 as “nay,” and ratings of 50 as abstentions. Hare, Liu, andLupton (2018) have since developed an OC method that does not require binary responses. Iwill use this in future work.
3. For the Reed-CCES data, a third dimension does improve classification more than itdoes for the VSG data – albeit marginally so. Computing marginal proportional reduction in
9
Pct. CC APREReed-CCES, 1 dimension 77.9 0.319Reed-CCES, 2 dimensions 81.6 0.434Reed-CCES, 3 dimensions 83.6 0.496VSG, 1 dimension 85.2 0.469VSG, 2 dimensions 88.4 0.581VSG, 3 dimensions 89.6 0.626
Table 3: Overall optimal-classification model fits: percent of answers correctlyclassified and aggregate proportional reduction in error.
Finally, tables 4 and 5 give question-level fit as percentage of responses cor-rectly classified for the stealth-democracy and rigged-system batteries, respec-tively. All but one question (discomfort with political argument) registers anacceptable fit.
Reed question Pct. CCJust take action 93.2Compromise is selling out 82.9Business run gov’t 85.0Unelected experts 72.5Arguing uncomfortable 69.6
Table 4: Reed stealth-democracy questions, percent correct classification in twodimensions.
In testing for the existence of a process (or populist-elite) dimension, weare mainly interested in the cutting lines for the respective survey questions.Whether a cutting line defines this or that dimension depends on its angle. Ifthe principal dimension is left-right conservatism, issues defining that dimensionwill have vertical cutting lines (e.g., around 90 degrees). In other words, thoseissues will separate conventional left from right. Second-dimension issues willseparate top from bottom; their cutting lines will be closer to 0 or 180 degrees.
error (Roberts, Smith, and Haptonstahl 2015) for the stealth-democracy questions, however,does not reveal much improvement for that group of questions. This may be due to therelatively smaller data matrix, which is known to depress APRE (Bucchianeri 2017), or due tothe presence of many unconventional questions (e.g., whether ranked-choice voting will reducecorruption.) For these reasons, I present results from the two-dimension model of the Reeddata.
10
VSG question Pct. CCElections don’t matter 80.9America is fair 87.4Economy is biased 90.4Can’t believe the media 86.3Ppl like me have no say 80.9Elites don’t understand 87.7
Table 5: VSG rigged-system questions, percent correct classification in two di-mensions.
Cutting lines also may run diagonally. When this is the case, they capture bitsof both dimensions, helping to define liberal-conservative as well as process (orpopulist-elite).
If the second dimension represents what we expect, several things will betrue about the cutting lines and/or angles for the stealth-democracy and rigged-system questions:
1. Traditional, left-right issues will define the first dimension. Most Democratswill appear on the left of two-dimensional space, and most Republicans willappear on its right.
2. The lion’s share of issues will have vertically-running angles, continuing todefine a liberal-conservative dimension.
3. Cutting lines for the batteries of interest will run horizontally or at leastdiagonally, with angles around or less than 45 degrees, then around orgreater than 135 degrees.
6 Results and discussion
Figures 1 and 2 present the spatial maps. Respondents’ ideal points are given as“D” if they said they voted for Hillary Clinton, “R” if Donald Trump, and “O”if they voted for a third-party candidate or did not vote at all. Cutting lines forthe stealth and rigged questions appear in each plot. In the Reed-CCES plot, tofacilitate interpretation, I also include cutting lines for two other questions:
• White people in the U.S. have certain advantages because of the color oftheir skin. (This is a standard CCES question. 53.4 percent agreed.)
11
• How much of the time do you think you can trust the federal governmentin Washington to do what is right? (This was from the Reed module. 15.4percent said “most of the time” or “always.”)
Both plots confirm the existence of a standard, liberal-conservative dimensionthat runs from left to right. Several cutting lines in the Reed data support this in-terpretation. The one for white-skin advantage separates virtually all Democratsfrom most Republicans. In the Reed data, and even though they were hypothe-sized to tap a second dimension, two stealth-democracy questions at least helpdefine liberalism and conservatism. One of them, for successful business peo-ple running the government, separates most Democrats from most Republicans– notably with a very small number of Democrats on the “agree” side, at thetop of the space, then a not-insignificant number of Republicans on its oppo-site. Similarly, the cutting line for “compromise is really just selling out” appearsto have become a liberal-conservative issue. Many would agree that refusal tocompromise has become a feature of conservatism (Chait 2011).
Turning to the VSG data, some of the rigged-system angles also run diago-nally. Those that stand out most are: “Our economic system is biased in favor ofthe wealthiest Americans” and “You can’t believe much of what you hear fromthe mainstream media.” The former separates many Democrats from many Re-publicans, and the latter does vice-versa. At the same time, a substantial set ofopposite-party persons holds each of these views: respondents toward the middleand bottom of the space.
Figure 3 provides a more direct test of the liberal-conservative hypothesis:that the lion’s share of angles will hover around 90 degrees. This figure also teststhe hypothesis that the process-populism questions will be clearly off-dimension.These two histograms represent the distribution of estimated cutting lines in eachdata set. Small triangles at the bottoms of the plots represent the angles for thestealth-democracy and rigged-system questions. In the Reed data, the modalcutting line is just around 90 degrees, with three of the five stealth-democracyquestions having angles of about 135 degrees or more. In the VSG data, all sixquestions have clearly diagonal cutting lines, if not almost perfectly horizontal.
Some reflection on the questions themselves and their positions in the spacewill help us begin to interpret the meaning of this dimension. Beginning withthe Reed data, trust in the federal government is limited to those at the topof the space. So is preference for government by “successful business people,”although fewer Democrats express this view. A bipartisan group at the top ofthe space reports discomfort with political argument and preference for rule by
12
Ideal points:CCES−Reed College
1st dimension
2nd
dim
ensi
on
R
D
D
D
R
R
DD
D D
R
R
R
D
R
R
R
D
O
D
R
D RO
R
RD
R
D
R
D
O
R
D
R
RD
R
D
R
D R
R
D
D
R
R
O
R
D
D
D
R
O
R
R
R
D
D
DR
R
D
O
R
R
R
R
D
R
R
R
R
O
R R
R
D
R
D
R
D
R
O
D
D
D
DD
R
R
D
R
R
O
R
D
D
R
D
D
R
R
R
D
D
R
R
D
R
D
R
D
O
O
D
D
D
R
R
R
R
O
R
R
R
R
R
R
D
R
D
D
D
R
D
R
R
O
R
D
R
D
O
R
R
R
R
D
R
D
R
D
R
R
D
O
R
RD
R
R
R
R
R
R
R
D
O
R
R
D
R
D
R
R
RD
D
O
R
R
D
R
R
R
R
R
R
D
DDR
R
D D RD
R
D
R
R
D
R
D
R
R
D
D
O
DR
D
D
D
R
R
R
D
D
D
D
D
RD
D
D
DD
R
D
D
DR
D
R
D
RR
R
R
D
R
D
R
D R
R
D
D
R
R
D O
R
R
D
D
D
R
D
D
D
R
R
O
D
O
D
R
R
R
D
D
R
R
D
R
D
R O
O
D
D RDR
D
O
O
D
R
R
D
D
R
R
R
D
O
DD
O
R
R
R
R
R
D
R
D
D
RD
D
R
R
R
D
R
O
D
D
R
R
D
D
R
D
D
D
R
D
R
O
D
O
D
D
DD D
DD
R
R
O
R
D
O
O
D
D
D
DR
D
D
D
D
D
D
DD
D
D
R
R
D
R
RD
D
D
D
O
D
D
R
O
D
D
D
D
D
D
D
O
R
D
R
D
R
D
DD
RD
D
D
R
D
D
D
DD
D
D
D
DD
D
R
O
D
R
R
R
R
O
D
D
R
O
R
DO
D
D
D
R
O
R
O
R
D
D
D
R
DD
R
D
D
D
R
D
R
OD
D
D
D
D
D
R
D
R
D
D
R
R
R
R
D
R
R
D
D
DD
D
R
RR
R
R
D
D
D R
R
D
D
D
D
D
R
R
D
D
R
R
D
D
D
O
D
DO
D
D
D
RR R
D
D
O
D
D
R
D
D
D
DD
D
D
D
O
D
R
D
R
D
R
D
D
O
D
R
D
R
D
D
O
D
D
D
D
R
R
D
O
R
DDO
D
D
R
D
D
D
D
D
R
D
D
D
R
D
O
R
R
O
DD
R
D
D
D
D
O
D
D
R
DDD
RO
D
R
R
R
OR
R
D
D
R
O
D
R
D
RR
O
R
D
O
R
D
O
DR
RD
R
O
R
D
D
D
D
R
DD
D D
RD
D
O
R
D
R
O
D
R
R
D
D
D
D
R
R
D
D
O
R
O
R
R
D
D
D
RR
R
D
D
R
R
R
R
D
D
D
O
D
R
O
D
D
DR
R
D
R
D
R
RD
R
R
RD
R
R
R
D
D
R
OR
D
R
D
RD
D
D
R
R
R
D
R
D
O
D
R
R
D
D
D
R
D
R
R
D
DO
D
D
D
R
D
D
D
R
D
R
D
D
D
R
D
O
D
−1.0 −0.5 0.0 0.5 1.0
−1.0
−0.5
0.0
0.5
1.0
Takeaction
No compromiseBiz. run gov't
Unelectedexperts
Pol. argue.uncomfortable
Whites have advantage
Trust fedto do right
Figure 1: Ideal points and stealth-democracy cutting lines (with white-peopleadvantage and trust in federal government), CCES-Reed College.
