Electoral Vulnerability,Party Affiliation,and Dyadic Constituency Responsiveness in U.S. Legislatures

Abstract

It is often argued that electoral vulnerability is critical to constituency responsiveness. We investigate this possibility using different measures of vulnerability, but argue that in the United States the Republican Party may be less responsive than the Democratic Party due to its core constituency and view of representation. We test our hypotheses relying on an innovative research design that exploits referenda in U.S. states to compare legislator voting behavior with voter preferences on exactly the same policy proposal, allowing us to overcome the measurement problems of much previous research. Based on a newly compiled data set of more than 3,000 voting decisions for 818 legislators on 27 referenda, we find high levels of congruence, but that congruence with the median voter is higher for legislators who are running for reelection. We also find that Democrats are more responsive after a close election but that Republicans are not sensitive to electoral margins.

Keywords

responsiveness representation parties referenda marginality

Introduction

In modern democracies, there is a normative expectation that legislators pursue the preferences of their constituents (Rehfeld, 2009), but a number of studies question whether this usually takes place (Bartels, 2008; Gerber, 1996; Gilens, 2011; Lax & Phillips, 2012; Masket & Noel, 2012; Rhodes & Schaffner, 2017; Snyder, 1996). Scholars have examined how electoral pressures may shape responsiveness (Gay, 2007; Griffin, 2006; Kuklinski, 1977), but we know less than we might because relatively few studies examine dyadic congruence because many studies rely on broad indicators of opinion or ideology (Wlezien, 2017) rather than the detailed policy proposals on which legislators actually vote. Furthermore, in modern democracies political parties are important to representation, and left and right parties have different representational orientations (Korpi, 1978), especially in the United States, where recent research suggests that Democrats and Republicans view their representational roles very differently (Grossmann & Hopkins, 2016) and interact with different types of constituents (Broockman & Skovron, 2018). This suggests that Democrats and Republicans may respond to constituency opinion and electoral pressures in different ways, which we examine in this article.

Whether policymakers generally act in line with the people that they represent is a critical question for students of democracy (Beyer & Hänni, 2018). There are many factors that might shape this process, but electoral considerations are thought to be critical in ensuring responsiveness to district opinion (Gay, 2007; Griffin, 2006; Kuklinski, 1977; Mansbridge, 2003). Scholars have noted that in winner-take-all systems like the United States, electoral pressures should cause responsiveness to the district median (Downs, 1957). However, it may be that primary elections mean that the median voter is not so important to U.S. legislators. In addition, there is emerging evidence that responsiveness varies considerably across parties. Conservative parties may generally have less of an ability to anticipate or respond to the preferences of the district median because they interact more with upper class individuals and groups (Broockman & Skovron, 2018; Korpi, 1978). In the United States specifically, it seems that Democrats view representing external groups as a key feature of the party’s mission, whereas Republicans tend to view themselves as fulfilling an ideological project (Grossmann & Hopkins, 2016), leading Democrats to be more responsive to public opinion in the U.S. Congress, according to recent research (Lax et al., 2019; Rhodes & Schaffner, 2017).

Despite numerous studies, we actually know less than we should about the factors associated with congruence between legislator actions and constituency opinion on the types of specific policy proposals that legislators vote on, as we rarely have good measures of district-level opinion on specific legislative proposals. Although it is equally important to understand if the broad ideology of voters is reflected in the ideology of their elected representatives (Griffin, 2006; Rhodes & Schaffner, 2017), it is a different question than whether legislators are responsive on individual legislative proposals. Furthermore, it is important to understand responsiveness in the state legislatures, where so much important policy is made. In this article, therefore, we examine responsiveness using referenda, where constituents and legislators vote on the exact same detailed policy proposal, allowing us to generate unusually detailed measures of constituency opinion on the types of detailed policy proposals on which legislators vote.

A handful of studies have used referenda to develop measures of constituency opinion and test responsiveness (Brunner et al., 2013; Giger & Klüver, 2016; Masket & Noel, 2012; Matsusaka, n.d.). Our analysis examines more than 3,000 voting decisions by 818 legislators on 27 referenda held in six U.S. states between 2011 and 2014. We observe fairly high levels of congruence between median voters and legislators. Our referenda are diverse in terms of topic and salience. Most of the referenda we examine are referred to voters by legislators (often as a result of requirements that constitutional provisions be approved by voters), but a handful of the issues are “veto” referenda, which appear on the agenda in response to earlier legislative actions. For all legislators, congruence is found to be higher when legislators are running for reelection. However, we find that Democrats are somewhat more responsive overall and much more responsive than Republicans when they have experienced a close election. These findings have important implications for how we understand the different types of representation and responsiveness that the two parties in the United States provide. Of course, although the benefits of using referenda to measure constituency opinion are clear, this approach may also limit the generalizability of our findings, which we discuss in the conclusion.

