Institutions and Representational Roles in American State Legislatures

What is the impact of a legislature’s institutions on the representational roles its members adopt? We address this question by examining the role orientation of state legislators in eight states, explaining why some legislators identify more with a trustee model of representation and others identify more with the delegate model. Using ordinal logistic regression analysis on data from a survey of 447 legislators, we test for the effects of multimember districts and term limits, along with several other factors. First, we find that representational roles and behavior are related; legislators who think of themselves as delegates are much more likely to hold frequent district office hours than their counterparts who think of themselves as trustees. Second, we find that, overall, legislators are more likely to consider themselves trustees than delegates. Third, we find that multimember districts and term limits increase the likelihood that legislators think of themselves as trustees. Thus, legislative institutions can influence the representational roles legislators adopt. a basic tenet of political science is that institutions can affect behavior (Shepsle and Weingast 1995). For example, when rules of a legislative body change, we expect that the policies produced by that chamber will change as a result (Larimer 2005). Likewise, as the rules of elections change, so do the outcomes of those elections (Cox 1990). Although institutions are generally believed to affect representation (Moncrief, Thompson, and Cassie 1996), most research on this relationship has focused on one case—the United States Congress—whose institutions have largely remained constant.1 In keeping with Jewell’s (1983, 310) call to uncover the “variables that may help to explain particular role orientations,” we use the institutional variation in the American state legislatures to test whether legislative institutions can influence legislators’ representational roles. We also demonstrate the influence of a legislator’s role orientation on his or her legislative behavior and describe 03.174-194_SPPQ.06.2.indd 174 4/6/06 1:53:33 PM summer 2006 / state politics and policy quarterly 175 the distribution of representational roles among American state legislators. In doing so, we gain a better sense of how representation is practiced and how institutions shape outcomes in American state legislatures. representational roles Scholars and practitioners of American politics have long debated the proper form of legislative representation. During deliberations over the ratification of the United States Constitution, anti-Federalists and Federalists sparred over whether legislators should make decisions they believe are in the “best interest of the state” or decisions that “follow the will of the governed” (Carman 2003, 2). More recently, scholars have placed these trade-offs on a scale characterizing role orientations from “delegate” to “trustee” (Pitkin 1967), with “politicos” residing somewhere in the middle (Wahlke et al. 1962). Delegates believe they are in office to follow the unfiltered opinion of the people. A pure delegate does not express his or her personal opinion on an issue, but rather votes based on the opinion of the constituents. On the other hand, a pure trustee believes that he or she is in office to act by making the best decision possible on some objective criteria, regardless of the constituents’ opinions. When district opinion and legislator opinion about the best course of policy action come into conflict, a trustee believes that the opinion of the people is less important than his or her own opinion. Many trace the evolution of the trustee model to Edmund Burke, who argued it was advisable for representatives to “promote the interests of constituents without consulting their wishes” (Rosenthal 1998, 8). Which of these representational roles is most commonly adopted by American legislators? Most scholars have found that state and national legislators in this country are more likely to characterize themselves as trustees than delegates (Cavanagh 1982; Gross 1978; Hanson 1989; Rosenthal 1998; Wahlke et al. 1962). Of course, while it is instructive to think in terms of the ideal types of delegate or trustee, most legislators fall somewhere in the middle. Many legislators even display different roles on different issues. Nonetheless, legislators’ general tendencies can teach us a great deal about how they view representation and the proper relationship between citizens and their elected officials. While scholars have debated the usefulness of these concepts for years (Cavanagh 1982; Gross 1978), most have found that representational roles vary in systematic and predictable ways (Carman 2003) and that they have important implications for understanding representation. These role orientations are particularly useful in explaining legislative behavior on salient 03.174-194_SPPQ.06.2.indd 175 4/6/06 1:53:33 PM 176 cooper and richardson issues where constituents may have well-established opinions (Kuklinski and Elling 1977; McCrone and Kuklinski 1979). For instance, a legislator’s rhetoric about representational roles may affect how he or she is viewed by the public (Lipinski 2003). Even in political systems with strong party discipline, representational roles vary and can affect behavior (Searing 1991, 1994). Studlar and McAllister (1996) showed that these roles affect both constituency service and vote margins in Australian legislatures, and Searing (1991) suggests these roles do a better job of explaining time spent in the district for British Members of Parliament than tenure, electoral marginality, party, or distance from home. Clearly, these roles affect not only a legislator’s own ideas about representation, but they also affect his or her legislative behavior. explaining representational roles