13
Ideal points:Voter Study Group
1st dimension
2nd
dim
ensi
on
D
RDO
RD
D
R
D
D D
D
R
D
D
D
D
R
D
D
R
D
R
O
R
R
O
D
R
DD RR
R
D
RR
RD
D
D
R
R
R
R
R
DD
R
R
D
D
D
R
R
R
D
D
D
D
D
D
R
R
D
D
D
D
DD
D
D
D
R
R
D
D
D
D
RD
R
D
D
D
RR
D
D
O
D
R
R
D
R
R
DD
D
O
D
D
D
RO R
D
O
D
R
D
D
D
D
R
R
R
O
O
D
O
D
DD
R
R
D
D
O
DD
R
D
D
DRR
O
O
D
R
R
RD
R
D
R
D
DD
R
D
R
D
R
O
DR
D
D
R
D
R
D
O
RD
R
RO
R
RD
R
DD
D
R
D
D RO
RD
D D
R
O
R
O
D
R
O
O
D
D
D
R
D
D
DR
D
D
D
D
R
O
D
R
D D
R
D
DDD
R
D
DD
O
D
DD
O
R
R
R
R
R
D
R
R
DR
R
R
R
RRR
D
R
D
RD
R
D
R
R
RR
R
R
RR
D
O
R
D
R
R
R
R
R
R
R
D R
R
R
R
R
R
D
R
D
RD
R
R
R
R
D
R
O
D
R
R
R R
DR
R
D
R
R
DR
R
D
D R R
RR
RD
R
O
R
R
R
R
RD
O
RR
D R
R
D
R
DR
R
R
O
R
R
O
R
R
R
R
R
R
DD
D
D
R
D
R
RD
R
R
D
O
DR
DR
R
R
D
O
R
R
D D
RO
D
D R
R
R
R
O
O
R
O
DR
R
R R
R
R
O
O
R
R
O
O
R
RD
R
DR
D
R
R
D D
D
R
RR
R
R
R
R
R
R
R
R
D
O
R
R
R
R
D
D
D
R
R
RD
D
R
RR
D
D
D
R
R
R
R
R
R
R
D
R
RR
R
RR
R
D
D
O
R
R
D
DD
R
RD
R
R
R
RD
D
R
D
RR
R
RDD
R
R
R
D
D
R
D R
R
O
O
O
D
R
R
O
R
D R
R
D
R
R
R
DR
D
OR
D
O
D
OD R
D
D
D
D
O
O
R
R
D
D
R
D
R
R
R
D
D
D
R
D
R
O
R
R
D
O
RR
R
R
R
D
D
DR
D
D D
DR
D
R
R
DD
R
D
D
D
R
D
DD
R
O
D
D
D
D
RR
D
R
D
R
D
D
R
D DR
R
DD
D
R
R
R
O
R
R
R
R
D
D
R
R
D D
O
D
DR
D
R
R
O D
D
R
D
R
R
DD
R
R
RD
RR
DD
D
R
D
D
D
D
D R
D
R
D
OD
R
R
R
D
R
DD
R
D
D
R
D
D
D
D
R
R
DR
D
D
O
D D DD
R
O
D
D
O
DD
R
R
R
DD
D
D
D
D
R
R
D
R
R
R
R
D
D
O
R
D
D
D
R
D
D
O
R
RD
R
DD
R
O
R
D
D
D
D
D
R
R
R
D
DO
O
R
RD
R
R
D
RR
R
D
D
R
R
RR
D
D
R
D
R
R
R
R
RO
R
D
D
R
D
R
D
D
O
RD
D
D
D
R R
D
D
O
D
R
R
R
D
DD
O
R
O
R
R
R
R
R
R
R
R
O
D
R
DD
R
R
D
O
O
R
D
DD
O
R
R
R
D
O
R
O
DDOD
R
O
R
R
RDD
D
D R
D
RDD
D
R
D
O
D
DR
D
D
D
D
R
D
R
R
R R
R
R
D
RDD R
D
DDD
O
R
D
D
R
O
D
R
D
DR R
RD
O
D
D
DD
DD
R
R
R
D
OR
D
R
R
DR
R
D
O
DR
D
O
O
R
R
R
D
R
RD
D
D
RD
D
O
D
DD
O
R
R
D
D
RR
D
R
D
R
O
DR
R
D
R
D
RD
D
D
O
R
D
R
DR
R
R
RR
D
RR
R
D
O
R
RD
D
D
D
D
DD
R
D
O
D
D
D
R
R
D
R
D
R
R
D
R
R
OD
D
D
D
R
D D
D
D
D
R
D
R
D
D
RO
D
R
D
D
R
R
D
R
D
R
O
R
D
D
D
DO
R
R
D
O
D
R
R
D
D
D
R
D
R
R
R
R
R
D
D
O
D R
D
RD
D
D
D
R
R
R
R
D
D
D
R
D
R
D
DR
D
R
O
DD
D
RD
R
R
R
D
R
D
D
D ODD
OD
D
O
DD
DO
D
D
D
D R
D
D
D
O
R
DO
D
D
D
D
D
O D
D
O
D
D
R
R
O
D
D
D
R
D
R
D
D
R
D
O
DR
D
RD
R
D
O
D
R
R
D
D
R
R
DD
DR
R
R
O
D
DD O
RD
O
D
RD
D
OD
R
R
D
D
D
R
D D
R
D
DD
D
O
D
R
RD
D
D
D D
D
D
R
R
D
R
D
DD
D
D D
D
DD
D
R
D
D
O
D
D
D
R
D D D
R
D
D D
R
R
D
D
D
D
D DO
R
D
D
DD
D
D
R
DD
D
D
DD
O
D
R
D
RD
R
D
O
RD
R
D
O
DD
DR
D
R
D
R
D
D
D R
R
D
R
R
R
OD
DD R
DD
D
O R
O
R
D
D
R
O
D
O
DD
D D
D
D
R
RR
D
D
D
DR
R
D
DR
D
RR
D
R
D
D
DDD
R
D
D
R
R
D
D
R
R
R
R
D
R
R
RR
D
D
R
RDD
D
RD
D
D
D
RR
R
R
D
R
D
D
D
D
D
D
D
R
O
D
O
D
D
D
R
O
R
R
O
DD
D D
O
R
R
D
DOD
D
DD
D
O O
R
D
R
O
R
O
D
D
D
R
R
D
R
D
D
D
D
O
D
R
D
R
D
D
D
D
R
R
D
R
D
R
D
O
R
DD
D
D
D
D
D
D
D
D
D
D
D
D
O
D
D
R
D
O
R
D
R
R
R
D
D
D
RD
DD
R
D
R
D
D
D
D
O
D
R
O
R
D O
O
RD
R
R
D
D
D
D D
O
D D
R
D
D
D
D
D R
DR
D
R
R
R
D
R
R
R
R
RDDD
DR
D
D
D
O
RR
O
R
O
O
RD
RD
D
R
R
D
DR
RRD
O
D
D
O
D D
D
RR
R
R
R
OD
R
RD
R
O
R
O
DDO
D
O
RD
D
R
D
O
R
R
D
D
RD
D
R
R
D
D
D
R
R
R
R
O
R
DD
R
D
R
D
D
R
O
D
R
RD
OD
D
D
R
R
D
D
D
D
D
D
D
D
D
DD
R
R
R
D R
O
D
D
D
D
R
D
R
O
D
D
D
D
R
D R
D
R
D
D
O
R
R
R
D
R
D
R
O
O
D
R
D D
D
D
D
O
R
D
D
D
D
R
O
D
D
DD
R
D
D
R
D
DR
D
D
R RDR
DD
R
D
R
O
R
D
R
R
D
D
D
D
D
R
R
O
D
D
D
D
R
D
D
O
R
R
D
RR
RD
D
O
RR
R
RD
D
D
RD R
D
R
D R
RO
O
D
R
D
R
R
D
D
R
R R
R
D R
D
D D
R
D
D
D
R
R
D
RR
D
DR
D
D
D
D
D
RR
R
O
D
O
R
R
R
D ORD
RD
O
O
ORR
O
D D
D
O
R
O
R
D
R
R
D
D
D
DD
D
D
D
R
R
D
DD
DDO
RD
D
RR
D
R
D
R
D
D
D
D
DO
D
R
D
D
D
D
D
R
D
D
D
R
D
D
D
DD
D
R
D
D
R
R
D
O
O
D
R
R
R
D
D
R
R
D
D
R
R
R
O
R
O
OD
R
R
R
D
R
O
O
O
D
D
D
DD R
D
D
D
D
R
O
R
R
D
D
D
D
R
R
D
R
D D
D
D OR
D
R
R
D
R
DD
D
R
D
D
D
R
D
D
R
R
RO
DR
D
R
D
RR
D R
D
R
R
RR
R
O
R
D
D
R
D
D
D
D
D
R
RD
RR
D
D
D
O
D
D
D
R
R
R
D
D
R
D
DD
D
D
D
D
DD
R
D
O
D
RD
DD
D
DO
D
D
D
D R
D
DD
D
D
D
D
D
R
D
D
R
R
R
R
DD DD
D
R
R
O
R
DD