The Determinants of Dyadic Congruence

There are numerous ways that legislators may be responsive to the needs and preferences of their constituents (Harden, 2013; Jewell, 1983), but here we focus on “dyadic responsiveness,” that is, that between a single constituency and its representative. Most studies implicitly test for responsiveness from the standpoint of the “delegate” model of representation (Rehfeld, 2009). For instance, recent studies arguing that representation in Congress is broken take the ideologies or opinions of voters on certain issues and examine how they correspond with vote-based measures of legislative ideology or voting on certain issues (Bartels, 2008; Gilens, 2011; Gilens & Page, 2014; Lax et al., 2019; Rhodes & Schaffner, 2017). Of course, there are other conceptualizations of representation (Mansbridge, 2003), and we explicitly consider the possibility that not all legislators view themselves as delegates. But, in keeping with most previous literature, our point of departure is a delegate model of representation: do representatives vote along with constituency policy preferences?

Competitive Elections and Future Congruence

Representatives almost necessarily vote along with the preferences of some of their constituents, but there is a normative expectation in a democracy that they vote along with the typical or majority view in the district. We expect legislators to be responsive to the majority opinion because majorities have legitimacy in democratic settings. This is why recent studies showing that representatives are responsive mostly to the wealthy are so troubling (Bartels, 2008; Gilens, 2011; Gilens & Page, 2014). In winner-take-all election systems, a member of the district majority, the median voter, is decisive in election outcomes (Downs, 1957). In practice, responsive to the median may not take place because responsiveness to other actors is more important (primary electorates, party leaders, and interest groups) or, as we discuss below, not all legislators view responsiveness as one of their key responsibilities.

In any case, it is often assumed that electoral pressures should increase responsiveness. But is it necessary to have very competitive elections or does the mere existence of elections ensure responsiveness? Many scholars have assumed that the former is the case. Indeed, scholars have argued that Congress may not be very responsive due to a lack of competitive elections (Ansolabehere et al., 1992), and there is even less competition in many state legislatures. A lack of competitive elections may result in a lack of responsiveness to district opinion and leave politicians free to pursue their own preferences, or be responsive to other political actors, such as party leaders (Kuklinski, 1977) or organized interests and wealthy individuals (Gilens & Page, 2014).

Even with a general lack of competition it may still be the case that more marginal legislators, that is, those who have experienced more competitive elections in the past, are more responsive to public opinion. Assuming that there is some probability of an electoral penalty for a deviation from district preferences, more marginal members should be more responsive. Constituents appear to value policy responsiveness from their representatives (Wolak, 2017) and research examining the U.S. Senate shows that constituents in competitive states pay more attention to politics and hold their Senator more accountable for deviations from their policy preferences in elections (Jones, 2013). Classic studies also show that legislators who represent competitive districts are more likely to defect from their party in roll call voting, presumably to vote more with their constituents (Froman, 1963; MacRae, 1952). In a more recent important study, Griffin (2006) finds that marginal members of Congress are more reflective of district ideological/partisan preferences in their roll call voting.

A number of other studies, however, fail to find any relationship or observe a weak or inconsistent relationship between electoral vulnerability and responsiveness. For instance, Kuklinski (1977) finds that marginal state legislators were only more responsive to constituents on some issues. Gay (2007) finds that legislators representing majority–minority districts in California, which have very low levels of competition, are not any less responsive to constituents than other legislators. Brunell and Buchler (2009) find that members of Congress who win in a landslide demonstrate greater ideological congruence with their constituents in their roll call voting. Giger and Klüver (2016) fail to find any relationship between electoral marginality and responsiveness to public preferences in Switzerland.

