The Macro Sort of the State Electorates

Abstract

Individual-level studies of partisan and ideological change find that individuals generally adjust their ideological preferences to match their partisan affiliation. In examining this process among the state electorates, we find that contrary to these studies, states have adjusted their partisanship to match their ideology. In addition, we use a measure of state elite ideology to show that state parties have a role in the character of the partisan sort of the states. These results are consistent with political explanations of party strategy and rational mass responses for the character of macro-political change in the states over the last half century.

Keywords

public opinion state ideology state partisanship party polarization

The relationship between partisanship and ideological self-identification has changed quite substantially over the last several decades. The political parties in the first half of the twentieth century were organized largely along group lines and were ideologically heterogeneous. This is no longer the case; after the 1960s and 1970s, the parties have organized primarily along ideological, rather than social, cleavages (Abramowitz 2010). Several scholars, often relying on data from the American National Election Studies, have demonstrated that the contemporary parties now enjoy a greater alignment of partisan and ideological identifications (Abramowitz 2010; Hetherington 2001; Levendusky 2009).

This aligning of arguably the two most important elements of voter decision making has a number of important consequences for the character of U.S. elections. With fewer voters conflicted by partisanship and ideological preferences, we have seen decreases in split ticket voting, including a substantial increase in levels of voting for the same party for both president and members of Congress (Hetherington 2001; Levendusky 2009). The growing congruence between party and ideological identification, frequently referred to as “sorting” or “partisan sorting,” has important implications for a number of aspects in political behavior, from parties changing their campaign strategy as they make more overt appeals to the base, to the electorate failing to punish candidates for ideologically extreme voting behavior (Bafumi and Shapiro 2009).

At the individual level, this can be seen in the CBS/New York Times polls we use for the current analysis. The correlation between party and ideological identification in 1977 was a very low .16, but as the parties began differentiating themselves, citizens began adapting their issue positions and/or partisanship to better suit the profile of party elites.¹ The relationship between party and ideology has strengthened considerably over time, to a correlation of .46 in 2011.

A similar pattern has emerged at the macro level, where we can also see a greater consistency in the correspondence between partisanship and ideology among the state electorates. Where thirty years ago there was an absence of a relationship between state ideology and state partisanship, today, there is an unmistakable robust positive relationship between the two. Aggregating these same survey data to the state level (roughly by decade), there was almost no relationship between aggregate state ideology and state partisanship in the 1980s (r = .11); the relationship increased to average out at r = .29 during the 1990s with a clear shaping up of partisanship and ideology in the most recent period (2001–2011) with the correlation growing to a strong r = .67. We should note that this was not only a matter of the Southern realignment but was rather a broad trend across the country of state electorates adapting their ideology and partisanship in response to elites.²

What has caused this pattern of increasing congruence of partisanship and ideology at the state level? At the individual level, there seems to be a general consensus that the prime mover in this process is political polarization by elites (Hetherington 2001; Layman and Carsey 2002; Layman, Carsey, and Horowitz 2006). Elites have moved in a way over the last three or four decades that has resulted in parties that have become more ideologically homogeneous and distinct from one another, and the most compelling work indicates that party activists may be the group that is influencing elite extremism (Aldrich 1983; Layman et al. 2010; Schofield and Miller 2007).

Our concern in this article is with the strengthened “macro sort” of state electorates we find in the increasing correlations of state partisanship and state ideology over time. How have the states changed in moving from the chaotic relationship between state partisanship and ideology of the 1970s and 1980s to the much greater alignment of 2000 and beyond? In particular, we are interested in what has driven this macro-state alignment—though we note that the alignment is by no means complete.

We frame our analysis around two related questions. One is, what has changed in the states? Does ideology adjust to state partisanship, is it the other way around, or do they exist in a reciprocal dynamic? Once we have answered this question, we are in a position to address the second: do the state parties have any role in the macro sort of state electorates? It is possible that the alignment of state electorates has been a function of changes in partisanship or in ideology, but that the state party elites are irrelevant to this—it is just a matter of aggregating individual changes (and replacements), all in response to changes in the parties at the national level.

Is It Partisanship or Ideology?

There is a great deal of research that testifies to the primacy of party identification in citizen attitude formation and behavior (Bartels 2000, 2002; Campbell et al. 1960; Goren 2005; Goren, Federico, and Kittilson 2009; Green, Palmquist, and Schickler 2002; Johnson 2006; Levendusky 2009). Under the “Michigan” model of partisanship, party is developed as a deeply held psychological attachment that is itself not only nearly permanent but also subsequently influences the individual’s ideological positions by way of a perceptual screen (Campbell et al. 1960; W. E. Miller and Shanks 1996). Similarly, the social-attachment theory presented by Green, Palmquist, and Schickler (2002) argues that partisan attachment is predicated by an emotional attachment that is rooted in social identity, without much basis in ideological considerations. Moreover, party affiliation influences how individuals interpret political events and evaluate politicians.

Under both schools of thought, partisanship is quite stable, ultimately shaping many aspects of individual behavior. Party identification is not only a strong predictor of individual vote choice (Bartels 2002) but also influences what sort of information individuals assimilate (Zaller 1992), the media they attend to (Iyengar and Hahn 2009), how they respond to new political information (Jacobson 2007), and even what neighborhoods to live in (Bishop 2008). Accordingly, partisanship serves as one of the key anchors in understanding how individuals structure their political behavior.

It should come as no surprise then, that partisanship serves as an important component in understanding how individuals respond to political change. When the ideological characteristics of the political parties began to adjust, most voters used their partisanship to understand these changes and adapted their issue attitudes to better suit the positions of the party elites (Carsey and Layman 2006). While there were some who maintained their core ideological positions and adjusted their partisanship, the vast majority of voters who converted their positions did so while adhering to a partisan anchor, and adjusting their issue positions.

Levendusky (2009) is similarly explicit about the importance of partisanship in shaping the ideological characteristics of electorate in recent decades. He argues that the polarization of the party elite has increased the salience of the party brand name, which has simplified the cue taking process among voters. Given that most voters do not have strongly integrated belief systems (Converse 1964), the clarification on issue positions by the elites trickles down to the voters, who adopt their party’s position on the issue. Thus, the major element in the growing convergence between partisanship and ideology in the electorate has been the result of Republicans adopting the ideological identification of conservatives and Democrats increasingly taking on the liberal and moderate labels (see also Bartels 1996; Goren 2005; Johnson, Hagen, and Jamieson 2004; Markus and Converse 1979).

