Incumbent ideology, district ideology, and candidate entry in U.S. congressional elections, 1954–2008

The Social Science Journal 51 (2014) 181–190

Contents lists available at ScienceDirect

The Social Science Journal journal homepage: www.elsevier.com/locate/soscij

Incumbent ideology, district ideology, and candidate entry in U.S. congressional elections, 1954–2008夽 Nicholas L. Pyeatt ∗ Pennsylvania State University—Altoona, 3000 Ivyside Park, Altoona, PA 16601, USA

a r t i c l e

i n f o

Article history: Received 7 March 2013 Received in revised form 12 February 2014 Accepted 13 February 2014 Available online 26 March 2014

Keywords: Candidate entry Congress Elections Ideology

a b s t r a c t This paper investigates the role of incumbent ideology on quality candidate entry in U.S. congressional elections. Looking at district ideology, the analysis shows that members of the House and Senate face different ideological incentives in terms of opponent quality. For officials from highly ideological districts, greater ideological extremity is advantageous as it discourages strong opponents. For members from cross pressured districts, however, greater extremity may increase the likelihood of facing a strong opponent. Although the magnitude of these findings vary by party, they illustrate that incumbents face diverging ideological motivations in terms of discouraging strong challengers based on the ideology of their district. © 2014 Western Social Science Association. Published by Elsevier Inc. All rights reserved.

1. Introduction Whether or not an incumbent faces a quality challenger matters to election scholars because of the strong link between candidate quality and election outcomes (Jacobson, 1989; Squire, 1992). While incumbents are notoriously difficult to defeat under most circumstances, challenger quality plays a major role in the election result. The central interest of this paper is the relationship between incumbent ideology and quality challenger entry. Relatively few works directly link challenger quality with

夽 Previous versions of this work have been presented at American Political Science Association Annual Meeting (2008) and the Midwest Political Science Association Annual Meeting (2009). The author would like to thank George Rabinowitz, Erik Engstrom, David W. Rohde, Jason M. Roberts, Thomas Carsey, Adam Chamberlain, and Patrick C. Wohlfarth for their helpful comments on versions of this piece. Finally, the author would like to thank Gary Jacobson for sharing his data on US House elections. ∗ Tel.: +1 814 949 5752; fax: +1 814 949 5161. E-mail address: [email protected]

incumbent ideology. Research from Bond, Covington, and Fleisher (1985), looking at the House, finds that greater ideological distance increases the potential for challenge; looking at the Senate, Adams and Squire (1997) find a very limited role for incumbent ideology on entry. While work on electoral outcomes such as Canes-Wrone, Brady, and Cogan (2002) find that more extreme incumbents received a lower general election vote percentage, the question remains whether incumbent positioning can affect the preceding step, the quality of challengers that incumbents face. As challenger quality strongly influences election outcomes, it is important to better understand the role of an incumbent’s ideology on challenger emergence. As quality challengers likely enter strategically, it is also important to look at the interplay between incumbent behavior and district ideology. Specifically, as district ideology clearly predicts election outcomes, quality opponents will likely consider district ideology when making their entry decisions. The expectation is that greater incumbent extremity serves to discourage strong opponents when the incumbent’s ideology and the district’s ideology are

http://dx.doi.org/10.1016/j.soscij.2014.02.003 0362-3319/© 2014 Western Social Science Association. Published by Elsevier Inc. All rights reserved.

182

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

congruent.1 In these districts, the majority of modern congressional seats, a Republican incumbent’s conservative signals in a conservative district, for example, should discourage quality Democratic opponents from entering. Conversely, for an incumbent oriented counter to the district’s ideological direction, such as a liberal Democrat representing a somewhat conservative district, ideological extremity should be a costly behavior encouraging quality candidates to challenge. Thus, the expectation is that the effect of incumbent ideology on candidate entry is conditional on district ideology. If the empirical models support these expectations, as the majority of districts are congruent, the electoral incentive for incumbents to moderate in terms of challenger quality will be limited to relatively few members of Congress. This study finds, by looking at House and Senate elections from 1954 to 2008, a conditional relationship between ideological extremity, district ideology, and quality challenger entry. After accounting for standard political and incumbent factors, the majority of congressional incumbents have no greater likelihood of facing quality opponents if they become more ideologically extreme. Only a small subset of congresspersons from places where district ideology runs counter to the incumbent’s ideology face an increased likelihood of being challenged by strong opponents based on greater ideological extremity. These findings offer the discipline additional understanding of incumbents’ strategic considerations and could offer some leverage over modern congressional behavior, such as polarization.

2. Candidate emergence and incumbent characteristics To understand the influence of incumbent ideology on entry, it is necessary to examine the decision making process of potential, high quality candidates. In their seminal work, Jacobson and Kernell (1983) find that strategic politicians enter when they see their chance of victory as the highest. Their work modeled the entry decision calculus as one where potential candidates measured the probability of winning against the value of the seat and cost of the election. This analysis of the costs and benefits of entry remains widely utilized, although more recent work has modeled ambition as a preceding step in the entry process (Maestas, Fulton, Maisel, & Stone, 2006). Looking at candidate entry, Bianco (1984) finds challengers are more likely to emerge when incumbents do less well in terms of previous vote share (Bianco, 1984; Jacobson, 1989; Krasno & Green, 1988; Squire, 1989). As Krasno and Green note, “[T]he cost of a tight election for