One reason for these inconsistencies may be that electoral marginality may not be the best indicator of the subjective electoral threat that legislators feel. Even with a small objective likelihood of defeat, legislators may be paranoid about losing elections (Mayhew, 1974). Some scholars have argued that the American system of frequent elections makes politicians too concerned about being responsive to the public (King, 1997). It is even possible that few competitive elections indicate that politicians are good at anticipating their constituents’ preferences. Although research finds that legislators are more responsive to the wealthy, the wealthy and others agree on many policy issues (Enns, 2015; Lax et al., 2019). From this perspective, politicians usually do a good job of responding to constituency preferences because they are worried about reelection, even if they have not experienced a near defeat. If this is so, merely running for reelection should be sufficient to ensure responsiveness.

Perhaps another reason that marginality is not closely tied to responsiveness in some studies is that the salience of an electoral challenge is not constant, but varies over one’s term in office. Electoral concerns may become more salient to legislators as elections approach. Existing research shows that in close proximity to elections elected officials are more responsive to the public’s preferences and interests (Ahuja, 1994; Miller et al., 2018). Thus, we also examine how proximity to an election may shape responsiveness. Finally, we consider the possibility that there may be interactions among these three threat variables by including a series of interaction terms between these variables in different models. Overall, with our various measures, we expect that the probability of legislator defection decreases with electoral vulnerability.

Party Variation in Responsiveness to Electoral Threat

So far, in keeping with much existing literature, we have assumed that all legislators are roughly equally familiar with district opinion and respond to electoral marginality in the same way. But representation in most democracies is mediated by parties. Generally speaking, we think conservative parties will be less likely to know the preferences of the median voter because they are less likely to interact with individual constituents or organized interests that share her preferences. Historically, conservative parties tend to have an upper class constituency, whereas left parties have a lower and middle class constituency (Korpi, 1978). If legislators tend to communicate more with their core constituents, then a natural result is that left or center-left parties will probably have a better sense of the preferences of the typical voter.

Recent research in the United States shows that both major parties are more likely to hear the preferences of upper class and more conservative voters (Butler, 2014). But this problem is more severe for legislators from the Republican Party. Broockman and Skovron (2018) show that all state legislators overestimate the conservatism of public opinion, but Republican politicians dramatically overestimate it. They suggest that this reflects that conservative voters are more likely to contact legislators, but especially Republican legislators. Republicans also appear to rely more on the business community to form their legislative policy positions, and business is often more conservative than the median constituent (Hertel-Fernandez et al., 2018).

Even if Republicans are aware of true constituency opinion, we think there is a strong likelihood that they will be less responsive to it because the modern Republican Party has a different understanding of representation than the Democratic Party. According to Grossmann and Hopkins’s (2016) analysis, the Republican Party is a conservative ideological movement, whereas the Democratic Party is a coalition of social groups with different policy priorities. Being responsive to the preferences of diverse groups of individuals is a key feature of the Democratic Party. Democrats approximate delegates, but Republicans practice what Mansbridge (2003) calls either “gyroscopic” or “promissory” representation. Republican candidates put forth conservative policy stances on the campaign trail and promise to pursue them if elected, sometimes even if they are quite unpopular at the time they are debated (consider the debate over the repeal of the Affordable Care Act or the tax cut legislation enacted during Donald Trump’s first term).

Neither of these forms of representation is necessarily better in a normative sense. However, if Republicans and Democrats have different ideas about how constituency opinion should shape their decisions, then they will necessarily be differentially sensitive to constituency opinion and electoral marginality. There is emerging evidence for these partisan differences in responsiveness. Gilens (2012) shows that Senators of both parties are rarely responsive to the poor, but Democrats are at least somewhat responsive to the middle class, whereas Republicans are only responsive to upper income voters. Rhodes and Schaffner (2017) find that lower and middle income constituents are represented as well as upper class constituents by Democrats, but that Republican members of Congress are much more responsive to the wealthy. Finally, Lax et al. (2019) find that Democratic U.S. Senators are considerably more responsive to all types of groups than are Republican Senators. There is little analysis of these partisan differences at the state level, but given the nationalization of the parties it is important to consider that the parties may be differentially responsive.

Taking these points together, we hypothesize that congruence is lower for Republicans and that Republicans are less sensitive to electoral vulnerability. Specifically, the probability of legislator defection is higher for Republicans than for Democrats and Democrats are more sensitive to marginality than Republicans.