There is, of course, an alternative view that sees partisanship as influenced by candidate and economic performance (MacKuen, Erikson, and Stimson 1989) as well as by issues, events, and previous voting behavior (Carmines 1991; Fiorina 1981; Franklin 1992; Franklin and Jackson 1983; Sundquist 1983). The “running tally” notion of partisanship (Achen 2002; Fiorina 1981), where partisan identification is determined by an individual’s assessment of the parties, suggests that changes in the electorate have been led by voters who use their ideology as an anchor as they adjust their partisanship (see also Abramowitz and Saunders 2008). However, perhaps due to the scarcity of citizens who have stable and well-articulated ideologies (Bartels 2002; Converse 1964), this view has not been linked to the meaningful and systematic changes in partisan sorting in the same manner as the party as anchor model.

These two views, of partisan or ideological supremacy, imply quite different macro manifestations of change. For example, as the party elites diverged first on race (Carmines and Stimson 1989) and later on social/moral issues (Abramowitz 2010; Layman 1997), the partisanship supremacy view implies that state electorates should have maintained their aggregate partisanship and changed their ideology. Alternatively, the ideological supremacy view implies that the liberal states should become more Democratic and conservative states more Republicans. Research that has examined the behavior of state electorates in the aggregate suggests support for the latter view. Erikson, Wright, and McIver (1993) suggest that relative state ideology is quite stable, at least into the 1980s.

What is important for the analysis here is that most of the theories that examine the stability of party identification also identify conditions under which partisanship is adjusted to comport with issue stances or ideology. Carsey and Layman (2006), for example, identify the conditions of respondents being aware of party differences and the specific issue being salient for issues to influence changes in partisanship. Claassen (2011) similarly finds that partisan and issues attitude change is found disproportionately among the most politically attentive respondents. Highton and Kam (2011) offer an additional condition, arguing that issue positions and partisanship influence one another at different times. Using long-term panel data,³ they report that in the latter period of their analysis (1982–1997), issue positions shaped partisanship, rather than vice versa, but prior to that (1973–1982), issue preferences seem to be brought into agreement with stable partisan identities. Using the same long-term data to examine party polarization, Stoker and Jennings (2008) show that attitude change varies with age and time period and that the macro results are influenced by the process of cohort replacement. Even Green, Palmquist, and Schickler (2002, Ch. 6), who are probably the strongest critics of the party identification as a policy-motivated heuristic, acknowledge in their explanation of change in the South that partisanship yields to changes in party elite policy behavior, albeit in their theory via changes in social identity.

Our focus here is on the macro-level manifestations of micro change. We are intrigued by the disconnect between the incremental, and possibly hard to measure, changes at the individual level and the incontrovertible patterns of change at the macro level. Ultimately, it is difficult to use the conclusions from these micro-level studies to offer an explanation for the massive regional realignment of partisan loyalties over the last forty years (Black and Black 2002; G. Miller and Schofield 2003). Unquestionably, the strongest test is the case of political change in the American South, which has transformed from the most consistently Democratic region of the nation to become the bedrock of a larger, more competitive and more conservative Republican Party (Knuckey 2006; Lublin 2004). Similar changes have occurred elsewhere, however, especially in urban areas of the Northeast and the upper Midwest (Brewer, Mariani, and Stonecash 2002).

Focusing on the aggregate state electorates offers an excellent perspective to the study of macro-level political change. The states get overlooked in micro-studies focus on individual behavior (e.g., Carsey and Layman 2006) or macro-level studies that focus on the national level (e.g., Carmines and Stimson 1989). The unique advantage that the states offer is their semi-autonomous relationship with the national parties (Gimpel 1996). While the state party elites tend to take positions generally similar to the positions party elites take at the national level, these are tailored to fit the political climate of the particular state. As a consequence, there is a considerable degree of heterogeneity in the positions and behavior of state political parties, and we suspect how citizens respond to these cues. We should expect to find that the electorate in states with polarized parties will respond differently to party cues than those who are in states with less polarized parties, for example. The fifty different state party systems offer more variation than we can find at the national level.

Our interest in the states is not simply due to the presence of variation, however. Most scholarship on partisanship tends to focus on national-level partisanship, but as we and others (e.g., Cotter et al. 1989; Gimpel 1996; Hadley 1985; Jennings and Niemi 1966) show, citizen conceptions of partisanship and ideology are simultaneously conditioned by state- and national-level cues.

Furthermore, our focus on the macro behavior of the state electorates matches the fact that political outcomes are commonly the result of aggregate, not individual processes. In particular, U.S. Senate and gubernatorial elections, as well as Electoral College votes in presidential elections take place at the macro-state level. This study’s focus on shifting partisanship and ideology at the aggregate state level helps bring these elections into account.

It is important to note that while there is an obvious logical connection between micro- and macro patterns of change, the correspondence is not necessarily clear in observed patterns. For example, analysis of individual-level voting decisions does not paint a complete picture of the electoral process. In voting for U.S. senators, issues appear relatively modest in their impact on individual voting decisions, but in the aggregate at the state level, candidates’ issues stances prove to be unmistakably important in accounting electoral outcomes (Wright 1989). Thus, in our focus on partisan sorting among state electorates, we follow in the tradition of previous macro analyses (Erikson, MacKuen, and Stimson 2002) of seeking an understanding for the aggregate behavior of citizens, while fully accepting that those patterns are compatible with a different configuration of causal processes at the individual level.⁴

We note that changes to state ideology and partisanship will occur due to a number of simultaneous processes. While shifts state macro partisanship and ideology will be largely influenced by individual-level decisions, changes in the states will also be due to population changes. Migration patterns will adjust or reinforce the existing partisan–ideological characteristics of the states (Bishop 2008). For example, when Republicans move, they are likely to move to areas that are even more Republican, while Democrats are likely to move to areas that are even more Democratic (Tam Cho, Gimpel, and Hui 2013).

In addition, state population naturally changes as a function of generational replacement. Green, Palmquist, and Schickler (2002) argue that this is one of the key factors leading to the partisan conversion of the South, as the older conservative Democrats passed away and were replaced by conservative Republicans. While these population shifts are clearly important in accounting for the partisan and ideological changes in the states, Abramowitz and Saunders (2006) show that the changes in the South and elsewhere are due less to generational replacement than the influence of ideology on party identification. It is precisely this last point that we seek to address here.