Many factors may influence the representational role a legislator adopts. Because the impact of institutional characteristics is the primary focus of this article, we address them first, concentrating on two major variables: multimember districts (MMDs) and term limits. Single-member districts (SMDs), in which only one legislator is elected to represent a district, are by far the most common electoral structure in the United States today, but several American state legislatures and many legislatures worldwide use at least some multimember districts (MMDs), in which more than one legislator is chosen on the same ballot to serve a single district. Because most research on legislative representation has focused on SMD systems, the impact of this aspect of district structure on representation has rarely been examined (Bowler and Farrell 1993). Extant research is dated and anecdotal, but it generally suggests legislators who serve in MMDs are less known to their constituents and, therefore, are considered less accountable. For example, Jewell (1982, 119) found that “legislators were more likely to be trustees in states using multimember districts, somewhat more likely to be trustees where there had been a recent shift from multimember to singlemember districts, and slightly more likely to be delegates where single-member districts had long been used.” Legislators in SMDs are better known to their constituents, and they must listen to the entire district, rather than just a small subconstituency, if they wish to be re-elected. As a result, SMD legislators are “more vulnerable to pressure from groups that are concentrated in the district” (Jewell 1982, 119), and they must hew closer to the demands of their constituents. In other words, “single-member districts . . . tend to bring members and constituents closer together” (Rosenthal 1998, 29). 03.174-194_SPPQ.06.2.indd 176 4/6/06 1:53:33 PM summer 2006 / state politics and policy quarterly 177 These differences in district structure may produce legislators with different role orientations because they create different incentives for candidates, with MMDs shifting the incentive away from the focus on the median voter that SMDs encourage (Cox 1990). For example, in a race involving four candidates running for two seats on the same ballot, one who receives less than 26 percent of the vote from any part of the constituency may win a seat. Furthermore, such a race may pit two fellow party members against one another as well as against members of the opposite party. In general, an MMD candidate does not have to seek the center of the district’s ideological distribution to win (Adams 1996; Cox 1990; Richardson, Russell, and Cooper 2004). Therefore, legislators who wish to seek re-election need only concentrate on their subconstituency of active supporters and can ignore the median voter. This situation leads us to our first major hypothesis: H1: Legislators who serve in multimember districts are more likely to claim to be trustees. The most recent wave of institutional reform in state legislatures has been the inclusion of term limits. In an effort to reduce careerism in state legislatures, in the 1990s, 21 states passed laws limiting the number of terms lawmakers can serve. While six states have since removed these limits, 29 chambers in 15 states still limit legislative service to between six and 12 years. By July 2005, 1,218 legislators had been termed out, but five chambers had not yet reached the date when their limits would start to take effect (National Conference of State Legislators 2005). Both reformers and scholars expected term limits to alter the relationship between legislators and their constituents. Our argument emphasizes the idea that ambitious legislators are driven by the re-election motive (Mayhew 1974), and the removal of long-term electoral pressures reduces their incentive to focus on constituent preferences (Glazer and Wattenburg 1996). Because term-limited legislators are not as motivated by re-election, they may spend more time engaged in lawmaking and less time on constituency casework, pork barrel projects, and other activities related to their districts (Carey, Niemi, and Powell 2000). Will (1992) argues that term limits will lead to the election of citizen legislators who will be more likely to follow their own consciences rather than hew strictly to constituency preferences. Therefore, a term-limited legislator may be more likely to act as a trustee. Other scholars have argued against such a “Burkean shift” (Corey et al. 2003) toward a trustee role. For example, Petracca (1991) argues that legislative professionalism distances lawmakers from their constituents and that term limits would be more likely to produce legislators who are closer to 03.174-194_SPPQ.06.2.indd 177 4/6/06 1:53:34 PM 178 cooper and richardson their constituents. Taking a different approach, Wright (2004, 19) uses nominate legislative voting scores for all state legislative chambers in 1999–2000, finding “absolutely no evidence that constituency preferences matter less in chambers where term limits have been implemented.” To test the effect of term limits, one could use a dummy variable to indicate a legislator in a state with term limits. But during our study period, many term-limit states have legislatures that are composed of members who were elected before and after term limits became law. The single dummy variable approach assumes that all legislators in such a body would change their behavior and attitudes in response to the removal of the electoral incentive when term limits are passed. This assumption may be unwarranted. In the four term-limit states in our study, many veteran legislators may not have considered term limits when they were first elected, but all first-term legislators were recruited after term limits had removed some legislators in the state. Therefore, we hypothesize two