Research Design

Using Referenda in the U.S. States to Study Responsiveness

A handful of studies solve the problem of an absence of measures of public opinion on detailed policy proposals actually before legislators by using referenda, where constituents and legislators vote on the exact same policy proposal. For instance, Rogers (2017) has used referenda to examine whether voters punish legislators for votes they disagree with. Scholars have examined legislative responsiveness and interest group influence with referenda in Switzerland (Giger & Klüver, 2016), how electoral incentives shape responsiveness (Matsusaka, n.d.), differential responsiveness to high- and low-income voters in California (Brunner et al., 2013), and patterns of partisan representation in California (Masket & Noel, 2012).

We have selected the referenda that were analyzed in this study as follows: First, we compiled a list of all state-wide referenda that took place between 2011 and 2014. The second criterion was the availability of roll call data and the availability of data on the results of public voting at either the electoral district level or the precinct level, which we could then use to construct district-level measures. Thus, we have arrived at a sample of 27 referenda in six U.S. states (California, Hawaii, Idaho, Illinois, Maryland, and Washington) that were held between 2011 and 2014. Overall, this results in data on 3,305 voting decisions of 818 legislators on the selected 27 referenda on a wide variety of issues. The incentives to respond to constituents may vary depending on whether an issue is a “hot-button” issue versus a more mundane matter. We do have a handful of hot-button issues in our sample, and in models in the Supplemental Appendix we control for whether a referendum was brought before the voters to overturn previous legislative action, which we can be confident are hot-button issues (in our sample two on gay marriage, one on collective bargaining for teachers, and one on Indian gaming in California).

Legislative voting data were collected from state legislative websites. The district-level popular vote was much more difficult to compile. Only a handful of U.S. states publish referenda voting by legislative district, in our sample California, Maryland, Washington, and Hawaii for the House. Most states only report the voting results of referenda at the precinct level, including Hawaii (for Senate districts), Idaho, and Illinois.¹ For these states, we collected data from on precinct names and the respective counties and legislative districts and constructed district-level measures of voting. Given the large number of precincts in many states (e.g., 10,030 precincts in Illinois), this is not a trivial task. Details of our precinct matching procedures are given in the Supplemental Appendix.

Although referenda voting provides us with excellent measures of constituency opinion, there are some limitations. Clearly, although they are geographically, racially/ethnically diverse and diverse in terms of size and region, we do not have a representative sample of states and, barring Idaho, our states are moderate to liberal. In addition, although we have a variety of salient versus non-salient, partisan versus non-partisan issues, the types of issues that are usually subject to referenda are not necessarily representative of all issues. In addition, referenda only measure the opinions of voters. However, for us this is advantageous, as we expect responsiveness to the median voter and not median constituent. We think these trade-offs are worthwhile, but we must be aware of them as we interpret the results.

An important feature of referenda for our purposes is that, except for on advisory referenda (which are not in our sample), the legislators vote before the public. This means that at the time that legislators vote they do not know, but rather must estimate constituency opinion, which is the normal situation in legislative decision making. In contrast with most studies, however, the subsequent public vote provides us with a good measure of actual public preferences, so that we can test how responsive legislators are to imperfectly known opinion at the time of the legislative vote by using actual opinion at the time of the constituency vote.

One potential concern is that we might be measuring public responsiveness to the legislative vote, rather than legislative responsiveness to unknown public opinion. This seems doubtful. First, most citizens have no idea how their legislators vote on most issue. Even on laws salient enough that they led to public veto referenda to overturn them, Rogers (2017) shows that constituents did not punish their legislators for defection from district preferences, indicating that constituents are probably not paying close attention to legislative votes, even on very hot-button issues. It may be the case that the overall public debate about salient issues (in our sample, e.g., gay marriage and teacher tenure) are shaping public opinion about these issues, but the exposure to the broad elite debate will be roughly constant across districts, allowing us to reliably estimate district-specific responsiveness. Unlike legislators, constituents lack an incentive to vote along with their representative if they disagree with her. Indeed, the correlation between the percentage of a chamber voting for a proposal and the percentage of the district voting for a proposal is only .38 in our sample.

Outcome Variable: Legislator Defections

Based on the voting behavior of legislators and the outcome of the referenda, we have generated a defection measure which is coded 0 if the voting decision of the legislator is in line with the majority decision of voters in her district, and it is coded 1 otherwise.²

Measuring Explanatory and Control Variables

We use three variables to measure electoral threat or vulnerability. First, a standard measure of electoral marginality, the percentage of the vote a legislator received in the previous election. Next, whether a legislator is participating in the next election. Finally, time to the next election. The next major explanatory variable is simply the Democratic party affiliation of the legislator. As noted, we also examine the interactions between the electoral threat/marginality variables and party affiliation.