Partisan and Ideological Trends

As the change we are most interested in is the long-term macro sort among the states, we use a local polynomial smoothing method to describe partisan and ideological change rather than inspecting noisy yearly scatterplots.⁵ We begin by fitting a polynomial smooth to yearly estimates of ideology for respondents living in the most Democratic and most Republican states (for the period 1977–1984). Here we use individual-level estimates of ideology and partisanship from CBS/New York Times polls. If mass ideological attitudes and issue preferences are drawn toward existing partisan dispositions as elites polarize, we expect the most Democratic states to become more liberal, and the most Republican states to trend conservative, and partisanship of the states to remain relatively stable. Looking first at the bottom two lines in Figure 1, this is not what we find. Rather, all states, both Democrat and Republican, trend conservative rather than liberal. Overall, the change is modest, which indicates that the relative ideological positions of the states are extremely consistent over time. For the state electorates, there is no indication that the movement toward macro alignment described earlier is due to state ideology conforming to state partisanship.

Figure 1.

Changes in partisanship (PID) and ideology (local polynominal smooths) for selected groups of states.

We do the same local polynomial smooth to yearly estimates of partisanship for respondents in the most liberal and most conservative states. If state ideology influences state partisanship, we should expect conservative states to trend Republican over time and liberal states trend Democratic. Referring to the top two lines in Figure 1, this is exactly what we find. The movement for the most conservative states is downward, indicating clear shifts in state partisanship toward the Republican Party. In contrast, the liberal states have consistently trended Democratic since the late 1980s. At the macro level, it seems quite clear from the figure that partisanship moves into alignment with ideology rather than vice versa.

Estimating the Contributions of Lagged State Partisanship and Ideology

We offer a more formal estimation of the relative contribution that state ideology has on state partisanship, and contribution that state partisanship has on state ideology; this uses aggregate estimates of state ideology and partisanship from the CBS/New York Times surveys (1977–2011)⁶ scaled so that higher values are associated with greater liberalism or Democratic support. We estimate two variations of the same model: first regressing current state partisanship on lagged values of partisanship and ideology, and then estimating current state ideology similarly on lagged values of both variables.⁷ We group the surveys into four-year portions—by presidential term, for more reliable state-level estimates (Carter, Reagan I, Reagan II, etc., through Obama I)—making the state-term the unit of analysis, and estimate for state partisanship and ideology:

{PID}_{t} = a + b_{1} {PID}_{t - 1} + b_{2} {Ideology}_{t - 1} + e .

{Ideology}_{t} = a + b_{1} {PID}_{t - 1} + b_{2} {Ideology}_{t - 1} + e .

If ideology serves as an anchor for partisan change, the coefficient for lagged ideology should be positive and significant in model 1a. Similarly, if party serves as an anchor for ideological change, the coefficient for lagged partisanship should be positive and significant in model 1b. Naturally, we expect that lagged values of partisanship will have a strong relationship with contemporary partisanship and that lagged values of ideology will have a strong relationship with contemporary ideology.

The results are shown in Table 1 and show clearly that while previous partisanship has no impact on current state ideology, previous ideology has a significant effect on current state partisanship.⁸ The positive sign for state ideology indicates, as suggested by Figure 1, that the more liberal states trended Democratic and the more conservative states trended Republican.

Table 1.

Regression of State Partisanship and Ideology on Lagged Partisanship and Ideology.

	Partisanship_t	Ideology_t
Partisanship_t−1	0.795*** (27.3)	−0.004 (−0.13)
Ideology_t−1	0.122*** (4.2)	0.817*** (26.2)
n	384	384
R ²	.707	.665

Standardized coefficients; t-values are in parentheses. Data are from the cumulative CBS/New York Times polls. The time periods are presidential administrations (Carter to Obama), as discussed in text.

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

Thus, at the state macro level, the evidence is clear that partisanship is adapted to conform to existing ideological preferences and that ideology does not change as part of the growing alignment. The point here is that at the aggregate level we find no evidence of issues or ideology being pulled into alignment with partisanship. Quite simply, although it may be true for many, even most, individuals that issue positions are changed to be consistent with party identification, this process does not apply to changes we see among state electorates. Rather, the story here is one of state partisanship adjusting around ideology. The relative ideological positions of the states appear to remain much as they were thirty years ago. State ideology appears to be the central feature by which we should be characterizing state electorates, and state partisanship appears to be changing as the public continues to bring their party and ideological identifications into ever closer alignment over time.

Effects of State Ideology and Partisanship on Behavior

While we have established the stability of state ideology relative to the more dynamic state partisanship, we have yet to establish the importance that this distinction makes for explaining the political behavior of the state electorates. Our findings in the previous section suggest that values and policy preferences are probably more important in characterizing and understanding state electoral and policy behavior than partisanship.⁹ To determine how each serves as a predictor of state behavior, we look at the relative contributions of partisanship and state ideology in presidential elections over time. While party is frequently argued to be the dominant factor in explaining citizens’ electoral behavior, at the macro level we find clear evidence that ideology has been the more important predictor of presidential voting of the states for most of the past forty years.

To estimate the relationship between presidential vote and state characteristics, we use the following model for each presidential election:

Dem . Share = a + b_{1} PID + b_{2} Ideology + e,

where Dem. Share is the percentage of the votes the Democratic presidential candidate received in the election, and state partisanship and ideology are estimated from the CBS/New York Times polls. Rather than presenting these estimates in a table, Figure 2 shows the standardized coefficients for each election.¹⁰

Figure 2.

Presidential vote regressed on state partisanship and ideology 1972–2012 (standardized coefficients).

For most of the period, and consistently since the 1984 election, we see that state ideology is the stronger predictor of state presidential vote. The period of the 1970s demonstrates the volatility of an electorate struggling to reconcile partisanship and ideology. The difference between the 1972 and 1976 outcomes is particularly instructive. The dominance of state ideology in the 1972 election shows that electorates were responding to the ideological choice being offered by the parties—in this case the highly issue-oriented McGovern campaign—and not just relying on habits of partisan voting. By contrast, when the policy differences between the candidates are more muted as in the 1976 Carter–Ford contest, and ideology played less of a role, standing partisan differences predict state presidential vote. As the electorates began to respond to the increasing polarization of party elites in the mid-1980s, and the candidates nominated by the parties reflect these differences, state ideology is increasingly the better predictor, while the predictive value of state partisanship wanes.

For macro-state sorting, the answer to the question of what is stable and what moves in response to party elite polarization is quite clear. State ideology is stable; the increased alignment we see among the states is due to shifts in partisanship.

Do the State Parties Matter?

For the macro-partisan sort of state politics, we have established that the alignment has been one of state partisanship coming into increasing alignment with stable patterns of state ideology. But this does not establish whether state parties have any part in this or not. We can gain snippets from the literature to argue either way. In looking at the resurgence of mass partisanship, for example, Hetherington (2001) and Levendusky (2009) identify explicitly the polarization of the congressional parties, thus implicitly making the argument (1) that everyone is exposed to the same levels of party elite polarization and (2) that what happens in party and electoral politics at the state level is probably of not much consequence for partisan sorting. Given the dominance of national politics in the media, this may not be too far off the mark.