an incumbent is the increased probability of facing an even tougher challenger (if possible) the next time around, in turn leading to another close call, or worse” (1988, p. 932). Thus, previous incumbent vote share is critical to potential entrants because it both reflects incumbent strength and district partisanship (Adams & Squire, 1997; Kazee, 1983; Maisel & Stone, 1997). Furthermore, Bianco finds that challengers are more likely to emerge from the in-party when the economy does well and challenges are more likely from the outparty when the economy does poorly (1984). Finding that national events drive House entry parallels Jacobson’s work on candidate emergence and strategic challengers (1989). Jacobson finds that more qualified candidates are likely to run when the national conditions favor their party, specifically when personal income or presidential popularity change (Born, 1986). These results are particularly important given that Jacobson finds clear evidence that quality challengers are more likely to lead to incumbent defeats. In Senate elections, a weaker relationship has been found between entry decisions and national political and economic factors (Lublin, 1994; Stewart, 1989), perhaps resulting from the longer time period between elections. While potential challengers react to the national political environment and the incumbent’s political strength, the focus here is on the role of incumbent ideology on entry decisions. Overall, the relationship between challenger entry and incumbent ideology has received relatively limited scholarly attention. The chief finding, from Bond et al. (1985), discovers a moderate effect for ideological distance in terms of promoting potential challenges. Their work found that previous incumbent percentage and district partisanship both strongly predict an incumbent facing a quality challenger but that ideological distance played a more modest role. Members further from the district’s median ideology were more likely to face better financed and more qualified challengers, although not necessarily more politically experienced candidates. Conversely, looking at the Senate, Adams and Squire (1997) find that incumbent ideology does not play a major role in an incumbent senator facing a quality challenger. While Bond et al. (1985) do not find that greater ideological distance leads to an increased probability of facing an experienced opponent, they do find that when incumbents are out of step with their districts’ voters, they increase their likelihood of facing a stronger opponent. This strongly suggests that higher quality opponents react to the relationship between an incumbent’s ideological position and the district’s political characteristics.

3. Expectations 1 Throughout this paper the word district will be used to denote the polities that elect members of the House and Senate. While Senators get elected at the state level and House members at the congressional district level, referring to the units that elect members as districts/states or electoral units seemed less clear than referring to all units as districts. In others words, this paper will use district in a general way to mean a geographically bounded political unit, rather than specifically a congressional district.

Fundamentally, this piece incorporates ideology into the Jacobson and Kernell candidate entry framework (1983). Their model describes the likelihood of entry as based on the probability of victory, the value of the seat in question, and the cost of the election (Black, 1972). Central to the current analysis is the idea that incumbent ideology and its congruence with district ideology has the potential

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

to influence the probability of challenger victory and thus change the likelihood of candidate entry.2 3.1. Probability of victory Strategic political elites considering whether to challenge an incumbent likely consider not only the incumbent’s ideology but also district ideology. Specifically, if the district’s ideology generally corresponds with the incumbent’s ideology, then these incumbents will project an ideological orientation attractive to the public. So the probability of quality challenger entry should decrease with ideological extremity in this type of district. Conversely, in places where the district’s ideology does not align with the incumbent’s ideological direction, greater ideological extremity should encourage quality challengers from the opposing party to enter. As the district is not inclined toward the incumbent’s policy direction, greater ideological extremity will make it more difficult for the incumbent to retain the district, even if other factors, such as visibility and constituency services are favorable. As the probability of the challenger winning increases under this scenario, quality challengers are more inclined to run for office. In both cases, political elites use information about public preferences and the incumbent’s visible ideology to determine the likelihood of victory. The theoretical assumption underlying both expectations is that political elites base their decisions on the relationship between the incumbent’s ideology and the district’s ideology. This assumption seems plausible as all three primary voting models from political science research expect the ideology of the district’s median voter and the incumbent’s ideology to affect the final vote margin and the likelihood of victory. In proximity theory (Davis, Hinich, & Ordeshook, 1970; Downs, 1957), the most common voting model, candidates receive the maximum electoral benefit from the ideological position nearest to the median voter in the district. Thus, for members representing highly ideological districts, greater ideological extremity is a good electoral strategy as it represents better congruence with the district. In less ideological districts, moderation is a better strategy as becoming more extreme would lead to increased distance from the median voter. While directional theory (Rabinowitz & Macdonald, 1989) and discounting theory (Grofman, 1985; Kedar, 2005) both use an ideal point some distance from the district’s median voter, both models expect relative congruence with the median voter. In directional theory, for instance, while the ideal position is the most extreme acceptable position on the same ideological side as the median voter, candidates still must react to the district median. A moderately conservative district would, under directional theory, find a liberal candidate unacceptable just as it would under proximity theory. While the analysis that follows does not attempt to test the voting theories, the expectation is that strategic elites would

2 The Jacobson and Kernell model will be used in this paper as way to conceptualize the potential predictors rather than as a direct use of the model.

183

consider the relationship between the incumbent’s ideological position and the district’s ideology. Therefore, as each of the three main voting models assume some level of incumbent-district ideological congruence, it is reasonable that strategic elites will be less likely to enter when the incumbent and district ideology are generally congruent. While ideological positioning is the principle interest of this piece, all relevant factors must be considered. As previously observed, district ideology strongly predicts challenger emergence. An incumbent holding a district oriented in their ideological direction should have a higher probability of victory than a member representing a more moderate seat. Therefore, the more ideological the district is in the incumbent’s direction, the less likely a quality challenger should emerge. Thus, district ideology should have both an independent effect and one in concert with incumbent ideology. Freshmen members, facing their first re-election effort, should see more quality challengers than experienced members. A freshman member will have the smallest possible incumbency advantage and should offer a potential challenger a higher probability of victory. Although female members have been found to win reelection as often as men, previous work has shown them to face a more competitive electoral environment in primary and general elections (Lawless & Pearson, 2008; Palmer & Simon, 2005, 2008). Thus, women in Congress may face more quality challengers than their male counterparts. Some incumbents have more political power than others. Therefore, one would expect that members with stronger political support would be less likely to draw challengers than comparatively weaker members. While multiple ways exist to consider incumbent strength, one straightforward method is to look at their previous electoral success. Members who won their previous election by larger margins should see a lower probability of facing a quality challenger than less successful members. 3.2. Value of the seat The value of the seat is harder to determine than the probability of winning. Potential candidates have a geographically constrained ability to seek seats. A rough equality of benefits also exists across congressional seats in terms of staff, salary, etc. Most differentiation between seats, such as office space and other institutional perks, is determined in Washington and not directly related to the seat in question. However, as most strategic elites would more highly value a seat in the majority party, minority party members should be more likely to face quality challengers. 3.3. Cost of the election Numerous factors can influence election costs. As Jacobson (1989) observes, challengers strongly respond to the overall state of the country, particularly the economy and presidential popularity. Thus, when adjusted by the incumbent’s party, presidential approval, income change, and midterm elections should all have the ability to predict quality challenger entry. For members of the president’s party, higher presidential approval ratings should decrease