We control for a number of variables that might otherwise confound the hypothesized relationships. First, we include constituency consensus, which is the percentage point margin of victory of the majority side in the referendum. For instance, a 60-40 split would result in a score of 20. Second, we control for whether a legislator served in the state lower chamber. Third, we control for the total legislative vote margin, the percentage of members in a legislator’s chamber voting yes, because on some issues legislators may feel institutional pressure to vote a particular way. Finally, we control for the years of service at, which may be associated with a greater ability to correctly estimate district opinion due to familiarity with the district (Glazer & Robbins, 1985), or it may be the case that legislators who have served for a long-time are of the type that are better able to respond to the preferences of their district or are less susceptible to electoral pressures. The Supplemental Appendix provides summary statistics for all the variables included in our analysis.

Model Specification

The observations in our data set are nested within states and clustered into 27 policy proposals on which legislators and districts sometimes vote repeatedly. Ignoring the non-independence may result in deflated standard errors (Steenbergen & Jones, 2002, pp. 219–220). We first attempted to estimate models with random intercepts at the state level, but the small number of states precluded estimation of confidence intervals, so we dropped these random intercepts from the model, as recommended (Barr et al., 2013). Our data are also clustered at the referendum and legislator/district level. Because legislators are not nested within particular referenda but rather vote on multiple referenda, we estimated a cross-classified mixed model with random effects for referenda and legislators/districts (Gelman & Hill, 2007; Leckie, 2013).³

Results

In this case, the mean of the dependent variable is substantively interesting. The legislative voting decisions were in line with the constituency median in 79.2% of all cases examined. Although defections are considerable (20.8%), for comparison they are fewer than in Switzerland where Giger and Klüver (2016) observe a defection rate of 32.8% (Table 1). There is considerable variation in defection rates across referenda, which we present in the Appendix.

Table 1.

Determinants of Legislator Defections.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5
Margin	0.846 (−0.50)	0.502 (−1.17)	0.850 (−0.48)	1.012 (0.988)	0.255*** (−2.93)
Run Next Elect	0.622** (−2.72)	0.435* (−2.23)	0.469 (−1.84)	0.621** (−2.73)	0.617*** (−2.80)
Time to Next Elect	0.999 (−1.25)	1.000 (−1.25)	1.000 (−1.06)	1.000 (−0.33)	1.000 (−1.44)
Democrat	0.778 (−1.62)	0.778 (−1.61)	0.778 (−1.61)	0.779*** (−1.60)	−0.232*** (−4.09)
Dist Consensus	0.001*** (−9.09)	0.001*** (−9.10)	0.001*** (−9.08)	0.001*** (−9.08)	0.001*** (−8.79)
Run Next Elect ×Margin		1.946 (1.08)
Run Next Elect × Time to Elect			1.000 (0.75)
Margin × Time to Elect				1.000 (0.798)
Democrat × Margin					7.799*** (3.68)
Yrs. Service	0.989 (−1.08)	0.990 (−1.07)	0.990 (−1.06)	0.990 (−1.09)	0.994 (−0.70)
Legis Vote Margin	0.071*** (−3.65)	0.072*** (−3.62)	0.070*** (−3.66)	0.073*** (−3.50)	0.071*** (−3.68)
Lower Chamber	0.866 (−0.83)	0.861 (−0.87)	0.870 (−0.81)	0.866 (−0.84)	0.929 (−0.43)
Constant	25.744*** (3.62)	33.537*** (3.78)	32.540*** (3.67)	22.004** (2.85)	45.444** (4.31)
Random effects
Referenda-level variance	6.61 (2.18)	6.60 (2.17)	6.62 (2.18)	6.62 (2.18)	6.16 (2.03)
District-level variance	0.266 (0.118)	0.267 (0.118)	0.269 (0.119)	0.268 (0.119)	0.152 (0.106)
Model fit
N	3,097	3,097	3,097	3,097	3,097
χ²	105.67***	107.17***	106.23***	105.64***	126.13***

Note. z statistics in parentheses, standard errors for variance terms.

p < .05. **p < .01. ***p < .001.