However, it is also the case that citizens vote for candidates who compete in very different electoral contexts. There is an unmistakable tendency for Democrats running in more conservative settings to moderate their stands, just as is a similar tendency for Republican candidates running in more liberal places to moderate their stances on issues (Ansolabehere, Snyder, and Stewart 2001; Canes-Wrone, Brady, and Cogan 2002; Erikson and Wright 1980, 2000, 2009).¹¹ To the extent that this generalizes across candidates and states, we expect that more conservative states will have candidates of both parties taking relatively conservative positions, and in more liberal states, that the parties’ candidates will take more liberal positions. If so, this suggests the party “brands” that citizens see are to some extent state-dependent. For example, a moderate in a conservative state might witness a pattern of relatively moderate Democratic candidates and conservative Republican candidates, especially compared with his ideological twin sister living in a liberal state. There, the Democratic candidates could afford to indulge liberal activists and supporters and still win (Aldrich and McGinnis 1989) whereas a Republican would clearly be damaged running as a “real conservative.” The real-world examples are easy to come by to make the point. Former Senator Jim DeMint won easily in conservative South Carolina, but would stand not chance of being elected in liberal Massachusetts, where even the moderate Republican Scott Brown could not hold on. The point is an obvious one: candidates do better in their general election efforts if they tailor their appeals toward the median voters in general elections.

Thus, our task is to address three related questions. First, do the state parties systematically vary? If the process of national elite polarization has been sufficient, maybe the differences among the state parties are too little to have a measurable impact. If the Democratic (Republican) parties in the states are ideological mirror images of Democratic (Republican) parties in other states, then we could not expect the state parties to have much of an influence on partisan sorting as they would all be echoing the message of polarization from Washington, D.C. By contrast, the state parties may differ from the national parties to the extent that they are influenced by the ideological preferences of their states.

Second, if there are differences, do these affect the efficiency of the partisan sort across the states? If the state parties in some places offer a clearer choice, then we anticipate that voters in those states will more effectively sort themselves into the ideologically appropriate parties. And, third, we ask whether the relative liberalism-conservatism of the parties in the states effects not only sorting but also the balance of partisanship in the states. Here, we explore the familiar Downsian thesis that electorates prefer parties (candidates) closer to their own ideological positions, so that in the long run, parties positioning themselves further from the median voter should lose supporters. Similarly, more moderate parties (relative to overall state ideology) should gain adherents.¹²

We use the Project Vote Smart National Political Awareness Tests (NPATs) as our source measure of state party elite ideology. We use the full set of responses of more than eighteen thousand Democratic and Republican candidates for both houses of Congress, the state legislature, and governor.¹³ The NPATs are those administered from 1996 to 2006. Each is composed of a long series of questions designed to tap the issues that the candidates are most likely to face. Across the questionnaires were a large number of questions that were administered to all, or almost all, candidates. Where candidates did not answer an item, or where it was not administered in a particular state and year, we used multiple-imputation methods to make up for what would be missing data. The construction of the overall ideology scores for the state parties is detailed in the supplemental appendix (http://prq.sagepub.com/supplemental/).

Our unidimensional measure of state party ideology is derived from factor analyses of all candidates using the imputed NPAT items. The virtue of this measure is that all the candidates and state parties are measured cleanly using the same underlying metric and that metric is intentionally designed to cover the full range of domestic issues candidates are likely to act on.¹⁴ Examples of the NPAT (now termed the “Political Courage Test”) can be found on the Project Vote Smart website (http://www.votesmart.org/).

To begin, we examine the distribution of all candidates by political party affiliation. The party median and interquartile range (IQR; 25th and 75th percentiles) of the first factor scores are shown in Figure 3, sorted in descending order by the liberalism of the state party system. The advantage of the IQR is that it demonstrates the central tendencies of the group—outliers are not presented in this plot. Rather, what is depicted is the middlemost 50 percent of the party: the mainstream.

Figure 3.

Party elite ideology in the states.

The Republican Party medians are on the left, and depicted by the solid triangles, the Democratic Party medians are on the right, and depicted by the hollow squares, and the midpoint of the party system is the smaller dot between the two.

Not surprisingly, we see that Democratic candidates are more liberal than their Republican counterparts within each state and that the most liberal state parties are in the Northeast, while most conservative state parties are in the South. The ideological relationship between the parties offers several interesting patterns worth noting, however. First, although state Democratic (Republican) parties have more similarities than differences with one another, there is considerable state-to-state interparty variation. The most obvious example of this is the contrast of the conservative Republican and moderate Democratic parties in the South with the moderate Republican and liberal Democratic parties in the Northeast.

While the Democratic Party is more liberal than the Republican Party in each state, Figure 3 shows clearly that the state party systems are relatively independent. For example, the 75th percentile for the New York Republican Party is well within the IQR of the Democratic parties in Pennsylvania, Ohio, Missouri, and eighteen other states. In addition, the IQR of the New Jersey Republican Party is almost the same as the IQR of the Mississippi Democratic Party. Similarly, the Arkansas Democratic Party’s 25th percentile is well within the IQR of Republican parties in Utah, South Carolina, Michigan, and twenty-seven other states. Put another way, Michigan Republicans and Arkansas Democrats may have quite a bit in common. Clearly, the ideological characteristics of the political parties vary considerably across states, indicating that the “party brand name” is not the same across the states and regions.

Second, state parties differ in their level of internal ideological heterogeneity. For example, the Oklahoma Republicans have a high level of internal consistency and the Oklahoma Democrats have a low level of internal consistency, whereas the parties in Virginia are similar in their heterogeneity, as evidenced by their nearly identical IQRs. The asymmetry in party brand information in Oklahoma is stark: voters supporting the Republican Party will have a very clear idea of what kinds of policies their candidate will support, while voters supporting the Democratic Party are supporting a more ideologically diffuse set of candidates. By contrast, the party brands appear to be equally as informative in Virginia, where the Republicans and Democrats are both relatively internally homogeneous.

Third, some state party systems are quite polarized, while in others, the two parties are relatively similar to one another ideologically. Among those with little polarization are those states where neither party is terribly extreme, such as North Dakota or Rhode Island as well as those states where one party occupies the ideological center when the opposing party is ideologically extreme, such as Mississippi’s moderate Democratic Party relative to its conservative Republican Party, or New Jersey’s moderate Republican Party relative to its liberal Democratic Party. However, there are several states where there is no convergence to the midpoint at all: the ideological middle is conspicuously vacant in the most polarized states shown at top of the figure, such as California, Colorado, Arizona, Texas, and Minnesota.