184

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

the chance of facing a strong challenger while lower ratings should make challenges more likely. Robust income growth or decline would similarly alter the electoral calculus for potential candidates. Finally, it has been widely observed that midterm elections generally see seat losses for the president’s party.3 These three factors, adjusted by party, should all influence the emergence of a quality challenger. 4. Data and measurement The unit of analysis is House and Senate general elections between 1954 and 2008 where an incumbent sought re-election. While important scholarship exists about the influence of incumbent ideology on strategic retirements (Stone, Fulton, Maestas, & Maisel, 2010), this analysis focuses only on members who chose to seek re-election. As it excludes members who did not run for re-election due to being out of sync with their electorate, it is fundamentally conservative about the impact of incumbent ideology on quality challenger entry.4 The only elections excluded are House seats redistricted since the last election, races where the two-party presidential vote percentage was not available by congressional district for the decade, and races from Louisiana. The latter were excluded as that state’s unique nonpartisan November primary cannot be easily compared to the elections in all other states. The principal dependent variable is a dichotomous measure of whether the incumbent faced a quality challenger, conceptualized using the Jacobson definition of whether the challenger previously held elected office. While holding previous office is an imperfect measure of candidate strength, this quality measure is both easy to conceptualize and available for the entire period. Furthermore, it has been widely observed that candidates with previous electoral experience raise more money, run more competitive campaigns, and are more likely to defeat incumbents than candidates without political experience. For the Senate, where electoral competition tends to be higher, an ordinal measure of candidate quality developed by Lublin (1994) will also be used as a secondary test.5 This measure has an effect on electoral margins in the Senate and is coded as no candidate or one without elected experience (0), a local elected official (1), a state legislator (2), a governor, other statewide elected official, a congressperson or a former senator (3).6

3

The recent exceptions to the general trend have been the midterm elections of 1998 and 2002. 4 It also should be noted that the work of Stone et al. (2010) focuses primarily on incumbent election percentages and personal vote, two factors related to but separate from the focus of this piece. 5 A similar measure was considered but ultimately rejected for the House. In the House, roughly 20% of all incumbents face quality challengers while in the Senate that number is almost three times higher (57%). This suggests that in the Senate, knowing that the candidate had been elected to a previous office provides less information about the strength of the candidate than in the House. Also, in the House, detailed candidate quality information is unavailable in a consistent manner over the entire time frame. 6 This coding scheme combines the top categories used by Lublin. This was done because the initial ordered logistic regression models violated

The two key predictors in the analysis are incumbent Ideological extremity and District ideology. The interaction of these variables will determine the effect of incumbent ideology as the district becomes more ideological in the incumbent’s direction. Ideology is measured using the incumbent’s first dimension DW-NOMINATE score.7 An adjustment is made to these ideological variables so that higher values reflect the general ideological direction of their party. For illustration, the most liberal Democrat and the most conservative Republican would both score roughly one on Ideological extremity. To measure District ideology, the average two-party vote percentage received by the incumbent’s presidential candidates in that decade is included.8 Using presidential vote to measure district ideology is common in the literature (Ansolabehere, Snyder, & Stewart, 2001; Buttice & Stone, 2012; Canes-Wrone et al., 2002; Erikson & Wright, 1980) and allows for a straightforward comparison between districts.9 To illustrate this measure, for a Republican-held district in 1998, District ideology would be determined based on the average of George H. W. Bush’s vote in 1992, Bob Dole’s vote in 1996, and George W. Bush’s vote in 2000. The results fully replicate when using a single presidential election to measure district ideology and, in fact, the magnitude of the critical variables increases in those models.10 However, the decade measure was preferable as it reduces the election to election variation that naturally occurs in presidential elections. Also, House districts boundaries usually get redrawn every ten years, while states effectively never change, so this allows for a time period common to both the House and the Senate. Freshman identifies members seeking re-election for the first time. Female identifies female members of Congress. Income change is the percentage change in personal disposable income, as computed by the Bureau of Economic Advisors, between the first and second quarters of the year of the election. Choosing which economic period to use presented some difficulty as, depending on state filing deadlines, candidates may decide to run at various times. The ultimate decision was to pick the quarter concurrent with the largest number of candidates announcing their intention to enter a congressional race. All analyses have been computed using alternative economic measures, including using annual changes and quarters earlier and later in the election cycle, and the results remain

the proportional odds assumption. Combining the top two categories solves the statistical problem. Theoretically, this change still represents increasing political experience and more competitive Senate challengers. 7 Models using ADA scores have also been computed and show substantively identical results. 8 There may be some concern that the District ideology and Ideological extremity measures correlate too highly. The two variables do correlate but their relationship is modest, less than .45 in the House models and less than .3 in the Senate models. 9 A recent work has used a large number of surveys in combination to assess district ideology (Tausanovitch & Warshaw, 2013). This sophisticated methodological approach leads to similar conclusions about district ideology (a correlation of roughly .9 with presidential vote) but would be impossible to replicate for the entire period under investigation in this paper. 10 The author will provide those models upon request.