Turning to the multivariate results which present odds ratios and z-scores, and beginning with the vulnerability variables, we see that the marginality of a legislator is not associated with defection. In contrast to some previous research, nor is the length of time until the next election. However, we do observe that legislators running again for election are significantly less likely to defect. Specifically, their odds of defection are almost one third less. Thus, it seems that the mere presence of an election, and not necessarily competitiveness, ensures some responsiveness.

For party affiliation, we see that being a Democrat rather than a Republican is associated with a reduction in odds of defection by approximately .22, though the coefficient is not quite significant (p = .106). Moving away from odds ratios to more straightforward predicted probabilities, in Figure 1 we plot the average marginal effect of Democratic party affiliation (relative to Republican partisan affiliation) and the average marginal effect of running in the next election (relative to not running in the next election) on the predicted probability of defection. The probability of defection is about 2% smaller for a Democrat, but the confidence interval crosses zero. Running again in the next election reduces the probability of defection by just more than 4 percentage points. This effect is not huge, but is meaningful.

Figure 1.

The marginal effect of democratic partisanship and running in next election.

Next, we turn to models 2 to 5, which include the various interaction terms. We fail to observe any statistically significant interaction effects between victory margin in the last election, participating in the next election and the length of time until the next election. Model 5 shows, however, that the interaction between the Democrat indicator variable and the margin variable is significant in the expected direction.

In Figure 2, we plot the effect of switching from being a Republican to being a Democrat on the probability of defecting across different values of the margin of electoral victory. We see that Democrats are particularly sensitive to narrow election victories. When they just barely won their last election they are much less likely to defect than Republicans on subsequent votes, nearly 15 percentage points less likely on average. Interestingly, even as their margin of victory approaches 60 percentage points, Democrats are less likely to defect than Republicans with similar margins of victory, suggesting that Democrats and Republicans respond differently to electoral threat.

Figure 2.

The marginal effect of democratic party membership on the probability of defection, by marginality.

Conclusion

In this article, we studied the link between legislative voting and constituency opinion using referenda, which allow us to get an accurate estimate of constituency opinion on the types of detailed proposals that legislators consider, across a variety of legislative institutions and issues. We have compiled a new data set that comprises thousands of voting decisions across hundreds of legislators on a few dozen referenda. Overall, we find what could be viewed as a reasonably high degree of responsiveness, about 80% congruence. On the contrary, in more than 20% of the cases legislators did defect from their constituents. Furthermore, in many districts it is certainly the case that voters are somewhat unrepresentative of non-voters (Franko et al., 2016), meaning that responsiveness to the preferences of the median of all constituents is lower.

In general, we do not find much evidence that electoral vulnerability is a major factor driving responsiveness, for all legislators. Neither the margin in the previous election nor the time to the next election are associated with a lower likelihood of defecting from constituents in any straightforward way. We do observe, however, that merely running in the next election (which the legislator presumably planned while making voting decisions) is associated with greater responsiveness. Thus, perhaps the mere existence of elections, which are not usually close but which could be close if legislators did things constituents disapprove of, are sufficient to ensure responsiveness. This has important implications for term limits, which based on our findings would tend to reduce responsiveness on average in the last term in which legislators serve, thus reducing responsiveness overall.

We also find important differences in the probability of defection and responsiveness to electoral vulnerability by party. Democrats seem to defect at slightly lower levels overall, though this result was not quite statistically significant. We clearly observe that Democrats are more responsive to the preferences of the median voter after a close election than Republicans are, however. This suggests that Republicans and Democrats do respond differently to electoral threat. Our finding about the partisan differences in representation is consistent with some other recent research at the federal level, showing that Democrats are more responsive to public opinion in voting (Gilens, 2012; Lax et al., 2019; Rhodes & Schaffner, 2017) and that at the state level Democrats might have a better sense of constituency opinion (Broockman & Skovron, 2018).

Although we have made important progress in understanding legislator–voter congruence by using referenda, there are important questions for future research that remain. Our design strengthens internal validity as it allows for directly comparing voter preferences and legislative behavior of legislators on exactly the same policy proposals. But due to the burdens of matching constituency voting with legislative districts which are not publicly available for many states, this study was limited to a relatively small number of issues in a fairly liberal group of states. In addition, the types of issues considered—though diverse—are nevertheless fairly atypical by nature of the fact that they were the subject of referenda. It may also be the case that referenda that are required to be passed on to voters because they are constitutional amendments enforce a degree of institutional (over constituent) loyalty that we would not see on most other issues and that supermajority requirements mean that defections are higher than we might otherwise see. In addition, in our sample of relatively liberal states it may be that Republicans need to be more concerned about primary electorates rather than general electorate, whereas in a sample of predominantly conservative states the opposite would be true.