This variation among the state parties is at least partially a function of state ideology. Comparing mass ideology (calculated from the 1996 to 2006 CBS/New York Times polls) and state party ideology (mean ideology of each of the state parties from NPAT scores during the same period), we find an obvious relationship between state party ideologies and mass ideological preferences with correlations of .74 for Democratic Party liberalism, and .62 for Republicans. Moreover, if we just look at the overall liberalism of state party systems (the midpoints) and state mass ideology, the correlation is a striking r = .81. It is clear that more liberal states have more liberal party systems and that the ideological profiles of the state Democratic (Republican) parties are not mirror images of one another. Moreover, the magnitude of differences between the parties varies across the states.

Our next question, then, is whether the differences in polarization evident in Figure 3 are picked up by the state electorates? That is, do the state party systems share any culpability in the growing macro sort of the state electorates we have documented above? One indication of whether this is the case is if the differences in party polarization at the state level are reflected in sorting levels among the state electorates. In the contemporary period, there is a healthy correlation between elite polarization (here measured as the difference in mean Democratic and Republican scores on our overall NPAT-based measure of elite ideology) and party/ideology sorting (correlation of party and ideological identifications) in the state electorates at r = .75. While this is a robust relationship, we can push the data a bit further to assess whether changes in elite polarization yield commensurate changes in mass sorting.

To do this, we employ two measures of party elite polarization. One is our NPAT-based measure of state party ideology that covers elites in the contemporary period. The other is from earlier in the period, roughly mid-1970s to mid-1980s (Erikson, Wright, and McIver 1993, Ch. 5).¹⁵

Unfortunately, the metrics of these different measures are not directly comparable. But we can classify states by changes, or lack thereof, in relative state party elite polarization. For this, we separate states into four groups, based on whether the level of elite polarization is above or below the median for the early and contemporary measures, and examine changes in mass party sorting in each of the four groups: states where elite polarization was below the median at both time points (low → low, n = 12); those where polarization was above the median in the early measure but below the median in the contemporary era (high → low, n = 10); states that were initially below the median measure of elite polarization but were above it subsequently (low → high, n = 11); and finally, those states above the median in both time periods (high → high, n = 13).¹⁶

If the state electorates are responsive only to changes in national elites, then we should expect no differences in the rates of partisan shorting. However, if the state parties do influence electorates’ perceptions of the parties, then we would anticipate that those in low polarization states consistently since the 1970s would show lower levels of partisan sorting. Similarly, those exposed to consistently highly polarized elites should show the highest levels of sorting. We also expect the rate of polarization in states that experienced relatively higher rates of polarization to exceed the rate of those states whose elites showed less relative change in polarization over time. Figure 4 shows the partisan sorting in our four groups of states, again classified by presidential administration. Here, we measure sorting as the ratio of “consistents” (liberal Democrats and conservative Republicans) to “inconsistents” (conservative Democrats and liberal Republicans).¹⁷

Figure 4.

Changes in partisan sorting over time by changes in state party polarization.

In Figure 4, all lines are going up, indicating that all groups of states are better sorted and increasingly so with time. Because we do not have independent measures of change in the states and in Washington, D.C., we cannot say how much each level of the party system contributes to the overall rise of mass polarization in the states. Equally important for our interests here, however, are the changes in the levels of polarization among the different groups of states. The differences in the rates of sorting—gaps between the lines in Figure 4—clearly suggest that state party elites matter. Throughout the time period, the states with higher levels of polarization (high → high) are better sorted than those states that are consistently low (low → low). In addition, and most importantly, the groups of states that experienced relative changes in elite polarization by our measures sorted at different rates. Those two groups (high → low and low → high) have virtually the same levels at the beginning of our series, but the low → high group shows an unmistakable faster rate of sorting than the states where changes in elite polarization were relatively less (high → low).

By the contemporary era, the two groups of states above the median in elite polarization are virtually indistinguishable in their levels of partisan–ideological convergence. Here, consistents outnumber inconsistents by about 5.5 to 1. The groups below the median in the contemporary period have ratios that average 4.2 to 1. We believe this demonstrates that the state parties matter and that the electorate responds to cues from both state and national-level party elites. The instance of relative changes in party elite polarization are especially telling in which we see in the separation in rates of the low → high from the high → low and low → low groups.

The only other obvious explanation for this pattern is whether mass polarization leads, rather than follows, the elites. While the idea that elites take their ideological cues from the electorate fits the narrative of Republican democracy, this argument would have to contend with virtually all of the scholarship that has looked at the mass responses to elite polarization, finding that elites move prior to the masses (Abramowitz 2010; Carmines and Stimson 1989; Hetherington 2001; Layman et al. 2010; Levendusky 2009). In short, we find substantial variation in the polarization of state party elites and, moreover, we find a clear pattern of sorting that follows those patterns. The state parties do matter.

The final part of this analysis is to examine whether state party ideology influences party identification. Our initial analyses suggest that if there is an “unmoved mover” (Johnson 2006) in macro-level state politics, it is the relative ideological character states, not citizens’ partisanship. Given the movement of aggregate state partisanship, and with some evidence that state parties do matter for partisan sorting, we ask whether partisanship in the states responds to the general tendencies of the states’ parties. That is, does the liberalism of the states’ parties condition a citizen’s propensity to identify as a Democrat or a Republican?

We believe that even in an era of polarized parties, most citizens are relatively moderate compared with party elites (Fiorina, Abrams, and Pope 2006), and that now as much as ever, people prefer parties and candidates who take positions closer to them, rather than further away. . This means that controlling for mass state ideology, party elites still face a strategic positioning dilemma, as they choose between pleasing their activists’ wings (and perhaps their own strong preferences) and winning the median voter. In a Downsian fashion, movement in a liberal direction for Democratic state parties means movement away from the median voter, while shifts by Republican parties in the states in the same direction indicate movement toward the median voter, or more generally, the average citizen.

In this framework, the average citizen finds herself gauging her preferences relative to the two parties. Or more precisely, she establishes a general expectation over time of how the parties in her state are likely to situate themselves relative to her preferences. We believe that to the extent the state parties matter, if over time she finds the Republican Party’s candidates and positions consistently more pleasing, then she is likely to come to identify as a Republican. This can be accomplished by any combination of the Democratic Party’s candidates situating themselves too far to the left, or the Republican Party moving in toward her preferred positions.