DR
R
R
DD R
R
D
D
O
R
R
R
D
O
R
D
D
R
R
RR
D
RD
D
R
D
R
R
R
D
R
D
R
R
O
R
D
R
R
D
R
D
R
D
D
D
RO
O
D
DD
D
D
RO
D
R
R
R
R
D
D D
D
D
O
D
DDD
RR
D
D
R
DD R
D
R
O
D
D
D
O
D
D
D
DD
R
D
DD
OO
D
D
O
O
O
D
R
R
D
DD
O
D
R
O
R
R
R
R R
O
O
D
D
R
R
O
D
R
D
R
R
D
D
O
D
D
D
D
R
O
OR
R
D
R
DD
R
R
O
R
D RR
R
O
R
R
D
R
O
R
DD
D
R
O
D
R
R
D D
D
R
R
DR
D
D
D
R
O
D
D
R
D DO
D
R
DD
R
O
D
D
DR
R
D
D
R
D
R
R
D
RD
D
R
R
R
R
O
DD
D
D
R
O
D
R
R
DD
D
RD
R
D
D
R
R
D
D
R
R
R
D
R
R
R
R
R
DR
D
R
R
R
R
D
O
R
D
D
D
D
R
D
DR
R
RD
D
D
D
O
D
D
R
D
RRR
R
D
D
D
O
O
D
D
D
DD
R
O
O
O
R
R
DD
D
D R
D
O
D
D
D
D
O
R
D
D
R
D
DO
D
D
RR
R
O
D
O
R
R
D
D
O
O
RD
D
DO
DR
D
R
D
R
D
D
D
O
D
DD
R
R
D
D
R
D
O
R
R
O
O
DR
R
DR
RR
D
R
R
D
R
D
D
DR
D R
D
D
D
R
D
R
R
O
D
R
D
OO
D
D
D
R
R
R
D
R
R
O
R
O
R
R
O
D
R
R
D
R
D
R
D
D
R
D
R
R
R
RD
R
D
D
D
D
O
D
R
R
D
D
R
R
O
R
O R
RR
O
R
D
D
RR
OR
DR
R
O
RR
O
RR
DO
R
D
D
R
D
R
D
R
R
O
R
D
D
O
R
RD
D
D
O
D
RR
D
R
R
DD
R
D
D
RR
D
O
R
RR
O
D
R
D
D
R
O R
R
R
R
R R
R
R
D
D
D
R
O
R
D
R
O
D
RRD
R
O
R
DR
O
R
R
D
R
R
D
R
R
R
R
R
RO
D
D
D
O
D
RD
R
R
R
D
D
D
D
R
R
R
D
R
O
RDD
D
R
R
D
R
R
R
D
O
RD
R
D
RR
O
R
R
DO
R
D
D
R
O
R
O
RO
D
O
D
R
R
D
O
D
R
D
R
D
D
O
D
OD R
D
RD
O
D
R
D
D
D
R
RD
R
D
R
D
O
D D
D
D R
O
R
D
R
R
R
R
R
R
O
D
R
O R
D
R
O
R
D
D
R
R
R
D
O
R
D
D
R
D
O
R
R
R
D
D
D
D R
D
R
D
D
D
D
DD
O
DD
R
R
RDD
R
R
D
D
D
D
RD
D
R
R
O
D
O
D
R
DR
D
O
R
D
D
O
D
D
R
O
D
R
R
D
D
D
D
O
D
D
R
D
RR
R
O
D
D
R
O
O
O
D
D
D
R
R
RR
R
R
R
D
R
D
D
D
R
D R
R
D
D
D
R
R
R
D
R
R
R
R
D
DR
O
D
R
D
R
O
R
R
RR
D
R
R
R
D D
D
R
R
D
R
R
R
D RR
RD
D
D
D
D
D
DR
R
R
R
R
R
R
D
R
R
D
D
O
R
R
DR
R
OO
D
R
R
R
R
RR
R
D
D
R
R
R
O
D
D
D RR
R
R
O
RD
O
R
O
R
R
R
DRR
RD
R
D
R
O
D
O
O
R
R
R
R
D
R
R
R
RR
R
D
O
D
D
DD
D
R
DR
R
D
D
R
D
R
D
D
R
D
R
RD
D
RR
D
R
R
DR
O
R
D
O
R
D
R
R
D
D
D
R
D
DD
D
R
O
D
O
R
D
R
R
R
D
R
O
R
R
D
O
R
O
D
D
D
R
O R
R
D
D
O
D
O
R
O
D
D
R
O
R O
R
R
O R
D
R
R
D
D R
R
D
R
D
R
DR
O
R
D
D R
R
D
R
D
D
D
O
R
D
R
R
D
DD
D
D
R
D
R
R
R
R
DD
D
R
R
D
D
R
D
R
D
O
D
O
D
R
R
O
O
OR
R
D
O
D
R
OD
R
R
D
D
R
D
R
R
R
O D
O
O
R
D
R
R
RO
O
DD
D
RO
D
R
R
R
RR
R
D
R
R
O
RR
R
D
OD D
R
R
D
R
D D
D
O
D
R
RRD
R
DD
D R
R
D
R
D
D
D
R
D
R
D
DD
R
D D
D
R
O
D
R
DD
O
D
D
R
RR
D
R
DO
R
R
D
R
D
R
O
D
R
R
D
O
D
D
R
O
O O
R
D
R
R
R
DD
R
R
R
R
D
D
R
D
O
R
O R R
D
R
D
D
R
D
RR
R
RR
R
R
D
R
D
R
D
O
R
R
D
R
O
D
RD
R
R
O
D
R
D
O
D
R
D
RD
D
R
OR
D
RR
O
D D
D
D
R
D
D
OD
D
R
D
RD RR
R
O
D
O
D
R
R
O
RR
R
D
R
R
O
R
D
D
D
D
O
R
R
R
O R
D
R
D
R
R
D
D
D
O
D
D
O
RO
D
D
O
D
D
D
O
D
R
D RO
O
D
R
O
R
D
R
R
D
DD
ROD
O
R
R
D
R
R
D
R
RD
D
D D
R
D
R
RR
O
D
O
D
D D
D
R
D
D
D
O
R
D
D
RD
R
R
D
D
R
D
O
D
R
DRD
D
D
D
D
D
D
D
D
R
D
D
D
D
D
D
R
D
D
D
D
R
R
DD
OD
O
D
D
R
OR
D
D
DD
D
D
O
RD
D
D
RR
D
O
O
D
D
R
D
O
O
D
R
D R
R
R
D
R O
RO
R
D
R
D
R
RO
D
O
R
D
D
D
D
D
O
D O
D
D
O
O
R
D
DD
OR
D
RD
D
R
R
O
D
R
RR
D
D
R
D
D
D
R
D
R
D
DR
R
D
R
O
R
R
D
R
D
D
D
R
O
D
R
D
DD
D
R
D
R
DR
D
D
ODD
D
R
OD
R
RD
O
DD
R
D
R
D
O
O
D
D
D
D
D
O
D
D
D
D
D
DD
D
OR
R
D
O
R
O
D
D
R
D
D
D
RR
R
R
D
R
R
R
O
D
D
D
R
D
D
D
R
D
D
RD
R
R
D
D
OD DR
R
D
D
R
D
D
D
D
D
O
D
D
R
D
R
R
O
RD
DR
R
R
D
D
D
D
RRD
R
D
RD
O
D
D
RR
D
D
O
D
R
D
D
D D
D
R
R
D
D
D
R
R
D
O
D
R
DD R
D
D
D
R
DR
DR
RD
R
D
D
RR
R
R
D
R
R
R
D
RDR
R
RRR
D
O
R
D
OR
DD
D
D
DD
D
D
R
D
R
D
R
D R
D
D
DR
R
R
OD
D
R
D
D
O
R
R
R
R
RR
D
O
O R
R
R
D
R
D
R
O
D
D
R
OR
R
R
D
R
D
R
D
R
R
D RR
R
O
R
R
R
R
O
R
R
R
O
RR RD
R
R
R
D
O
D
R
D
R
R
RR
R
R R
R
R
R
R
D
D
R
R
D
O
D R
R
D
R
R
D
R
R
RR
O
R
R
R
D
D
R
R
O
R
D
D
D
R
R
D
O
RR
DD
D
R
D
R
R
O
R
D
R
D
R
D
R
D
R
R
R
RR
R
R
O
R
R
R
R
R
RO R
RR
R
R
D
R
D
R
D
RD R
D
O
D
R
D
D
R
O
R
O
O
R
R RD
R
RR
O
D
R
R
R
R
R
RD
R R
R
D
D R
D
R
RO
O
D R
R
R
RD
R
D
R
R
O
DD
RR
D
O
R
R
D
O
D
R
R
D
D
DR
R
R
R
D
D
D
O
DD R RD
R
R
D
O
D
D
O
D
D
D
O
R
D
D
OD
R
O
RR
R
DR
R
R
O
O
R
R
D
R R
R
D
R
D
OR
R
O
O
D
O
D
R
R
O
R
D
R
R
D
R
R
D
R
RR
R
R
D
D
R
R
D
R
D
D
RD
R
D
RD
D D
R
R
D
D
D
O
D
R
RD
RO
D
R
O
OR