Future research should explore a larger number of issues to understand general responsiveness better and to understand differences in the parties more. One question for future research is how responsiveness varies depending on the salience of the issue, which our data set was not well-suited to address. In addition, scholars should consider whether our finding regarding the differential responsiveness of the Republican Party holds for other conservative parties, or whether this is a somewhat unique feature of contemporary U.S. politics, and also further explore what differential responsiveness from the parties means for policy making and interactions with interest groups.

Supplemental Material

OnlineAppendix – Supplemental material for Electoral Vulnerability, Party Affiliation, and Dyadic Constituency Responsiveness in U.S. Legislatures

Supplemental material, OnlineAppendix for Electoral Vulnerability, Party Affiliation, and Dyadic Constituency Responsiveness in U.S. Legislatures by Nathalie Giger, Heike Kluver and Christopher Witko in American Politics Research

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Christopher Witko

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Nathalie Giger is an associate professor in the Department of Political Science and International Relations at the University of Geneva. She has published on a range of themes related to the linkage of citizens and political elites, unequal representation and party competition. She is currently working on a project on the political consequences of the perceptions of inequality funded by the Swiss National Science Foundation.

Heike Klüver is a professor and chair of Comparative Political Behavior at the Humboldt-Universität zu Berlin. Her research interests include Comparative European Politics, Political Parties, Coalition Governments, Interest groups, Political Representation as well as Quantitative Text Analysis.

Christopher Witko is a professor and associate director of the School of Public Policy at the Pennsylvania State University. His research focuses on the interrelationships between inequalities in the representative process, public policy and economic outcomes, especially economic inequality.

References

Ahuja

(1994). Electoral status and representation in the United States Senate does temporal proximity to election matter? American Politics Quarterly, 22(1), 104–118.

Ansolabehere

Brady

Fiorina

(1992). The vanishing marginals and electoral responsiveness. British Journal of Political Science, 22(1), 21–38.

Barr

D. J.

Levy

Scheepers

Tily

H. J.

(2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 68(3), 255–278.

Bartels

L. M.

(2008). Unequal democracy: The political economy of a new gilded age. Princeton University Press.

Beyer

Hänni

(2018). Two sides of the same coin? Congruence and responsiveness as representative democracy’s currencies. Policy Studies Journal, 46, S13–S47.

Broockman

D. E.

Skovron

(2018). Bias in perceptions of public opinion among political elites. American Political Science Review, 112(3), 542–563.

Brunell

T. L.

Buchler

(2009). Ideological representation and competitive congressional elections. Electoral Studies, 28(3), 448–457.

Brunner

Ross

S. L.

Washington

(2013). Does less income mean less representation? American Economic Journal: Economic Policy, 5(2), 53–76.

Butler

D. M.

(2014). Representing the Advantaged: How Politicians Reinforce Inequality. Cambridge University Press.

10.

Downs

(1957). An economic theory of democracy. Harper & Row.

11.

Enns

P. K.

(2015). Reconsidering the middle: A reply to Martin Gilens. Perspectives on Politics, 13(4), 1072–1074.

12.

Franko

W. F.

Kelly

N. J.

Witko

(2016). Class bias in voter turnout, representation, and inequality. Perspectives on Politics, 14, 351–368.

13.

Froman

L. A.

(1963). Congressmen and their constituencies. Rand McNally.

14.

Gay

(2007). Legislating without constraints: The effect of minority districting on legislators’ responsiveness to constituency preferences. Journal of Politics, 69(2), 442–456

15.

Gelman

Hill

(2007). Data analysis using regression and multilevel/hierarchical models. Cambridge University Press.

16.

Gerber

E. R.

(1996). Legislative response to the threat of popular initiatives. American Journal of Political Science, 40, 99–128.

17.

Giger

Klüver

(2016). Voting against your constituents? How lobbying affects representation. American Journal of Political Science, 60(1), 190–205.

18.

Gilens

(2011). Policy consequences of representational inequality. In Enns

P. K.

Wlezien

(Eds.), Who gets represented? (pp. 247–284) Russell Sage Foundation.