different effects of term limits on representational roles. We assess separately whether term limits affect all sitting legislators or whether the main effect is through the recruitment of legislators with different attitudes than those who are attracted to legislatures without term limits. H2A: All legislators subject to term limits are more likely to claim to be trustees. H2B: Legislators elected for the first time under term limits are more likely to claim to be trustees. In addition to the impacts of these legislative institutions, a number of other factors may influence which representational roles legislators adopt. To make unbiased tests of our hypotheses of institutional effects, we must control for these forces in our models. First, consider that legislators are elected from very different types of constituencies even within the same state. We expect that legislators who represent heterogeneous districts are more likely to consider themselves as trustees since they cannot easily identify a single district opinion (Fiorina 1974). Districts can be heterogeneous on a variety of dimensions, but an important one in American politics is ethnic diversity. To create ethnic diversity in state legislative districts, we collected data directly from each state’s Secretary of State. When such data were not available from a Secretary of State, we used geographic information systems software to determine the percentage of each census block group in a state in our sample (2000 summary tape 3 or SF3) that fell within each state legislative district and divided the demographic data accordingly (United States Bureau of the Census 2005). We 03.174-194_SPPQ.06.2.indd 178 4/6/06 1:53:34 PM summer 2006 / state politics and policy quarterly 179 then totaled the demographic data for each legislative district based on the block group fragments contained within the district, resulting in an estimate of the demographic profile of each legislative district. We then used Hero and Tolbert’s (1996) formula to calculate a racial heterogeneity score for each district.2 Theoretically, this variable could vary from 0 to 1.0, with a district composed entirely of one ethnic category receiving the low score and one with several equally represented ethnic groups receiving the high score. In our dataset, our lowest scoring legislative district (.007) was in South Dakota with 99.7 percent non-Hispanic whites, and the highest scoring district (.714) was in New Jersey with 22.0 percent Latinos, 37.0 percent non-Hispanic whites, 29.0 percent blacks, and 12.0 percent Asian Americans. Another characteristic of a legislative district that could affect a legislator’s representational role is distance from the state capital. A legislator who lives far from the capital may find it more difficult to travel to the district and spend time with his or her constituents, suggesting that he or she would adopt more of a trustee role because of the difficulty of determining the district’s wishes. A legislator’s personal characteristics may also be important in shaping his or her representational style. Because African-American legislators often represent more homogeneous majority-minority districts, black state legislators may act more like delegates than their white colleagues (Swain 1993; Whitby 1997; Burnside and Haysley-Jordan 2003). On the other hand, there is little evidence supporting this effect in other minority legislators. While the states included in our study were served by black, Latino and Latina, NativeAmerican, and Asian-American legislators, in addition to white legislators, there were too few survey respondents in some ethnic categories to create separate variables for each. Therefore, we include a single ethnic minority variable in our models to control for any effect, based on the legislator’s self-identified minority status. Freshman legislators may enter the legislature with different ideas about representation than will develop later in their careers. Specifically, because of electoral uncertainty and traditional homestyle patterns, we would expect that first-time legislators are more likely to consider themselves as delegates. Lipinski (2003) finds evidence supporting this hypothesis in the congressional context, but no work explicitly examines this relationship in the state legislature. There is also reason to believe that the manner in which female legislators view the job differs from that of their male colleagues. Differences in sex-role socialization may lead women to spend more time in their districts and conduct more constituency service than men (Carey, Niemi, and Powell 03.174-194_SPPQ.06.2.indd 179 4/6/06 1:53:34 PM 180 cooper and richardson 1998; Richardson and Freeman 1995). As a result, we expect that female legislators are more likely to perceive themselves as delegates than their male counterparts. Finally, the strength of a legislator’s ideology may affect his or her role orientation. Legislators who consider themselves to be extreme liberals or extreme conservatives are likely to hold strong worldviews on a range of policies and, therefore, may feel less compelled to follow the guidance of others in their legislative activities, including their constituents (Richardson, Russell, and Cooper 2004). As a result, these ideological extremists are more likely to view themselves as trustees than their more moderate colleagues. Thus, theory and previous research suggests a number of potential predictors of representational roles. Specifically, we expect that, all else being equal, multimember districts and term limits encourage legislators to perceive themselves as trustees, and ideological extremists will also be more inclined to self-identify as trustees. On the other hand, we expect that legislators who are members of ethnic minorities, in their first term, female, or represent homogenous districts are more likely to view themselves as delegates.