R
D
D
RR
R
R
D
R
R
R
D
R
R
D
R
RD
R
R
D
D
O
D
D
D
R
D
D
RD
D
R
O
R
D
R
O
D
D
R
D
R
R
R
D
R
O
DD
D
D
D
D
D
D
O
D
D
R
DR
R
D
D
RD
D
D
R
R
D
D
R
D
R
D
O
R
D
R
R
R
D
R
R
R
DO
D
R
O
D
R
D
D
RD
R
D
RD
R
RD
RD
R
D
D
R
O
O
DD
D R
D
R
D
R
O
OD
R
R
R
D
D
D
DR
R
RD R
O
DDRD
R
R
D
D
D
D
D
DD
O
R
R
R
R
R
R
RRD
R
R
R
D
D
RR
O
DO
D
R
D
R
R
RRR
D
R
R
D
D
R
D
R
R
R
D
RD
D
D
D
R
O
R
RR
D
D
R
R
O
R
D
R
D
R
R
D
D
D
O
R
R
DD D
R
R
R
R
D
D
D
D
R
R
D
O
D
O
R
D
D
O
D
D
R
RO
R
D
D
O
R
R
R
R
D
R
O
O
D
D
D
D
R R
O
DD
D
R
D
R
DD DD
O
D
R
R
O
D
O
D
R
D
R
D D
D
D
D
O
OO
R
R
D
R
D
RR
R
R
D
D
O
D
R
D
RR
O
R
R
D
D
DD
D
RD
R
D
D
OR
D
D
R
D
D
D
RD
D
O R
D
D
R
O
D
R
R
R
R
R
O
R
R
D
R
D
O
RR
R
D
R
D
R
R
D
D
D
R
RD
O
O
R
R
D
O
D
R
RR
D
O
D
R
O
DO
R
RO
DR
R
R
DD D
R
O
DR
D
R
R
D
D
R
R
R
R
R
DR
D D
RRD
R
O
D
R
D
DD
R
D
R
D
R D
O
D
R
R
D
R
R
D
O
R
R
R RR
RR
R
O
O
RR
D
R
R
R
R
R
R
R
D
R
D
R
D
RD
O
D RR
D
O
DD
D
R
D
D
DR
O
D
O
D
R
RD
D
R
O
R
D
R
R
R
O
DD
D
R
R
O
RO
R
D
D
D
D
R
R
R
R
R
R
D
RDR
D
D
D
R
D
D
O
R
D
D
R
D
DD
D
R
D
D
D
R
D
RR
R
O
RR
R
R
O
R
D
O
R
D
D
R
R
R
R
D
D
R
D
D O
R
R
R
D D
D
D
D
R
R
R R
R
O
O
R
D
D
D
R
O
D
R
D
R
D
R
D
D
D
R
D
R
O
R
D
O
D
OD
R
O
R
R
RRD R
D
D
D
R
O
D
D
D
R
D
D
O
R
D
R
D
R
DR
R
D
DR
D
R
D
R
R
RO
R
D
R
R
R
R
DD
D
O
R
D
D
R
R
D
R
D
D
R
R
D
OD
R
R
R
R
DD
O
D
O
O
D
O
R
DR
D
D
O
D
D
D
R
R
D R
RRR
D O R
D
D
R
D
D
R
R
D
R
D
D
R
R
O
R
D
R
D
D
DD
RDD
DO
R
DO
R
R
D
R
R
D
R
R
RD
O
D
DD
RD
D
D
D
D
RR
R
R
D
D
R
R
D
R
D
O
O
R
O
R O
R
R
R
R
RD
D
R
R
O
R
R
D
D
D
D
R
DO
R
D
R
DD
R
O
D RD
OR
D
O
DO
R
DD
R
O
D
D
O
O
D
D
R
R
D
R
D O
R
O
D
D
R
O
O
DD
R
D
R
D
R
D
O
RD
R
R
D
R
D
D
R
R
R
D
R
DR
O
R
D
D
R
D
R
R
R
D R
RR
D
OD
R
O
O
D
RD
D
R
D
R
R
R
R
O
R
D
RR
D D
D
RD
R
R
R
D
R
O
R
R
R
O
D
O
RD
D
O
R
R
R
RD
D
O
D
R
D
D R
DO
O
D
D
R
D
D
D
R
R
D
DD
R
R
R
O
R
D
O
R
R
R
D
D
R
RO
O
D
O
R
D
R
DD
D
R
R
D
O
R
R
D
R
D
D
R
D
R
R
R
D
R
D
RRR
D
O
R
R
R
R
D
OD
O
O
R
R
D
RR
R
R
D
R
D
D
D
D
D
D
OD
R
D
R
D R
R
D
R
D
R
D
O
D DD
D
D
O
O
DD
R
D
R
O
D
D
O
R
D
DO
D
R
D
R
RDD
RR
O
O
D
D
D R
R
RD
D
R
D D
R
R
D
D
D
RD
D
O
D
R
D
DO
OR
D
D
R
O
D
R
D
D
O
R
R
D
R
R
D
D
D
R
O
R
R
D
D O
D
R
R
R
O
R
D
O
RR
D
RD
DR
R
R
D
D
D
D
RD
O
D
D
D
D
D
R
R
D R
O
R
D
D
R
D
D
R
D
DD D
D R
O
R
R
D
R
RD
D
O R
R
O
R
D
D
D
O
D
R
R
D
R
D
R
R
D
R
R
R
D
O
R
R
DD
O
R
O
O
O
R
R
O
D
D
R
D
R
R
ODD
D
D
R
DD
D
R
R
D
R
R
R
D D
R
D RR
R
RD
D
D
D
D
D
D
R
RD
R
O
R
D
DD
R
DD
D
R
D
R
R
D
D
R
D
D
R
D
R
R
RD
D
D
R
R
D
R
D
R
R
D R
R
RR
R
D
R
D
R
R
OR
D
D D
O
R
D
D
D
R
D
R
R
R
D
D
D
R
R
R
D
D
R
R
D
R
D R
D
O
D
R
D
RD
D
D
D
R
D D
D
O
D
D R
O
DR
R
R
D
R
D R
RDR
R
RD
D
D
R
D
R
D
R
RD
D
D
D
DR
D
R
R
R
R
R
R
D
D
R
D
D
R
O
D
RR
R
D
D
RD
R
D
RR
O
R
R
D
D
D
O
D
D
R
D
O
D RR
R
R
D
D
R
D
D
R
D
R
D
D
RD
DO
D
R
D
D
R
D
D
R
D
R
O
D
R
DD
R
R
D
R
D
R
DR
R
O
D
D
R
D RR
D D
O
D
D
D
R
R
R
O
DD
D
D
RO
O
R
D
D
D
R
D
D
D
D
DD
D
D
D
D
D
D
R
DD
D DO
R
OR
D
D
D
R
R
D
D
O
R
D
RO
R
D
D
R
R
R
D
D
D
D
D
OD
R
R
D
D
R
D R
R
DD
RD
R
RD
R
R
O
R
D
RR
D
D
R
R
D
R
D
D R
D
R
R
R
R
D
D D
RD
D
R
R
R
D
R
D
D
D
D
D
D
RR
OR
D
R
D R
O
O
O
R
R
R
D
R
R
R
D
R
R
D
D
D
DD
DD
RO
D
R
R
R
RR
O
D
R
O
DR
R
R
O
O
D
R
D
D
D
R
D
D
R
D R
R
D
O
D
O
R
D
D
R
O
D
DD
R
D
D
RD
RR
R
D
D
R
D
O
RR
D
D
D
DD
D
D
R
D
D
O
RD
D
O
R
O
D
O
D
O
R
D
D
D
D
D
R
R
R
DD
O
R
R
D
D
D
D
D
R
D
R
D
O
D
R
R
RD
R
D
O
R
R
RD
R
D
R
R
D
D
OR
R
DDO
D
R
O
R
RR
R
D
D
D
O
D
R
R
O
D
D
R
D R
D
R
D
R
D
D
D
O R
R
D R
R
R
D
R
R
R
D
O
DO
R
D
D
D R
D
D
D
O
O
R
R
O
O
RD
R
O
D
O
D
R
DRD
R
RR
D
R
R
D
D
D
D
R RD
DD
RD
D
R
R
R
D
D
O
RO
R
R
R
R
O
R
R
R
O
R
R R
R
O
R
O
R
D
D
O
D
R
O
R
D
O
D
R
R
OR
R
R
D
R
R
R
D
D
O
R
O
D
RD
D
D
DRD
R
R
R
D
O
D
R
R
O
DD
R
O
O
R
D
R
O
R
D
D
D
R
R
D
D
D
RO
RD
DO
R
D
D
R
D
R
D
R R
RR
D
D
DD
O
R
R
RR
R
D
R
D
R
R
D
O
D
RO
R
R
R
D
RD
RD
R
O
D
RD
R
R
R
D
D
R
D
RD
D
R
R
RR
D
R
R
R
RD
R
RD R
D
RR
R
R
D
D
R
R
R
OD
R
R
D
R
D
D
R R
D
R
R
O
R
R
D
OD
D
O
O DD RD
D
R
R
R
R
D
D R
D
R
R
DD
R
D
RR
R
R
D
R
DR
R
R
O