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

185

Table 1 The effect of ideological extremity on the probability of facing a quality opponent 1954–2008 (DW-NOMINATE). I House Key variables Ideological extremity District ideology Control variables Freshman Female Income change Prev. incumbent percentage Incumbent majority party Pres. approval Midterm Constant Constant cut 1 Constant cut 2 Constant cut 3 N Pseudo R2

II Senate

0.01 (0.22) −0.04 (0.01)*** 0.33 (0.08)*** 0.05 (0.13) −0.13 (0.04)*** −0.06 (0.00)*** −0.51 (0.09)*** −0.00 (0.00) 0.40 (0.05)*** 4.80 (0.33)*** – – – 6276 0.13

III Senate ordinal quality

−0.67 (0.45) −0.02 (0.01)**

−0.49 (0.40) −0.02 (0.01)*

−0.22 (0.18) 0.31 (0.41) 0.18 (0.09)** −0.07 (0.01)*** −0.27 (0.17) −0.01 (0.01)* 0.01 (0.13) 6.68 (0.89)*** – – – 685 0.09

−0.08 (0.16) 0.18 (0.34) 0.08 (0.08) −0.07 (0.01)*** −0.15 (0.15) −0.01 (0.01)** 0.17 (0.12) – −6.11 (0.80)*** −5.25 (0.79)*** −4.90 (0.79)*** 685 0.05

Note: The universe of cases is all congressional elections from 1954 to 2008 outside of Louisiana where an incumbent sought re-election, the seat has not been redistricted since the last election, and ideological scores are available for the incumbent. The dependent variable in Models I and II is whether the incumbent faces a quality challenger (0–1) and in Model III it is a scale of challenger quality (0–3). Accordingly, Models I and II are logistic regressions and Model III is an ordinal logistic regression. Predictors in italics have been adjusted by party. * p < .10, ** p < .05, *** p < .01.

substantively identical. Income change is adjusted by party. If the incumbent comes from the president’s party and income increases, it is positive; the variable is also positive if income decreases and the incumbent does not come from the president’s party. Previous incumbent percentage, an important control for incumbent success, measures the two-party vote percentage that the incumbent received in the previous congressional election. Incumbent majority party reflects whether the incumbent is a member of the majority party in Congress. Presidential approval measures the president’s job approval using the Gallup Poll closest to March 1st of the election year. This variable is multiplied by negative one if the incumbent congressperson does not come from the president’s party. Midterm reflects whether or not the election year is a midterm, a non-presidential year. If the incumbent comes from the president’s party, this variable is coded as a one; for opposing party members this variable takes on the value of negative one. It should be noted that all subsequent analyses exclude time and incumbent fundraising as predictors of opposing candidate quality. While year to year variation in the number of quality candidates that emerge definitely exists, there is no apparent theoretical reason to believe that time would affect the central relationship in a systematic way. Therefore, for the sake of parsimony, time was excluded as a predictor. However, all models have been alternatively computed with time and no results change in a substantive way.11 The decision to exclude incumbent fundraising was made consciously based on Goodliffe’s (2001, 2004, 2007) research about the link between incumbent war chests and candidate emergence. Fundamentally, the assumption that an incumbent, even a damaged and vulnerable one, can raise the necessary funds to compete seems reasonable. As such, the precise amount of incumbent fundraising is

11

The author will provide those models upon request.

largely endogenous to the race at hand and, at least in part, the quality of opponent that emerges or is expected to emerge. Moreover, using such data would limit the time period under investigation to 1980 and later. Alternative versions of all models presented were run with incumbent fundraising included and the findings remain substantively comparable. 5. Analysis The analysis will begin by investigating the influence of incumbent ideology and district ideology on quality candidate entry. Table 1 includes models with all key predictors, but Ideological extremity and District ideology will not be interacted at this stage to observe the independent effects of the variables. In this analysis, Ideological extremity does not have a clear effect on the likelihood of the incumbent facing a quality opponent in any model. In other words, more ideologically extreme incumbents are no more or less likely to face quality challengers. Districts more ideological in the incumbent’s direction, however, have fewer quality challenges to incumbents. This is a reasonable finding as strategic challengers are less likely to enter where district ideology makes their electoral odds less favorable.12

12 It should be noted that the observed effect for district ideology differs from Goodliffe (2007), who found no effect on quality candidate entry in Senate elections for ideological distance or state partisanship. While that paper’s broader focus was the effect of incumbent fundraising, it asked a similar empirical question to this paper about the effect of various factors on quality challenger entry. While it might be ideal to compare those results directly to the ones presented here, major differences in modeling make the two results difficult to compare. For example, in Goodliffe’s work ideological distance is measured using the approach described in Abramowitz (1988), based on interest group ratings and a measure of state ideology (Erikson et al., 1993). Conversely, this analysis models district ideology using presidential voting over a decade, an approach also common in the literature (Ansolabehere et al., 2001). More importantly, the measure used by Goodliffe (2007) is not available for the House of

186

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

Table 2 The effect of ideological extremity on the probability of facing a quality opponent 1954–2008 (DW-NOMINATE). I House Key variables Ideological extremity District ideology Ideological extremity* District ideology Control variables Freshman Female Income change Prev. incumbent percentage Incumbent majority party Pres. approval Midterm Constant Constant cut 1 Constant cut 2 Constant cut 3 N Pseudo R2

II Senate

4.92 (1.15)*** −0.01 (0.01) −0.09 (0.02)***

0.34 (0.08)*** 0.08 (0.13) −0.13 (0.04)*** −0.06 (0.00)*** −0.50 (0.09)*** −0.00 (0.00) 0.39 (0.05)*** 3.18 (0.49)*** – – – 6276 0.13

III Senate ordinal quality

6.13 (3.11)** 0.02 (0.02) −0.13 (0.06)**

4.99 (2.80)* 0.02 (0.02) −0.10 (0.05)**

−0.22 (0.18) 0.30 (0.41) 0.19 (0.09)** −0.07 (0.01)*** −0.30 (0.17)* −0.01 (0.01)* 0.01 (0.13) 4.56 (1.28)*** – – – 685 0.10

−0.07 (0.16) 0.18 (0.34) 0.08 (0.08) −0.07 (0.01)*** −0.17 (0.15) −0.01 (0.01)** 0.17 (0.12) −4.41 (1.16)*** −3.55 (1.16)*** −3.20 (1.16)*** 685 0.05

Note: The universe of cases is all congressional elections from 1954 to 2008 outside of Louisiana where an incumbent sought re-election, the seat has not been redistricted since the last election, and ideological scores are available for the incumbent. The dependent variable in Models I and II is whether the incumbent faces a quality challenger (0–1) and in Model III it is a scale of challenger quality (0–3). Accordingly, Models I and II are logistic regressions and Model III is an ordinal logistic regression. Predictors in italics have been adjusted by party. * p < .10, ** p < .05, *** p < .01.