To assess this dynamic in the states, we replicate part of the policy model developed in Statehouse Democracy. This is a simple regression of state partisanship on the midpoint of the party elites (which gauges the relative liberalism-conservatism of the state party system) with a control for state party ideology:

State Partisanship = α + β_{1} (State Ideology) + β_{2} (Party Elite Ideology) + e .

As more liberal publics are going to tend to be Democratic, we expect β₁ in this equation to be positive, and as party elite ideology is measured with higher scores being more liberal, a higher midpoint indicates a more liberal party system. This in turn, when controlling for state ideology, should nudge party identifications toward the Republican Party, or lower scores on state partisanship. Thus, we expect β₂ to be negative. We estimate two regressions, one using contemporary values for state partisanship and elite ideology, and one using lagged values (roughly the early 1980s). The results in Table 2 show the standardized coefficients for current state partisanship regressed on current mass and elite ideology, as well as their lagged (1977–1987) values. The story is the same in both cases: partisanship in the states today is strongly influenced by the ideological preferences of the mass public, whether measured in the current era or with its (highly correlated) values from the earlier period. But most importantly, we see that state party liberalism has the predicted effect on state partisanship. When controlling for mass opinion, states with more liberal party elites have lower (more Republican) levels of partisanship. Moreover, the effects are substantively important: a standard deviation change in the stances of elites is predicted to yield three-quarters of a standard deviation change in state partisanship in the opposite direction. Or, using the older values of mass and elite ideology, slightly over half a standard deviation of change is predicted.

Table 2.

The Effects of Party Elite Ideology on State Partisanship.

Dep Var: Contemporary partisanship (1996–2006)	(1) Contemporary	(2) Lagged
Mass ideology	1.09*** (5.7)	1.00*** (6.1)
Elite ideology	−0.763 (−4.0)	−0.569 (−3.5)
R ²	.43	.48
n	49	46

Standardized coefficients; t-values are in parentheses. Column (1) shows results using contemporary values for mass and elite ideology (1996–2006). Column (2) shows results using lagged values for mass and elite ideology (1977–1987).

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

Discussion and Conclusion

We have shown in this analysis that the story of macro changes in partisanship and ideology in the states is one of remarkable stability in ideology and considerable movement in state partisanship. In the 1980s, there was virtually no relationship between state partisanship and ideology, while today there is a reasonably strong relationship with more liberal states rather consistently being more Democratic and the more conservative states being Republican.

By contrast, work at the micro level has pointed to a pattern opposite what we have found at the macro level: studies of micro change stress the importance and stability of partisanship and the apparent malleability of issues stances, ideology (Carsey and Layman 2006; Levendusky 2009), and even “core values” (Goren 2005; Goren, Federico, and Kittilson 2009). The unexplored implications from the work at the micro level are that the aggregate behavior of the electorate would simply be the sum of behaviors at the individual level, with state changes matching individual changes. In this case, their conclusions would be that states would maintain their aggregate partisanship as the states’ ideologies adjust around their partisan anchors. Rather, the aggregate-level finding is that state ideology is stable while partisanship has changed over time.

Our argument about ideological stability and dynamic partisanship stands in contrast to the party-centric notion of party sorting. While partisanship may be a far more concrete concept for citizens than the abstract concept of ideology, we find that the aggregate behavior of the electorate operates at a fundamentally different level. This is consistent with the Highton and Kam’s (2011) study that explicitly considers changes over longer time periods in the processes of sorting, finding that ideology can shape individual partisanship.

We are unable to analyze the causal mechanism in the process of the macro sort in this study. We suggest, though cannot test, that the mechanism in the process of the macro sort is similar to the models of political change outlined by Carmines and Stimson (1989). That is, the party elites adjusted their positions in a strategic move to appeal to disaffected partisans of the other group. With their worldview unchanged, but the political circumstances adjusting, elites and informed citizens respond to these cues by gradually shifting their partisanship, with the masses following suit through the two-step flow of information.

What if it were the other way? What if partisanship was to serve as the anchor at the state level, and ideology were to adjust? We argue that the result would be changing psychology, more than merely changing politics. It is difficult to imagine a conservative state such as Mississippi becoming more tolerant of same-sex marriage, less hawkish on terrorism, or more willing to initiate immigration reform, without some large-scale change in psychology, fundamental values and outlook, or tremendous population replacement.¹⁸ Accounting for patterns of macro change, fortunately, we do not need to stretch our assumptions about psychology and values being fairly stable. Indeed, the changes in the states have been political and entirely reasonable: state electorates respond to the shifting positions of elites and appropriately modify their partisanship. We suspect that there is a lot of psychological noise in the individual findings and that in this the important (for state electorates) political signal gets lost. Yet, that signal is quite clear when we focus on state electorates. We see a real positive message here, which is that change in the states makes perfect sense, given the familiar political story of party strategies over the last half century.

In addition, our evidence demonstrates clearly that state context matters. Because citizens participate in different state political systems, they do not receive the same messages about what the party systems offer in terms of electoral choice. The heterogeneity of ideological preferences of the states (and the parties’ electorally induced responses to that diversity of opinion) may thus limit the levels of ideological homogeneity achievable by the national parties. This ideological variance in the electorate should provide some counterweight to the attempts by activists to pull parties to the extremes, as doing so would force the parties to risk losing in all but the safest districts.

We suggest that researchers using state-level data to address changes in national-level partisanship would do well to heed the differences in the state parties, both in the elite and mass level. While there are clearly similarities across party systems, the differences across states continue to play an important role in politics. Although the parties have an incentive to cultivate a strong party brand (Snyder and Ting 2002), this is counterbalanced by the needs to appeal to electorates with widely varying ideological predispositions.

We have employed only a simplified unidimensional measure of state ideology here. While there are situations where the unidimensional model accurately captures the dominant political cleavages, there are many situations where economic and cultural dimensions of conflict do not coincide, and these have important implications for political behavior (Ensley, Tofias, and Marchi 2009). This greater complexity in the state electorates may condition the pressures for continued alignment of the states, but this awaits further analysis.

In closing, we note that the growing relationship between ideology and partisanship has hardly reached its upper limit, which indicates that there is considerable room for the two to continue the process of integration. Moreover, given the polarization of party elites on every major policy dimension—economic, social, and foreign—we expect the relationship between state ideology and partisanship to continue to tighten as long as party elites continue to polarize. The parties have long shown patterns of change with the current era representing a high point (to date) in polarization. If that holds or even increases, the mass electorate will continue to sort and the macro-level manifestations of that are redder red states and bluer blue states need to be explored. What we have accomplished here, we hope, is to firmly establish that the explanation of macro-level change as largely political. Individual attitude change may be explained primarily in terms of psychological processes, but story of change among state electorates is one of voters responding to the signals of party elites and activists, and adjusting their partisan attachments to match long-standing ideological, or even cultural, predispositions.