R
D
D
DR R
D
R
OO
D
D
RR
O
D
D
D
D
R
R
D
R
R
R
D
D
R
R
R
D
R
D
O
R
D
R
D
R
R
R
D
R
O
D
D
R
R
D
D
R
D
O
D
R
O
D
D
O
D
R
RR
O
OR
D
O
D
R
D
R
R
D
D
R
D
O
RD
R
D
D
R
D
R
RD
R
R
O
DD
R
O
R
R
RR
D
RO
D
D
D
D
R
O
O
O
R RD
R
D
DO
D
D
O
D R
D
R
D
R
D
R
D
R
R
O
R
R
D
RR
D
RR
R
R
R
R
O
D
R
D
R R
DD
O RD
D
D
R
D
D
R
D
R
D
O
D R
D
R
R
R
R
D
R
R
R
D
D
D
R
R
D
D
R
R
RD
D
O
R
R
R
DR
R
R
O
R
R
R
O
R
D
R
R
R
D
O
R
RR
R
R
D
R
R
D
RO
R R
D
O
D
R
R
DR
R
D
D
R
D
R
R
R
R
D
R
R
OO
O
R
R
O
DD
OR
D
DR
R
R
OO
D R
D
D
D
R
R
R
RR
D
D
R
R
D
D
RD
O
RD
R
R
R
D
RD
R
R
DR
D
O
R
R
R
R
DR
O
O
O
D
DD
D
O
O
OD
R
D
R
O
O
D
R
R
R
D
D
R
R
R
R
RR
R
O
R
R
R
D
O
D
D
R
D
R
D
RR
D
D
D
D
D
R
R
D
D
R
D
D
D
R
D
R
DD
R
R
R
R
R
DD
D
O
DD
RR
D
O
R
D R
R
R
R R
R
O
DR
R
D
R
D
D
R
D
R
D R
R
D
D
D
R
R
D
D D
RD
O
R
R
RD
D
O
R
D
D
R
R R
D
R
R
R RD
R
R
D
R
R
DD R
RR
RR
R
R
R
RR
D
R
R
RD
D
D
R
RR
D
R
DR
D
R
R
R
R
R
RR
R
R
D R
D
RR
DD RD
D
R
R
R
R
RD
R
OD
D
D
RD
O
R
R
R
D
R
D
R
RR R
R
D
R
D
D
R
R
R
R
R
D
RR
R
O
D
R
R
D
D
RR R
DR
RR
R
R
R
D
R
DO
O
R
R
O
O
D
R
DR
D DD
R R
D
R
R
D
R
D R
R
D
R
R
R
R
R
D
D
R
D
D
RR
D
R
O
D
R
R
R
D
OR
R
RO
RR
O
D
RD
R
RR
RR
R
R
O
D
O
D
D
R
D
D
O
R
O
D
R
O
R
D
D
R
R
R
R
R
O
R
DD
D
D
D
R
R
R
D D
R
D
D
D
O
D
R
D
R
R
R
R
D
R
R RRD
R
D
R
R
R
O
D
D
D
RR
O
RR
R
R
R
D
R
O
R
R
D
R
R
R
O
D
R
R
R
R
R
D
R
D
R
D
R
R
DR
R
O
D
R
R
O
R
R
D
O
R
R
D
R
D
R
D
O
D
R
O
RR
R
RR
R
D
O
D
D
R
R
R
D
D
R
R
R
RR
D
O
R RD
R
R
R O
R
R
R
O R
R
D
D R
R
R
RD
D
D
R
D
D
D
D
O
D
O
R
R
R
R
R
O
RD
R
R
R
R
R
D
R
R
D
RR
D
D
D
D
R
D
R
D
D
D
O
R
RD
R
R R
D
O
D
R
R
D
D
D
D
O
RO
D
R
D
R
O
R
O
R
D
R
R
R
R
D
R
R
D
R
R
R
R
R
D
O
R
R
R
D
D
D
D
D
R
R
D
D
R
DD
D
D
D
D
R
R
R
D
DR
RD
R
RO
R
O R
D
D
R
D
O
D
R
D
O
R
R
D
R
D
R
R
DDD
R
D
D
R
D
RR
RO
D
D
D
D
D
O
O
D
DD
O
R
R
O
D
D
R
R
O
R
D
D
D
O
R R
D
R
R
R
D
R
O
D
R
D
D
D
D
O
D
O
RR
R
D
R
D
D
R
R
D
R
D
R
R
R
D
R
RDD
R
R
DR
O
OD
D
D
O
D
D
D
R R
R
R
R
ORR
D R
R
R
D
O
D
D O
D
R
R
R
O
D
D
R
R
R
DR
O
D
R
D
D
D
D
O
RR
R
R
O
D
O
R
D D
D
R
O
O
D
D
D
D
D
OD
D
R
R
DR
D
R
D
O
D
R
R
R
DD
D
D
D D
D
D
RO
O
D RRD
R
OR
RD D
R
R
D
D
RD
R
R
R
R
DR
R
O
R
R
R
D
O
R
O
DD
R
D
RO
R
R
DD
D
R
R
R
O
D
D
D
D
R
DO
D
R
O
O
D
R
D
DD
R
D
R
D
R
D
D
R
O
R
R
D
R
D
D
R
R
R
R
R
RR
D
O
D
R
R
D
R
R
R
RR
OD
R
R
R
DD
R
D
R
R
D R
D
R
R
D
OD
D
RO
R
D
DD
R
R
OO
R
R
O
R
R
RD R
R
R
R
D
D
D
R
O
RD
O
D
O
D
R
D
R
D
D
O
R
R
R
D
D
DD
D
D
O
R
D
R
R
R
O
RD
D
O
R
D
O
D
R
R
R
D
R
DR
D
D
O
D
R
DD
D
D
O
D
R
R
D
D
D
D
O
D
D
R
D
R
R
D
O
DR
DR
D
D
O
D
D
R
R
R
R
D RR
O
D
D
R
R
R
R
RR
R
R
D
D
O
D
D
D
D
O
O
D
R
D
O
D O
D
O
D
R
R
DR
O
D
D
DR
R
D
D
D
D
R
R
D
RD
D
R
R
D
R
O
RD
R
D
R
D
R
R
D
DD
O DD
D
D
DD
R
R
RD
D
R
D
D
RD
O
D
DD
O
D
R
D
R
D
R
R
R
R
RR
D
R
D
D
D
D
D
D
R
D
R
D
D
D
R
D
O
D
R
D
D
D
O
D
R
R R
D
D
D
D
D
R
R
R
R
O
RD
D
R
O
O
D
D
O
D
D
R
D
D
R
R
R
R
R
RDR
R
D R
D
D
R
R
R
D
D
D
R
R
D
R
R
D
D
D
D
D
R
D
R
D
R R
D
D
D
R
O
O
D
DD
R
D
R
R
D
D
O R
D
D
D
R
D
D
D
R
R
R
D
D
R
DR
D
R
D
D
D
R
R
R
D R
D
O
D
O
D
R
R
RR
D
R
D
DD
R
R
O
D
D
RD
D
D
R
R
OD
R
R
D
R
R
R
R
R
R
R
D
O
RD
R
D
DR
O
R
R
D
O
D
O
D
R
R
RRD
OD
O R
O
R
RD
O
OR
R
D
D
D
D
R
D
RR
D
R
D
D D
O
O
R
D
R
D
RR
D
D
D
D
R
R
O
D
O
R
DR
D
R
RD
O
O
D
D
R
O
D
R
D
DR
D
O
R
D
D
R
R
R
R
D
R
DR
O
D
R
D
DR
R
DO
R
D
O
DD
D
O R
R
R
D
D
D
R
RD
R
RO
D
R
O
D
R
R
R
R
R
R
R
D D
R
O
R
R
R
D
R
O
RO
R
O
R
R
D
D
D
RO
D
R
RR R
D O
R
RR
D
D
R
D
R
O R
D
R
D
R
R
R
R
D
D
RRR
D
RD
R
D R
R RR
D
O
R
D
R
D
R
D
D
D
O
R
O
D
R
D
D
R
O D
R
D
D
R
D
R
R
D
O
D
RR
D
R
D
R
D
D
ROR
R
D
D
R
DR
D
R
R R
D
R
R
O
D
D
RR
D
R
R
D
R
D
D
R
D
O
R
R
RD
D
D
R
R
D
R
RR
D
R
RO
D
D
D
R
R
DR
D
R
D
DR
D R
D
R
O
D
R
D
R
OD
O
D
D
DR
O
D
D
DR
RD
R
R
O
O
R
R
DD D
R
D
R
D R
D
D
R
O
O
D
D
O
R
R
O
O
OD O
O
D
D
R
O
O
O
OR
R
R
DR
R
D
O
O
R
D
R
DRD
R
D
D
R
R
D
D
DD
R
O
O
D
O
D
R
D
D
R
OD
D
D
D
D
D
D
O
R
D
DD
D
R
D
R
O
ROD
D R
D
R
D O
RR
D
R
R
O
D
OD
D
D
O
D
RR
R
D
−1.0 −0.5 0.0 0.5 1.0
−1.0
−0.5
0.0
0.5
1.0
Electionsdon't matter