The House and Senate models both exhibit comparable results. It should be noted that although not centrally important, the control variables behave as expected with most reaching statistical significance in at least one model. The key variables in Table 2, Ideological extremity and District ideology, are interacted to examine the effect of greater ideological extremity as districts become more ideological. As an interaction, this and all substantive analyses will be interpreted considering Brambor, Clark, and Golder (2006). While the interaction reaches statistical significance in all three models, simulations have been run and the changes in probabilities have been graphed to determine the certainty and magnitude of the relationship. In Fig. 1, the effect of an incumbent moving from one standard deviation below the mean on ideology to one standard deviation above is shown at three categories of district ideology measured using presidential vote at 45%, 55%, and 65%.13 These values reflect three distinct district types: the cross pressured incumbent district, the moderate district, and the ideologically extreme district. The effect of the ideological change is surrounded by a 95% confidence interval to visualize the certainty of the estimate. Fig. 1’s left-hand panel illustrates the House results from Model I, Table 2. In this figure, the results clearly show that incumbent ideology affects the likelihood of challenge in two of the three district types. In the cross pressured district, where the district is oriented ideologically counter to the incumbent, greater Ideological extremity increases

Representatives and as the central research goal here was to investigate the effect of incumbent congressional ideology on entry, it was necessary to find a measure common to both houses of Congress. 13 All figures use the CLARIFY package for Stata from King et al. (2000), Tomz et al. (2003).

the likelihood of facing a quality challenger roughly 32%, rising from a probability of 19% to 25%. This effect is statistically significant. An incumbent in this position should hesitate before engaging in more ideological behavior as doing so would increase the likelihood of facing a quality challenger, exactly the sort incumbents would most like to avoid. Looking at the moderate district, there is no distinct effect for ideology in this district type. House incumbents in these districts neither increase nor decrease their likelihood of facing a quality challenger through greater ideological extremity. In the ideologically extreme district, greater extremity benefits incumbents as their predicted likelihood of facing a quality challenger drops roughly 29%, falling from a probability of 14% to 10%. For these incumbents, the likelihood of facing a quality challenger is already low, but becoming more conservative or liberal will further improve their electoral position. The right panel illustrates the key variables from Model II, and a similar pattern emerges, albeit with a much lower level of certainty. Senators in the cross pressured districts show a small increase in the likelihood of facing quality opponents as they become more extreme, but this effect does not reach statistical significance. Senators in moderate districts show a roughly 16% decrease or a 10 percentage point drop in the likelihood of facing a quality challenger as they become more extreme; this effect narrowly misses significance. In ideologically extreme districts, the effect is statistically significant with senators from the most ideological districts having a strong incentive for more ideological behavior. Fig. 1 reveals noticeable distinctions between cross pressured districts where incumbents have incentives in terms of opponent quality to moderate, and the ideologically extreme districts where incumbents can lower their chances of facing a quality opponent through greater extremity.

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

Senate

.2 0 Decreases from 61% to 51%

-.4

-.2

Decreases from 14% to 10%

Increases from 60% to 63%

-.2

Decreases from 16% to 15%

Change in probability of facing a quality challenger

.2 0

Increases from 19% to 25%

-.4

Change in probability of facing a quality challenger

.4

.4

House

187

Cross pressured (45%)

Moderate (55%)

Ideologically extreme (65%)

District ideology

Cross pressured (45%)

Decreases from 62% to 39%

Moderate (55%)

Ideologically extreme (65%)

District ideology

Fig. 1. The change in probability of an incumbent facing a quality challenger. As incumbents become more ideologically extreme in different district contexts. Note: Congressional elections 1954 to 2008 where an incumbent sought re-election. Changes in simulated probabilities based on moving an incumbent from one standard deviation below the mean on ideology to one standard deviation above. All other values held to their means. Each change is surrounded by a 95% confidence interval.

The results discussed thus far have dealt with whether the incumbent faces an opponent previously elected to another political office. In Model III, Table 2, the analysis uses a more nuanced measure of quality and looks at whether an incumbent senator faces an opponent of increasing quality as defined by Lublin. In other words, while in Fig. 1 all quality opponents have met a basic standard for quality, in Fig. 2 each step represents candidates with appreciably more powerful and visible political experience. In that analysis, the data behaves in the manner expected by theory. In cross pressured districts, incumbents going from the 10th percentile on ideology to the 90th percentile increase their likelihood of facing a governor, statewide official, U.S. Representative, or a former

Moderate (55%) Cross pressured (45%)

.1

Pr(Challenger of each type) .2 .3 .4 .5

.6

Ideologically extreme (65%)

10th Percentile

50th 90th Percentile Percentile

10th Percentile

50th 90th Percentile Percentile

10th Percentile

50th 90th Percentile Percentile

Senate incumbent ideological position No challenger or challenger without electoral experience Governor, statewide official, U.S. Representative or former senator

Fig. 2. The probability of facing a challenger of selected types. As incumbents become more ideologically extreme in different district contexts. Note: Senate elections 1954 to 2008 where an incumbent sought reelection. All other values held to their means.