19.

Gilens

(2012). Affluence and influence: Economic inequality and political power in America. Princeton University Press.

20.

Gilens

Page

B. I.

(2014). Testing theories of American politics: Elites, interest groups, and average citizens. Perspectives on Politics, 12, 564–581.

21.

Glazer

Robbins

(1985). Congressional responsiveness to constituency change. American Journal of Political Science, 29, 259–273.

22.

Griffin

J. D.

(2006). Electoral competition and democratic responsiveness: A defense of the marginality hypothesis. Journal of Politics, 68(4), 911–921.

23.

Grossmann

Hopkins

D. A.

(2016). Asymmetric politics: Ideological republicans and group interest democrats. Oxford University Press.

24.

Harden

J. J.

(2013). Multidimensional responsiveness: The determinants of legislators’ representational priorities. Legislative Studies Quarterly, 38(2), 155–184.

25.

Hertel-Fernandez

Mildenberger

Stokes

L. C.

(2018). Legislative staff and representation in congress. American Political Science Review, 113, 1–18.

26.

Jewell

M. E.

(1983). Legislator-constituency relations and the representative process. Legislative Studies Quarterly, 8(3), 303–337.

27.

Jones

P. E.

(2013). The effect of political competition on democratic accountability. Political Behavior, 35(3), 481–515.

28.

King

A. S.

(1997). Running scared: Why America’s politicians campaign too much and govern too little. Free Press.

29.

Korpi

(1978). The working class in welfare capitalism: Work, unions, and politics in Sweden. Routledge & Kegan Paul.

30.

Kuklinski

J. H.

(1977). District competitiveness and legislative roll-call behavior: A reassessment of the marginality hypothesis. American Journal of Political Science, 68(4), 627–638.

31.

Lax

J. R.

Phillips

J. H.

(2012). The democratic deficit in the states. American Journal of Political Science, 56(1), 148–166.

32.

Lax

J. R.

Phillips

Zelizer

(2019). The party or the purse? Unequal representation in the US senate. American Political Science Review. http://www.columbia.edu/~jrl2124/partypurse.pdf

33.

Leckie

. (2013). Cross-classified multilevel models using stata: How important are schools and neighborhoods for students’ educational attainment? In Garson

G. D.

(Ed.), Hierarchical linear modeling: Guide and applications (pp. 311–322). SAGE.

34.

MacRae

(1952). The relation between roll call votes and constituencies in the Massachusetts House of Representatives. American Political Science Review, 46(4), 1046–1055.

35.

Mansbridge

(2003). Rethinking representation. American Political Science Review, 97(4), 515–528.

36.

Masket

S. E.

Noel

(2012). Serving two masters: Using referenda to assess partisan versus dyadic legislative representation. Political Research Quarterly, 65(1), 104–123.

37.

Matsusaka

J. G.

(n.d.). When do legislators follow constituent opinion? Evidence from matched roll call and referendum votes. Stigler Center for the Study of the Economy and the State.

38.

Mayhew

D. R.

(1974). Congress: The electoral connection. Yale University Press.

39.

Miller

Witko

Woods

N. D.

(2018). How the unorganized mass public (sometimes) gets represented in regulatory politics. Political Research Quarterly, 71, 88–101.

40.

Rehfeld

(2009). Representation rethought: On trustees, delegates, and gyroscopes in the study of political representation and democracy. American Political Science Review, 103(2), 214–230.

41.

Rhodes

J. H.

Schaffner

B. F.

(2017). Testing models of unequal representation: Democratic populists and republican oligarchs? Quarterly Journal of Political Science, 12(2), 185–204.

42.

Rogers

(2017). Electoral accountability for state legislative roll calls and ideological representation. American Political Science Review, 111, 555–571.

43.

Snyder

J. M.

Jr. (1996). Constituency preferences: California ballot propositions, 1974-90. Legislative Studies Quarterly, 21, 463–488.

44.

Steenbergen

M. R.

Jones

B. S.

(2002). Modeling multilevel data structures. American Journal of Political Science, 46(1), 218–237.

45.

Wlezien

(2017). Public opinion and policy representation: On conceptualization, measurement, and interpretation. Policy Studies Journal, 45, 561–582.

46.

Wolak

(2017). Public expectations of state legislators. Legislative Studies Quarterly, 42(2), 175–209.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.22 MB