Amer. fair society
Econ. is biasedCan't believe MSM
Ppl haveno say
Elites don'tunderstand
Figure 2: Ideal points and rigged-system cutting lines, VSG.
14
Voter Study Group
Degrees
Fre
quen
cy
0 45 90 135 180
0
5
10
15
CCES−Reed College
Degrees
0 45 90 135 180
0
5
10
15
Figure 3: Distributions of cutting-line angles with positions of rigged-system andstealth-democracy questions.
“non-elected, independent experts.” Each of these cutting lines suggests elitism,although none clearly disentangles process attitudes from policy ones.
In the VSG data, by contrast, we get a sense of anti-elite sentiment. Thecutting line for “elites don’t understand the problems I am facing” clearly sepa-rates a handful at the very top from the mass of people in both parties. Movingdown this figure, we next encounter “people like me don’t have any say” andthen the more extreme statement: “elections today don’t matter; things stay thesame no matter who we vote in.” Where the Reed data give evidence of elitism,the VSG data suggest a sort of process populism.
In the next two sections, I turn to the process-versus-policy question. Onesection looks at the people along this dimension. The second looks at othercutting lines defining it.
Who defines the second dimension?
Can we identify the sorts of people who tend to occupy either end of this seconddimension? Figures 4 and 5 provide some insight here. For each data set, Iestimated a simple, dummy-variable regression (with appropriate survey weights)of the second-dimension score on several demographic attributes. The plots
15
represent estimated coefficients, sorted in descending order of magnitude. Similarvariables cluster in similar spaces across the two data sets.
Perhaps due to the relatively small sample size, fewer coefficients in the Reeddata are precisely estimated. Several variables do stand out as predictors of up-ward positions on the second dimension: college education, graduate education,and family income over $60,000. The picture in the VSG is clearer, with its Nof 8,000. The highest-income respondents have the highest second-dimensionscores. Meanwhile, high-school education or less predict negative scores on thesecond dimension. So does reporting a racial identity of “other.”
Interestingly, vote choice in the 2016 primaries does not separate putativepopulists from elites. We might expect it to do so because two candidates ranself-consciously populist campaigns: Donald Trump and Bernie Sanders. I reraneach regression above with dummy variables for reported primary vote: BernieSanders, Hillary Clinton, John Kasich, Ted Cruz, and Donald Trump, with “didnot vote” and “other” as a combined reference category. Each of these variablesregisters with a statistically significant and positive coefficient. It may be that thepeople at the very bottom of the second dimension tend to be nonvoters. Thatwould make sense because primary electorates tend to be the most politicallyengaged.
Even though primary voting does not systematically predict negative second-dimension scores, a visual inspection of the ideal points is telling. Figures 6 and7 offer density plots of ideal points on each dimension, for each primary bloc,for each data set. Sanders and Cruz voters anchor the left and right of theliberal-conservative dimension in each. On the second dimension, Sanders votersanchor the left most clearly in the VSG data, with a clear mass of ideal pointsdistinct from those in every other camp. Sanders voters also anchor the second-dimension left in the Reed data, although much less starkly than in the VSG data.For those who might cast Sanders as a populist, these ideal-point distributionsgive some support to a populist-elite interpretation. It is worth noting, though,that many Sanders voters still had positive scores.