senator. While the increase is modest, the trend is positive. Additionally, increasing ideological extremity reduces the likelihood of facing a politically inexperienced challenger. For senators from moderate and ideologically extreme districts, the effect completely reverses. For those officials, greater extremity decreases the likelihood of facing the most experienced challengers and increases the likelihood of facing a challenger without political experience. Fig. 2 offers further evidence for the conditional incentives that incumbents face for greater ideological extremity based on their districts’ ideology. The analysis thus far has looked at both parties in Congress together. In Table 3, the analysis investigates whether the effects remain similar across parties, a significant question as previous scholars have noted that the entry decision of quality opponents can vary by party (Jacobson, 2009). In Fig. 3, we observe the effects from Table 3 with the House models in the top panels and the Senate models in the bottom panels. In the top panels, we observe differences between the two parties. For Democrats in cross pressured districts, greater extremity increases the likelihood of facing a quality opponent. For Republicans in similar districts, greater extremity neither increases nor decreases the likelihood of facing such an opponent. In moderate districts, Democrats have no clear result from greater extremity while their Republican counterparts receive a modest but statistically significant advantage. In the most ideological districts, Republicans receive a reduction in the likelihood of facing a quality opponent from greater extremity while their Democratic counterparts receive no similar benefit. In short, from the House models, it appears the Democrats receive a greater disadvantage from ideological extremity than Republicans,

188

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

Table 3 The effect of ideological extremity on the probability of facing a quality opponent 1954–2008 (DW-NOMINATE). I House Dems. only Key variables Ideological extremity District ideology Ideological extremity* District ideology Control variables Freshman Female Income change Prev. incumbent percentage Incumbent majority party Pres. approval Midterm Constant N Pseudo R2

II House Reps. only

4.80 (1.55)*** −0.04 (0.01)*** −0.08 (0.03)**

0.51 (2.33) −0.02 (0.02) −0.03 (0.04)

0.56 (0.12)*** 0.37 (0.18)** −0.09 (0.06) −0.05 (0.01)*** −0.53 (0.13)*** 0.00 (0.00) 0.27 (0.08)*** 3.28 (0.68)*** 3476 0.17

III Senate Reps. only

III Senate Dems. only 4.44 (5.34) −0.01 (0.04) −0.07 (0.11)

5.27 (4.57) −0.01 (0.03) −0.12 (0.08)

−0.28 (0.25) 1.07 (0.64)* 0.13 (0.12) −0.05 (0.01)*** −0.02 (0.26) −0.01 (0.01) −0.23 (0.19) 3.77 (2.12)* 360 0.09

0.20 (0.12)* −0.22 (0.20) −0.10 (0.05)* −0.06 (0.01)*** −0.21 (0.12)* −0.01 (0.00)** 0.36 (0.07)*** 4.50 (0.94)*** 2800 0.09

−0.12 (0.28) −0.49 (0.59) 0.37 (0.15)** −0.08 (0.02)*** −0.23 (0.27) −0.01 (0.01) 0.17 (0.21) 6.97 (1.84)*** 325 0.14

Note: The universe of cases is all congressional elections from 1954 to 2008 outside of Louisiana where an incumbent sought re-election, the seat has not been redistricted since the last election, and ideological scores are available for the incumbent. The dependent variable in all models is whether the incumbent faces a quality challenger (0–1). Accordingly, the models are logistic regressions. Predictors in italics have been adjusted by party. * p < .10, ** p < .05, *** p < .01.

at least in cross pressured districts. Conversely, Republicans other than those from cross pressured districts receive more obvious benefits from ideological extremity. Due to the smaller number of cases in the Senate, it is difficult to observe clear effects. For Senate Democrats, the predicted changes are similar to the House Democrats, positive for cross pressured and moderate districts, essentially zero for strongly ideological districts, but the large confidence intervals all cross the zero line. While some evidence that Senate Democrats receive similar results from greater ideological extremity exists, the effects are not clear. For

Senate Republicans, the pattern resembles that of their House colleagues. In the bottom right panel, we observe no distinct effects in cross pressured districts but in both the moderate and ideologically extreme districts, greater extremity leads to a statistically significant reduction in the probability of facing a quality opponent. The results in Table 3 and Fig. 3 point toward interesting differences between the two parties. For the Democrats, greater ideological extremity in both moderate and ideologically extreme districts offers fewer benefits than it does for Republicans. This difference may stem, in part, from the lower initial probabilities of facing quality challengers. A

.2 .4

Increases from 14% to 22% Increases from 10% to 12%

Cross pressured (45%)

Moderate (55%)

Decreases from 32% to 26% No Change

Ideologically extreme (65%)

Decreases from 27% to 20% Decreases from 22% to 15%

Cross pressured (45%)

Increases from 48% to 62%

Increases from 44% to 51%

Cross pressured (45%)

No Change Moderate (55%)

District ideology

Moderate (55%)

Ideologically extreme (65%)

Senate Republicans

Ideologically extreme (65%)

-.4 -.2 0 .2 .4

Senate Democrats -.4 -.2 0 .2 .4

Change in probability of facing a quality challenger

House Republicans

-.4 -.2 0

.2 .4 -.4 -.2 0

Change in probability of facing a quality challenger

House Democrats

Decreases from 75% to 73%

Decreases from 71% to 57%

Cross pressured (45%)

Decreases from 66% to 39% Moderate (55%)

Ideologically extreme (65%)

District ideology

Fig. 3. The change in probability of an incumbent facing a quality challenger. As incumbents become more ideologically extreme in different district contexts. Note: Congressional elections 1954 to 2008 where an incumbent sought re-election. Changes in simulated probabilities based on moving an incumbent from one standard deviation below the mean on ideology to one standard deviation above. All other values held to their means. Each change is surrounded by a 95% confidence interval.