Footnotes

Acknowledgements

Birkhead gratefully acknowledges support from the Center on American Politics at Indiana University. The authors thank Ian Anson, Tom Carsey, and Steve Rogers for helpful comments.

Authors’ Note

The party elite data were collected and processed under a grant from the Russell Sage Foundation. The data on the mass electorate were initially collected under a grant from the National Science Foundations. Revised version of a paper initially prepared for the State Politics and Policy Conference, Rice University/University of Houston, February 16–18, 2012, Houston, Texas.

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

Our data on state party elites were collected with support from grants from the National Science Foundation and the Russell Sage Foundation. Early collection of the CBS/New York Times data series was supported by grants from the National Science Foundation.

Notes

References

Abramowitz

Alan I.

2010. The Disappearing Center: Engaged Citizens, Polarization, and American Democracy. New Haven: Yale University Press.

Abramowitz

Alan I.

Saunders

Kyle L.

1998. “Ideological Realignment in the American Electorate.” Journal of Politics 60 (3): 634-52.

Abramowitz

Alan I.

Saunders

Kyle L.

2006. “Exploring the Bases of Partisanship in the American Electorate: Social Identity vs. Ideology.” Political Research Quarterly 59 (2): 175–87.

Abramowitz

Alan I.

Saunders

Kyle L.

2008. “Is Polarization a Myth?” The Journal of Politics 70 (2): 542–55.

Achen

Christopher H.

2002. “Parental Socialization and Rational Party Identification.” Political Behavior 24 (2): 151–70.

Achen

Christopher H.

Shively

W. Phillips.

1995. Cross-Level Inference. Chicago: University of Chicago Press.

Aldrich

John H.

1983. “A Downsian Spatial Model with Party Activism.” The American Political Science Review 77 (4): 974–90.

Aldrich

John H.

McGinnis

Michael D.

1989. “A Model of Party Constraints on Optimal Candidate Positions.” Mathematical and Computer Modelling 12 (4-5): 437–450.

Ansolabehere

Stephen

Snyder

James M.

Jr. Stewart

Charles

III . 2001. “Candidate Positioning in U.S. House Elections.” American Journal of Political Science 45 (1): 136–59.

10.

Bafumi

Joseph

Shapiro

Robert Y.

2009. “A New Partisan Voter.” The Journal of Politics 71 (1): 1–24.

11.

Bartels

Larry M.

1996. “Pooling Disparate Observations.” American Journal of Political Science 40 (3): 905–42.

12.

Bartels

Larry M.

2000. “Partisanship and Voting Behavior, 1952-1996.” American Journal of Political Science 44 (1): 35–50.

13.

Bartels

Larry M.

2002. “Beyond the Running Tally: Partisan Bias in Political Perceptions.” Political Behavior 24 (2): 117–50.

14.

Bishop

Bill

. 2008. The Big Sort: Why the Clustering of Like-Minded America is Tearing Us Apart. Boston: Houghton Mifflin.

15.

Black

Earl

Black

Merle

. 2002. The Rise of Southern Republicans. Cambridge: Harvard University Press.

16.

Brady

David W.

Han

Hahrie

Pope

Jeremy C.

2007. “Primary Elections and Candidate Ideology: Out of Step with the Primary Electorate?” Legislative Studies Quarterly 32 (1): 79-105.

17.

Brewer

Mark D.

Mariani

Mack D.

Stonecash

Jeffrey M.

2002. “Northern Democrats and Party Polarization in the U.S. House.” Legislative Studies Quarterly 27 (3): 423–44.

18.

Campbell

Angus

Converse

Philip E.

Stokes

Donald E.

Miller

Warren E.

1960. The American Voter. New York: Wiley.

19.

Canes-Wrone

Brandice

Brady

David W.

Cogan

John F.

2002. “Out of Step, Out of Office: Electoral Accountability and House Members’ Voting.” The American Political Science Review 96 (1): 127–40.

20.

Carmines

Edward G.

1991. “The Logic of Party Alignments.” Journal of Theoretical Politics 3 (1): 65–80.

21.

Carmines

Edward G.

Stimson

James A.

1989. Issue Evolution: Race and the Transformation of American Politics. Princeton: Princeton University Press.

22.

Carsey

Thomas M.

Layman

Geoffrey C.

2006. “Changing Sides or Changing Minds? Party Identification and Policy Preferences in the American Electorate.” American Journal of Political Science 50 (2): 464–77.

23.

Claassen

Ryan L.

2011. “Political Awareness and Partisan Realignment: Are the Unaware Unevolved?” Political Research Quarterly 64 (4): 818–30.

24.

Converse

Philip

. 1964. “The Nature of Belief Systems in Mass Publics.” In Ideology and Discontent, edited by Apter

David E.

, 206–61. New York: Free Press.

25.

Cotter

Cornelius C.

Gibson

James L.

Bibby

John F.

Huckshorn

Robert J.

1989. Party Organizations in American Politics. Pittsburgh: University of Pittsburgh Press.

26.

Ensley

Michael J.

Tofias

Michael W.

de Marchi

Scott

. 2009. “District Complexity as an Advantage in Congressional Elections.” American Journal of Political Science 53 (4): 990–1005.

27.

Ensley

Michael J.

2012. “Incumbent Positioning, Ideological Herogeneity and Mobilization in U.S. House Elections” Public Choice 151 (1): 43-61.

28.

Erikson

Robert S.

MacKuen

Michael

Stimson

James A.

2002. The Macro Polity. New York: Cambridge University Press.

29.

Erikson

Robert S.

Wright

Gerald C.

1980. “Policy Representation of Constituency Interests.” Political Behavior 2 (1): 91–106.

30.

Erikson

Robert S.

Wright

Gerald C.

2000. “Representation of Constituency Ideology in Congress.” In Continuity and Change in U.S. House Elections, edited by Brady

David

Cogan

John F.

Fiorina

Morris P.

, 148–77. Stanford: Stanford University Press.

31.

Erikson

Robert S.

Wright

Gerald C.

2009. “Voters, Candidates, and Issues in Congressional Elections.” In Congress Reconsidered, edited by Dodd

Lawrence C.

Oppenheimer

Bruce I.

, 71–95. Washington, DC: CQ Press.

32.

Erikson

Robert S.

Wright

Gerald C.

McIver

John P.