7 What other issues define the second dimen-sion?
So far, I have shown that attitudes about who ought to rule help define a seconddimension in public opinion. It is tempting to interpret this as a “process” di-
16
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
Correlates of second−dimension score:CCES−Reed College
Bold = significant at 95%.Partial correlation +/− 2 SDs
−1.0 −0.5 0.0 0.5 1.0
Other race
y−intercept
Middle Eastern
$120,000 − $149,999
$10,000 − $19,999
Prefer not to say
Native American
$250,000 − $349,999
$30,000 − $39,999
Mixed
Hispanic
Age−squared
Female
$70,000 − $79,999
2−year
$50,000 − $59,999
$500,000 or more
$100,000 − $119,999
$200,000 − $249,999
Black
$40,000 − $49,999
High school graduate
$20,000 − $29,999
4−year
$150,000 or more
$350,000 − $499,999
Some college
$150,000 − $199,999
Post−grad
$80,000 − $99,999
Asian
$60,000 − $69,999
Figure 4: OLS estimates of association with second-dimension score, Reed-CCESdata.
17
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
Correlates of second−dimension score:Voter Study Group
Bold = significant at 95%.Partial correlation +/− 2 SDs
−0.4 −0.2 0.0 0.2 0.4
Other race
y−intercept
No HS
Mixed
High school graduate
Middle Eastern
Some college
4−year
Post−grad
Age−squared
White
Male
$20,000 − $29,999
$60,000 − $69,999
$50,000 − $59,999
Black
Less than $10,000
Native American
$80,000 − $99,999
$70,000 − $79,999
Prefer not to say
$40,000 − $49,999
$30,000 − $39,999
$100,000 − $119,999
$150,000 − $199,999
$150,000 or more
Hispanic
$250,000 − $349,999
$120,000 − $149,999
$500,000 or more
$200,000 − $249,999
$350,000 − $499,999
Figure 5: OLS estimates of association with second-dimension score, VSG data.
18
Distribution of ideal points by 2016 primary vote:CCES−Reed College
First−dimension score
Den
sity
Clinton (D)Sanders (D)Trump (R)Cruz (R)Rubio (R)Kasich (R)
−1.0 −0.5 0.0 0.5 1.0
0.0
0.5
1.0
1.5
2.0
2.5
Second−dimension score
Den
sity
Clinton (D)Sanders (D)Trump (R)Cruz (R)Rubio (R)Kasich (R)
−1.0 −0.5 0.0 0.5 1.0
0.0
0.5
1.0
1.5
2.0
2.5
Figure 6: Ideal points by primary vote, CCES-Reed College.
19
Distribution of ideal points by 2016 primary vote:Voter Study Group
First−dimension score
Den
sity
Clinton (D)Sanders (D)Trump (R)Cruz (R)Rubio (R)Kasich (R)
−1.0 −0.5 0.0 0.5 1.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Second−dimension score
Den
sity
Clinton (D)Sanders (D)Trump (R)Cruz (R)Rubio (R)Kasich (R)
−1.0 −0.5 0.0 0.5 1.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Figure 7: Ideal points by primary vote, Voter Study Group.
20
mension, with people simply griping about a hard-to-understand political system(Hibbing and Theiss-Morse 2002).
Does the second dimension have any substantive policy content? One wayto check is to ask what other survey questions have orthogonal (to liberal-conservative) cutting lines. One systematic way to do that is to look at itemswith angles greater than 135 degrees or less than 45 degrees. These are questionsthat do more to define the second dimension than the first.
In the VSG data, several off-dimension questions capture economic issues.Others capture tolerance, perhaps consistent with Lipset’s two-dimension thesis.These questions are:
• Life in America for people like me, compared to 50 years ago (139 degrees,79 percent correct classification);
• Increased opportunities for women have significantly improved the qualityof life in the U.S. (144 degrees, 88 percent);
• Importance of infrastructure investment (166 degrees, 96 percent);
• Importance of the abortion issue (170 degrees, 94 percent);
• A five-question battery on expanding free trade (167, 173, 173, 173, and174 degrees, respectively, and 90, 93, 74, 90, and 92 percent correct clas-sification, respectively).
• Importance of immigration (39 degrees, 89 percent);
• Importance of religious liberty (32 degrees, 90 percent);
• And importance of jobs (2 degrees, 91 percent).
Turning to the CCES data, there are several such issues. Many deal withelectoral integrity and policy toward campaign contributions. The most extremeangles (greater than 170 or less than 10 degrees) deal with attitudes towardlocal government, zoning policies, local roads, and education policy. Correctclassification for each item is 76 percent or greater, with most items gettingscores of more than 80 percent.
In sum, people at the bottom of the second dimension tend to want moresocial services and less foreign trade. They are also skeptical about women’srights and religious tolerance. Going back to Lipset (1959a: 495) may be usefulhere:
21
. . . the lower-class individual is more likely to have been exposedto punishment, lack of love, and a general atmosphere of tensionand aggression since early childhood, experiences which often pro-duce deep-rooted hostilities expressed by ethnic prejudice, politicalauthoritarianism, and chiliastic transvaluational religion. His educa-tional attainment is less than that of men with higher socio-economicstatus. . .
8 Conclusion
This paper hypothesized and gave evidence of a second, process-populism di-mension in American public opinion. This dimension is orthogonal to liberal-conservative ideology, which continues to define competition between the majorparties, at least in the 2016 presidential election. People with high scores on thisdimension tend to make more money and have more formal education. In thelarger of my data sets, people with the lowest scores tend to have less education.Also in the larger set, membership in one racial out group predicts lower scoreson elitism. Finally, people who voted for Sanders in the 2016 Democratic primaryare to the left of most other respondents on this second dimension, although notin an extreme way.
Survey items not in the batteries of deductive interest also lined up with thissecond dimension: attitudes toward education, foreign trade, local government,and social services. These are policy outputs that some might say have atro-phied or changed a lot of late. Meanwhile, several hints of intolerance show uphere: toward women, immigrants, and people of other religions. Issue leadershipmatters, especially when times are tough.
Given available data, I cannot say whether this dimension existed in the past.An historical view suggests it has, at least in Congress, at least at certain points.Future research should more fully interrogate the second dimension in Americanpolitics (Bateman and Lapinski 2016). The trend in large opinion surveys willmake this easier to do. One interesting thing to watch will be how this dimensioninteracts with liberal-conservative, especially as the next presidential electionapproaches.
22
References
Abramowitz, Alan I., and Kyle L. Saunders. 2008. “Is Polarization a Myth?”Journal of Politics 70 (2): 542–555.
Aldrich, John H., and John D. Griffin. 2018. Why Parties Matter: Political Com-petition and Democracy in the American South. Chicago: University ofChicago Press.
Bailey, Michael A. 2007. “Comparable Preference Estimates across Time andInstitutions for the Court, Congress and Presidency.” American Journal ofPolitical Science 51 (3): 433–448.
Bailey, Michael, and Kelly Chang. 2001. “Comparing Presidents, Senators, andJustices: Inter-institutional Preference Estimation.” Journal of Law, Eco-nomics and Organization 17 (2): 477–506.
Bateman, David A., and John Lapinski. 2016. “Ideal Points and American Polit-ical Development: Beyond DW-NOMINATE.” Studies in American PoliticalDevelopment 30 (2): 147–171.
Berelson, Bernard R., Paul F. Lazarsfeld, and William N. McPhee. 1954. Vot-ing: A Study of Opinion Formation in a Presidential Campaign. Chicago:University of Chicago Press.
Bonica, Adam, Nolan McCarty, Keith T. Poole, and Howard Rosenthal. 2013.“Why Hasn’t Democracy Slowed Rising Inequality?” Journal of EconomicPerspectives 27 (3): 103–24.
Bucchianeri, Peter. 2017. “Political Parties, Local Cleavages, and the Dimen-sionality of Legislative Conflict: Evidence from City and County Councils.”Paper presented at the 2017 Annual Meeting of the Midwest Political Sci-ence Association, Chicago, IL.