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

House Democrat with a moderate voting record in a somewhat conservative district faces a probability of a quality opponent of only 14% while a comparably placed Republican faces a probability of 32%. Or to put it another way, such a Democrat would face a quality challenger roughly once every eight election cycles, while a Republican would face a quality opponent once every three cycles. These differences exist in the Senate as well, although they are somewhat more muted. The difference between the parties in the likelihood of facing a quality challenger persists across all district orientations with Republicans systematically more likely to face quality opponents than their Democratic counterparts regardless of incumbent ideology. Numerous plausible explanations exist for this systematic variation, such as differences in party recruitment strategies, the relative size of the candidate pools, or candidate variation in willingness to take political risks, but any explanation requires additional empirical investigation. The relevance for this analysis is that Republicans appear to receive greater gains for ideological activities in Congress, in terms of discouraging quality opponents, in moderate and conservative districts than Democrats do in moderate and liberal districts. Democrats, on the other hand, face much lower initial probabilities of facing quality opponents, but those members from cross pressured or somewhat conservative districts face strong incentives to moderate. These differences may play a role in the greater ideological homogeneity among Republicans than Democrats in Congress. 6. Conclusion The results point to a common finding: looking solely at opponent quality, incumbents have different electoral incentives in terms of ideological positioning based on the ideology of the districts they represent. Generally, incumbents from districts where the district median is counter to the incumbent’s party will have incentives to moderate or risk facing more politically experienced opponents. Members from more ideological districts will have incentives toward greater extremity. Representatives from moderate districts, however, will not receive strong benefits or penalties from more ideological behavior. It should be noted that these results come from general elections rather than primary elections. Depending on the specific district, some members may focus more on the primary election than the general election. Brady, Han, and Pope (2007) find that House members have electoral incentives toward extremity in terms of primary elections. The results here complement those findings and extend them to general elections, at least for the majority of members. Specifically, as most members do not represent cross pressured districts, the majority of elected representatives in Congress have electoral incentives to behave ideologically. The results presented here complement Brady et al. well, showing that ideological extremity is doubly advantaged in the most partisan and ideological districts as incumbents have incentives toward extremity at both the primary and the general election stage. In moderate districts, there is no overall penalty for extremity in terms of opponent quality in general elections. Thus members from such districts may choose to focus their attention on primary elections

189

and behave more ideologically in Congress to reduce their likelihood of facing a primary opponent. The advantage for extremity disappears when the incumbent represents a cross pressured district. For these representatives, their electoral position will suffer if the incumbent moves away from the center. While the number of members from cross pressured districts varies from Congress to Congress, using a cutoff of 45% and below, in no decade does the total number of members exceed 40% in the Senate or 26% in the House.14 These cross pressured members have very different incentives than their counterparts from more ideological districts. The conditional relationship found in this analysis adds a new dimension to the literature on candidate emergence, illustrating that incumbents can behave relatively extremely and avoid facing strong opponents in most cases, provided that they remain on the ‘correct’ side of the district. The broader point does not tell the entire story however. Instead of the electoral incentives being consistent across the two parties, it appears that they differ somewhat between Republicans and Democrats. Increasing conservatism among Republicans leads, in the majority of districts, to a reduction in the likelihood of facing a quality challenger. Increasing liberalism among Democrats leads to an increased likelihood of challenges in somewhat conservative districts without clear electoral advantages in more Democratic districts. This differing effect of incumbent ideology on candidate entry may stem from the fact that Republican incumbents are initially more likely to face quality opponents than their Democratic counterparts. Although, outside the scope of this paper, further studies should investigate the link between party differences and candidate emergence. Such an investigation would also create opportunities for a broader examination of differences between the two parties in terms of electoral incentives. As ideological cohesion has increased for both parties in Congress, but not uniformly, one potential explanation for the difference between the parties comes from how an elected official’s behavior affects the quality of their opposition. Specifically, the incentives for greater conservatism for most Republicans may affect the increasing ideological homogeneity of the Republicans in Congress. While additional investigation may require alternative candidate quality measures, future analysis could compare the results presented here to state legislatures. The diversity of these legislatures would allow scholars more opportunities to refine the linkage between incumbent ideology and opposing candidate quality.

14 Looking at the cases by decade, the highpoint for districts below 45% comes in the 1980s. In that decade, roughly 26% of Representatives and 39% of Senators represented districts below that total. The highpoint for districts above 65% came in the 1980s for the Senate (roughly 11%) and the 2000s for the House (roughly 22%). In all decades, in both houses, there have been more districts above 45% than below. In the House, the majority of the districts exceed 55% in all decades, while in the Senate the majority of the districts range between 45% and 55%. All of these values come from the decade measure; using a single presidential vote increases the amount of variation as landslide elections like 1964 and 1972 can substantially affect the results. The author will provide those distributional results upon request.