1993. Statehouse Democracy: Public Opinion and Policy in the American States. New York: Cambridge University Press.

33.

Erikson

Robert S.

Wright

Gerald C.

McIver

John P.

2006. “Public Opinion in the States: A Quarter Century of Change and Stability.” In Public Opinion in the States, edited by Cohen

Jeffrey E.

, 229–53. Stanford: Stanford University Press.

34.

Fiorina

Morris P.

1981. Retrospective Voting in American National Elections. New Haven: Yale University Press.

35.

Fiorina

Morris P.

Abrams

Samuel J.

Pope

Jeremy

. 2006. Culture War? The Myth of a Polarized America. 2nd ed. New York: Pearson Longman.

36.

Franklin

Charles H.

1992. “Measurement and the Dynamics of Party Identification.” Political Behavior 14 (3): 297–309.

37.

Franklin

Charles H.

Jackson

John E.

1983. “The Dynamics of Party Identification.” The American Political Science Review 77 (4): 957–73.

38.

Gimpel

James G.

1996. National Elections and the Autonomy of American State Party Systems. Pittsburgh: University of Pittsburgh Press.

39.

Goren

Paul

. 2005. “Party Identification and Core Political Values.” American Journal of Political Science 49 (4): 881–96.

40.

Goren

Paul

Federico

Christopher M.

Kittilson

Miki C.

2009. “Source Cues, Partisan Identities, and Political Value Expression.” American Journal of Political Science 53 (4): 805–20.

41.

Green

Donald

Palmquist

Bradley

Schickler

Eric

. 2002. Partisan Hearts and Minds. New Haven: Yale University Press.

42.

Groenendyk

Eric

. 2012. “Justifying Party Identification: A Case of Identifying with the “Lesser of Two Evils.” Political Behavior 34 (3): 453–75.

43.

Hadley

Charles D.

1985. “Dual Partisan Identification in the South.” The Journal of Politics 47 (1): 254–68.

44.

Hetherington

Marc J.

2001. “Resurgent Mass Partisanship: The Role of Elite Polarization.” The American Political Science Review 95 (3): 619–31.

45.

Highton

Benjamin

Kam

Cindy D.

2011. “The Long-Term Dynamics of Partisanship and Issue Orientations.” The Journal of Politics 73 (1): 202–15.

46.

Iyengar

Shanto

Hahn

Kyu S.

2009. “Red Media, Blue Media: Evidence of Ideological Selectivity in Media Use.” Journal of Communication 59 (1): 19–39.

47.

Jacobson

Gary C.

2007. “Referendum: The 2006 Midterm Congressional Elections.” Political Science Quarterly 122 (1): 1–24.

48.

Jennings

M. Kent

Niemi

Richard G.

1966. “Party Identification at Multiple Levels of Government.” American Journal of Sociology 72 (1): 86–101.

49.

Johnson

Richard

. 2006. “Party Identification: Unmoved Mover or Sum or Preferences?” Annual Review of Politics 9:329–51.

50.

Johnson

Richard

Hagen

Michael G.

Jamieson

Kathleen H.

2004. The 2000 Presidential Election and the Foundations of Party Politics. Cambridge: Cambridge University Press.

51.

Knuckey

Jonathan

. 2006. “Explaining Recent Changes in the Partisan Identifications of Southern Whites.” Political Research Quarterly 59 (1): 57–70.

52.

Layman

Geoffrey C.

1997. “Religion and Political Behavior in the United States: The Impact of Beliefs, Affiliations, and Commitment from 1980 to 1994.” The Public Opinion Quarterly 61 (2): 288–316.

53.

Layman

Geoffrey C.

Carsey

Thomas M.

2002. “Party Polarization and Party Structuring of Policy Attitudes: A Comparison of Three NES Panel Studies.” Political Behavior 24 (3): 199–236.

54.

Layman

Geoffrey C.

Carsey

Thomas M.

Green

John C.

Herrera

Richard

Cooperman

Rosalyn

. 2010. “Activists and Conflict Extension in American Party Politics.” The American Political Science Review 104 (2): 324–46.

55.

Layman

Geoffrey C.

Carsey

Thomas M.

Horowitz

Juliana M.

2006. “Party Polarization in American Politics: Characteristics, Causes, and Consequences.” Annual Review of Political Science 9:83–110.

56.

Levendusky

Matthew

. 2009. The Partisan Sort: How Liberals Became Democrats and Conservatives Became Republicans. Chicago: University of Chicago Press.

57.

Lublin

David

. 2004. The Republican South: Democratization and Partisan Change. Princeton: Princeton University Press.

58.

MacKuen

Michael B.

Erikson

Robert S.

Stimson

James A.

1989. “Macropartisanship.” The American Political Science Review 83 (4): 1125–42.

59.

Markus

Gregory B.

Converse

Philip E.

1979. “A Dynamic Simultaneous Equation Model of Electoral Choice.” The American Political Science Review 73 (4): 1055–70.

60.

Miller

Gary

Schofield

Norman

. 2003. “Activists and Partisan Realignment in the United States.” The American Political Science Review 97 (2): 245–60.

61.

Miller

Warren E.

Shanks

J. Merrill

. 1996. The New American Voter. Cambridge: Harvard University Press.

62.

Schofield

Norman

Miller

Gary

. 2007. “Elections and Activist Coalitions in the United States.” American Journal of Political Science 51 (3): 518–31.

63.

Shafer

Byron E.

Johnston

Richard

. 2006. The End of Southern Exceptionalism: Class, Race, and Partisan Change in the Postwar South. Cambridge: Harvard University Press.

64.

Snyder

James M.

Jr. Ting

Michael M.

2002. “An Informational Rationale for Political Parties.” American Journal of Political Science 46 (1): 90–110.

65.

Stoker

Laura

Jennings

M. Kent

. 2008. “Of Time and the Development of Partisan Polarization.” American Journal of Political Science 52 (3): 619–35.

66.

Sundquist

James L.

1983. Dynamics of the Party System: Alignment and Realignment of Political Parties in the United States. Washington, DC: Brookings Institution Press.

67.

Tam Cho

Wendy K.

Gimpel

James G.

Hui

Iris S.

2013. “Voter Migration and the Geographic Sorting of the American Electorate.” Annals of the Association of American Geographers 103 (4): 856–70.

68.

Wright

Gerald C.

1989. “Level-of-Analysis Effects on Explanations of Voting: The Case of the 1982 US Senate Elections.” British Journal of Political Science 19 (3): 381–98.

69.

Zaller

John

. 1992. The Nature and Origins of Mass Opinion. New York: Cambridge University Press.