Burnett, Craig M. 2017. “Parties as an Organizational Force on Nonpartisan CityCouncils.” Early online version. Party Politics. doi:10.1177/1354068817737996.
Burnham, Walter Dean. 1982. “Into the 1980s with Ronald Reagan.” In TheCurrent Crisis in American Politics, 273–276. New York: Oxford UniversityPress.
Campbell, Angus, Philip Converse, Warren E. Miller, and Donald E. Stokes. 1960.The American Voter. New York: John Wiley & Sons.
23
Chait, Jonathan. 2011. “Why Liberals Like Compromise and Conservatives HateIt.” The New Republic (March 3). https://newrepublic.com/article/84630/why-liberals-compromise-and-conservatives-hate-it.
Converse, Philip. 1964. “The Nature of Belief Systems in Mass Publics.” InIdeology and Discontent, edited by David E. Apter, 206–261. New York:Free Press of Glencoe.
Dewan, Torun, and Arthur Spirling. 2011. “Strategic Opposition and GovernmentCohesion in Westminster Democracies.” American Political Science Review105 (2): 337–358.
Drutman, Lee, Larry Diamond, and Joe Goldman. 2017. Follow the Leader: Ex-ploring American Support for Democracy and Authoritarianism. ResearchReport. Democracy Fund Voter Study Group.
Dyck, Joshua J., and Mark Baldassare. 2009. “Process Preferences and Votingin Direct Democratic Elections.” Public Opinion Quarterly 73 (3): 551–565.
. 2012. “The Limits of Citizen Support for Direct Democracy.” CaliforniaJournal of Politics & Policy 4 (1): 1–20.
Fiorina, Morris P. 2009. Disconnect: The Breakdown of Representation in Amer-ican Politics. Norman, OK: University of Oklahoma Press.
. 2010. Culture War? The Myth of a Polarized America. 3rd ed. New York:Pearson Longman.
Hare, Christopher. 2015. “Constrained Citizens? Ideology and the AmericanVoter.” Doctoral dissertation, University of Georgia.
Hare, Christopher, Tzu-Ping Liu, and Robert N. Lupton. 2018. “What OrderedOptimal Classification Reveals about Ideological Structure, Cleavages, andPolarization in the American Mass Public.” Early online version. PublicChoice.
Herrmann, Michael, and Ulrich Sieberer. 2018. “The Basic Space of a Revo-lutionary Parliamenty: Scaling the Frankfurt Assembly of 1848/49.” Earlyonline version. Party Politics.
Hibbing, John R., and Elizabeth Theiss-Morse. 2002. Stealth Democracy: Amer-icans’ Beliefs about How Government Should Work. Cambridge, UK: Cam-bridge University Press.
24
Hill, Seth J., and Chris Tausanovitch. 2015. “A Disconnect in Representation?Comparison of Trends in Congressional and Public Polarization.” Journal ofPolitics 77 (4): 1058–1075.
Hinich, Melvin J., and Michael C. Munger. 1994. Ideology and the Theory ofPolitical Choice. Ann Arbor: University of Michigan Press.
Hix, Simon, Abdul Noury, and Gerard Roland. 2006. “Dimensions of Politics inthe European Parliament.” American Journal of Political Science 50 (2):494–520.
Jackman, Simon. 2001. “Multidimensional Analysis of Roll Call Data via BayesianSimulation: Identification, Estimation, Inference, and Model Checking.” Po-litical Analysis 9 (3): 227–241.
Jenkins, Jeffrey A. 2000. “Examining the Robustness of Ideological Voting: Ev-idence from the Confederate House of Representatives.” American Journalof Political Science 44 (4): 811–822.
Jessee, Stephen. 2016. “(How) Can We Estimate the Ideology of Citizens andPolitical Elites on the Same Scale?” American Journal of Political Science60 (4): 1108–1124.
Kellermann, Michael. 2012. “Estimating Ideal Points in the British House ofCommons Using Early Day Motions.” American Journal of Political Science56 (3): 757–771.
Kinder, Donald R., and Nathan P. Kalmoe. 2017. Neither Liberal nor Conserva-tive: Ideological Innocence in the American Public. Chicago: University ofChicago Press.
Lazarsfeld, Paul F., Bernard Berelson, and Hazel Gaudet. 1944. The People’sChoice: How the Voter Makes up his Mind in a Presidential Campaign. NewYork: Duell, Sloan, / Pearce.
Lipset, Seymour Martin. 1959a. “Democracy and Working-class Authoritarian-ism.” American Sociological Review 24 (4): 482–501.
. 1959b. “Some Social Requisites of Democracy: Economic Developmentand Political Legitimacy.” American Political Science Review 53, no. 1(March): 69–105.
25
Martin, Andrew D., and Kevin M. Quinn. 2002. “Dynamic Ideal Point Estimationvia Markov Chain Monte Carlo for the U.S. Supreme Court, 1953-1999.”Political Analysis 10 (2): 134–53.
Masket, Seth, and Boris Shor. 2015. “Polarization without Parties: Term Limitsand Legislative Partisanship in Nebraska’s Unicameral Legislature.” StatePolitics and Policy Quarterly 15 (1): 67–90.
Miller, Gary, and Norman Schofield. 2003. “Activists and Partian Realignmentin the United States.” American Political Science Review 97 (2): 245–60.
. 2008. “The Transformation of the Republican and Democratic PartyCoalitions in the U.S.” Perspectives on Politics 6 (3): 433–450.
Mudde, Cas. 2017. “Populism isn’t dead. Here are five things you need to knowabout it.” The Guardian (July 17). https://www.theguardian.com/
commentisfree/2017/jul/07/populism-dead-european-victories-
centrists.
Mudde, Cas, and Cristobal Rovira Kaltwasser. 2017. Populism: A Very ShortIntroduction. New York: Oxford University Press.
Noel, Hans. 2013. Political Parties and Political Ideologies in America. New York,NY: Cambridge University Press.
Page, Benjamin I., and Martin Gilens. 2017. Democracy in America? What HasGone Wrong and What We Can Do About It. Chicago: University of ChicagoPress.
Poole, Keith T. 2000. “Nonparametric Unfolding of Binary Choice Data.” Polit-ical Analysis 8 (3): 211–237.
. 2005. Spatial Models of Parliamentary Voting. New York: CambridgeUniversity Press.
Poole, Keith T., and Howard Rosenthal. 1997. Congress: A Political-EconomicHistory of Roll Call Voting. New York, NY: Oxford University Press.
Poole, Keith, Howard Rosenthal, and Christopher Hare. 2015. “House: Vote onClean DHS Funding Bill.” Voteview Blog (March 6). https://voteviewblog.wordpress.com/2015/03/06/hello-world/.
26
Roberts, Jason M., Steven S. Smith, and Stephen R. Haptonstahl. 2015. “TheDimensionality of Congressional Voting Reconsidered.” American PoliticsResearch (October 15). doi:10.1177/1532673X15608940.
Rosenthal, Howard, and Erik Voeten. 2004. “Analyzing Roll Calls with PerfectSpatial Voting: France 1946-1958.” American Journal of Political Science48, no. 3 (July): 620–632.
Santucci, Jack. 2018. “Evidence of a Winning-cohesion Tradeoff Under Multi-winner Ranked-choice Voting.” In press. Electoral Studies. doi:10.1016/j.electstud.2017.11.003.
Schofield, Norman, Gary Miller, and Andrew Martin. 2003. “Critical Elections andPolitical Realignments in the USA: 1860–2000.” Political Studies 51:217–240.
Shor, Boris, and Nolan McCarty. 2011. “The Ideological Mapping of AmericanState Legislatures.” American Political Science Review 105 (3): 530–51.
Sides, John, Chris Tausanovitch, Lynn Vavreck, and Christopher Warshaw. 2018.“On the Representativeness of Primary Electorates.” Online. British Journalof Political Science. doi:10.1017/S000712341700062X.
Tausanovitch, Chris, and Christopher Warshaw. 2013. “Measuring ConstituentPolicy Preferences in Congress, State Legislatures, and Cities.” Journal ofPolitics 75 (2): 330–42.
Treier, Shawn, and D. Sunshine Hillygus. 2009. “The Nature of Political Ideologyin the Contemporary Electorate.” Public Opinion Quarterly 73 (4): 679–703.doi:10.1093/poq/nfp067.
27