190

N.L. Pyeatt / The Social Science Journal 51 (2014) 181–190

In total, this analysis contributes to the literature as it illustrates that the majority of the House and Senate have incentives to avoid moderation based on the factors motivating high quality challengers. Additionally, this work has the potential to add some insight to the study of congressional polarization. If there is no penalty and some reward in terms of opponent quality for being more ideologically extreme, then one potential explanation for polarization stems from changes in the number of moderate and ideologically extreme districts (Theriault, 2008). While the work of McCarty, Poole, and Rosenthal (2009) concludes that polarization does not come from redistricting, their work does not investigate the effect of candidate emergence. If quality candidates react differently to more extreme incumbents in cross pressured districts than moderate and ideologically extreme districts and the number of cross pressured districts declines, then few incumbents would have incentives for moderation. This would be independent of the effect of voter constraints on incumbents. Further work should examine this relationship while also considering the impact of the conditional role of ideology on party loyalty within Congress, both on the floor and within committees. Expanding this work could also examine the relationship of ideology with candidate quality on members of historically underrepresented groups and in other legislatures. References Abramowitz, A. I. (1988). Explaining Senate election outcomes. American Political Science Review, 82, 385–403. Adams, G., & Squire, P. (1997). Incumbent vulnerability and challenger emergence in Senate elections. Political Behavior, 19, 97–111. Ansolabehere, S., Snyder, J. M., Jr., & Stewart, C., III. (2001). Candidate positioning in U.S. House elections. American Journal of Political Science, 45, 136–159. Bianco, W. T. (1984). Strategic decisions on candidacy in U.S. congressional districts. Legislative Studies Quarterly, 9, 351–364. Black, G. (1972). A theory of progressive ambition: Career choices and the role of structural incentives. American Political Science Review, 66, 144–159. Bond, J., Covington, C., & Fleisher, R. (1985). Explaining challenger quality in congressional elections. The Journal of Politics, 47, 510–529. Born, R. (1986). Strategic politicians and unresponsive voters. The American Political Science Review, 80, 599–612. Brady, D. W., Han, H., & Pope, J. C. (2007). Primary elections and candidate ideology: Out of step with the primary electorate? Legislative Studies Quarterly, 32, 79–105. Brambor, T., Clark, W. R., & Golder, M. (2006). Interaction models: Improving empirical analyses. Political Analysis, 14, 63–82. Buttice, M. K., & Stone, W. J. (2012). Candidates matter: Policy and quality differences in congressional elections. The Journal of Politics, 74, 870–887. Canes-Wrone, B., Brady, D. W., & Cogan, J. F. (2002). Out of step, out of office: Electoral accountability and House members’ voting. The American Political Science Review, 96, 127–140. Davis, O., Hinich, M., & Ordeshook, P. (1970). An expository development of mathematical model of the electoral process. The American Political Science Review, 64, 426–448. Downs, A. (1957). An economic theory of democracy. New York: Harper and Row.

Erikson, R. S., & Wright, G. C., Jr. (1980). Policy representation of constituency interests. Political Behavior, 2, 91–106. Erikson, R. S., Wright, G. C., & McIver, J. P. (1993). Statehouse democracy: Public opinion and policy in the American states. Cambridge: Cambridge University Press. Goodliffe, J. (2001). The effect of war chests on challenger entry in U.S. House elections. American Journal of Political Science, 45, 830–844. Goodliffe, J. (2004). War chests as precautionary savings. Political Behavior, 26, 289–315. Goodliffe, J. (2007). War chests and challenger quality in Senate elections. Legislative Studies Quarterly, 32, 135–156. Grofman, B. (1985). The neglected role of the status quo in models of issue voting. The Journal of Politics, 47, 230–237. Jacobson, G. (1989). Strategic politicians and the dynamics of U.S. House elections, 1946–86. The American Political Science Review, 83, 773–793. Jacobson, G. (2009). The politics of congressional elections (7th ed.). New York: Pearson Longman. Jacobson, G., & Kernell, S. (1983). Strategy and choice in congressional elections. New Haven: Yale University Press. Kazee, T. (1983). The deterrent effect of incumbency on recruiting challengers in U.S. House elections. Legislative Studies Quarterly, 8, 469–480. Kedar, O. (2005). When moderate voters prefer extreme parties: Policy balancing in parliamentary elections. The American Political Science Review, 99, 185–199. King, G., Tomz, M., & Wittenberg, J. (2000). Making the most of statistical analyses: Improving interpretation and presentation. American Journal of Political Science, 44, 347–361. Krasno, J., & Green, D. P. (1988). Quality challengers in House elections. The Journal of Politics, 50, 920–936. Lawless, J. L., & Pearson, K. (2008). The primary reason for women’s underrepresentation? Reevaluating the conventional wisdom. The Journal of Politics, 70, 67–82. Lublin, D. I. (1994). Quality, not quantity: Strategic politicians in U.S. Senate elections. The Journal of Politics, 56, 228–241. Maestas, C. D., Fulton, S., Maisel, L. S., & Stone, W. J. (2006). When to risk it? Institutions, ambitions, and the decision to run for the U. S. House. The American Political Science Review, 100, 195–208. Maisel, L. S., & Stone, W. J. (1997). Determinants of candidate emergence in U.S. House elections: An exploratory study. Legislative Studies Quarterly, 22, 79–96. McCarty, N., Poole, K. T., & Rosenthal, H. (2009). Does gerrymandering cause polarization? American Journal of Political Science, 53, 666–680. Palmer, B., & Simon, D. (2008). Breaking the glass ceiling: Women and congressional elections (2nd ed.). New Haven: Routledge. Palmer, B., & Simon, D. M. (2005). When women run against women: The hidden influence of female incumbents in elections to the U.S. House of representatives, 1956–2002. Politics & Gender, 1, 39–63. Rabinowitz, G., & Macdonald, S. E. (1989). A directional theory of issue voting. The American Political Science Review, 83, 93–121. Squire, P. (1989). Competition and uncontested seats in U.S. House elections. Legislative Studies Quarterly, 14, 281–295. Squire, P. (1992). Challenger quality and voting behavior in U.S. Senate elections. Legislative Studies Quarterly, 17, 247–263. Stewart, C., III. (1989). A sequential model of U.S. Senate elections. Legislative Studies Quarterly, 14, 567–601. Stone, W. J., Fulton, S. A., Maestas, C. D., & Maisel, L. S. (2010). Incumbency reconsidered: Prospects, strategic retirement, and incumbent quality in U.S. House elections. The Journal of Politics, 72, 178–190. Tausanovitch, C., & Warshaw, C. (2013). Measuring constituent policy preferences in Congress, state legislatures, and cities. The Journal of Politics, 75, 330–342. Theriault, S. (2008). Party polarization in Congress. New York: Columbia University Press. Tomz, M., Wittenberg, J., & King, G. (2003). Clarify: Software for interpreting and presenting statistical results. (2.1 ed.). Stanford University, University of Wisconsin, and Harvard University.