The Dynamics of Gender, Ideology, and Policy in a Polarized Congress. Megan M. Moeller

Transcription

1 The Dynamics of Gender, Ideology, and Policy in a Polarized Congress Megan M. Moeller 17 March 2012

2 ABSTRACT This paper focuses on the dynamics of the relationship between gender, ideology, and policy in a polarized Congress. I determine the extent to which the ideology of female Republican members of the Congress distinct from that of male Republican members of Congress and in what policy areas is the voting of Republican women in the House changing the most, the least, or not changing at all. Using Policy Agendas Project coding, I examine DW-NOMINATE and W-NOMINATE scores by policy area from the 102 nd through 108 th Congress. Results indicate that the overall effect of gender on ideology diminishes steadily over time, and the voting behavior of Republican women in Congress is becoming more and more like that of Republican men. These trends are most pronounced in issue areas that are the most unidmensional and receive the most attention in Congress. Gender does not seem to have an effect on voting behavior for issues that receive little attention in Congress. Overall, Republican women not only are becoming more ideological over time, they are doing so at a faster rate than Republican Men. This suggests that Republican women are more responsive to party polarization than their male counterparts. Women s issues, however, are highly resistant to these trends. Gender continues to have a significant effect on Republican women s ideology and voting behavior in Education, Health, and Civil Rights votes. 2

3 INTRODUCTION Gender differences in political behavior have received much attention from scholars, and a variety of gender differences in political behavior have been demonstrated to persist. Women vote more often but are less politically knowledgeable and otherwise participatory than men (Burns, Scholzman, Verba 2001). Partisanship by gender has diverged over time, and in recent years women are more likely to identify with the Democratic party (Box-Steffensmeier, De Boef, Lin 2004). Gender differences not only exist in mass political behavior and participation but also in legislative behavior. Female legislators are more likely to take a consensus building approach to legislating (Volden, Wiseman, Wittmer Forthcoming), rely on a wider variety of resources when formulating policy solutions, suggesting they take a different approach to conceptualizing policy problems (Kathelene 2005), are generally more task-oriented (Rosenthal 2005), spend more time negotiating (Epstein, Niemi, Powell 2005), retire earlier (Theriault and Lawless 2005), place more emphasis on female issues in press releases, and are less likely to take credit and blame others in press releases as well (Fridkin and Woodall 2005). Mirroring the more liberal partisanship of women in mass publics, women in Congress are more likely to be Democrats. In policy preferences women of both parties typically have had ideal points to the left of mean ideal point of their respective party and are more likely to sponsor and cosponsor women s issue legislation (Swers 2002), Party polarization, however, has increased (Sinclair 2006), and in ideologically charged sessions all members of Congress tend to stay closer to party lines (Swers 2002). This increased ideological extremism is even true in cases when female legislators more liberal ideology should predict otherwise. For example, when the Republicans gained majority party status in 1994, Republican men s support for liberal positions on women s 3

4 issues dropped 26% while Republican women s support dropped 50%. This change was partially due to the influx of conservative freshman women, but veteran moderate Republican women also reduced their support for liberal positions on women s issues (Swers 2002). As party polarization increases the degree of Republican women s ideological voting has changed, though to what extent and in which issue areas remains to be seen. To date, there has yet to be a systematic analysis of the dynamics of women s ideology in various policy areas as party polarization increases. This study will address this gap in knowledge by exploring the relationship between gender, ideology, and policy in roll call votes in the Republican Party during an era of increased partisan polarization how has it changed over time? Specifically, I examine the direction and magnitude of these changes, to what extent it is distinct from that of male Republican House members, and in what policy areas is their voting changing the most, the least, or not changing at all. This paper proceeds as follows: First I note important theoretical considerations for this study and present relevant hypotheses. Then I discuss the data and methods I utilize and introduce a model to study the effect of gender on ideology. Next, I put forth the results of my analyses and provide interpretations of the findings. Last, I offer conclusions of this study and consider avenues of future research. THEORY and HYPOTHESES A variety of theoretical considerations may have implications for the dynamics of the relationship between gender, ideology, and policy in a polarized Congress. Firstly, a relationship exists between increased polarization and increased unidimensionality in policy issues (Poole and Rosenthal 2006; Jochim and Jones Forthcoming). Issue dimensionality refers to a spatial arrangement of preferences for a given issue. In a 4

5 unidimensional space, for example, preferences fall along a single continuum. Policy choices are structured by coalitions, but different issues should generate different coalitions in Congress (Riker 1990). This is because not all issues affect members of Congress in the same way. A set of votes concerning public transportation might affect urban constituencies differently than rural constituencies, thus generating legislative coalitions of urban representatives versus rural representatives. These coalitions need not fall along partisan cleavages. Democrats and Republicans from rural districts would likely vote together against Democrats and Republicans from urban districts in this issue area. Often, however, political parties do structure the coalitions for a given issue, and when this occurs the choice space is more likely to be unidimensional. In an era of heightened party polarization, an increased number of issues collapse onto a unidimensional choice space (Aldrich 1995; Jochim and Jones Forthcoming). Some issues continue to display multidimensional characteristics in terms of voting behavior, despite party polarization (Jochim and Jones Forthcoming). Jochim and Jones have calculated the mean extracted dimensions for eighteen policy areas from 1965 to 2004 and find that, on average, Science, Trade, and Agriculture demonstrate the highest dimensionality while Labor, Housing, and Economics demonstrated the most unidimensionality. The policy areas ordered by mean dimensions in Jochim and Jones study are presented below: [Insert Table 1] Since the primary dependent variable in this study is W-NOMINATE scores, which only account for one issue dimension, I hypothesize that patterns in roll call voting behavior 5

6 for multidimensional issues will not be captured in the dynamics of W-NOMINATE scores. A second relevant consideration is the scarcity of attention. Given that humans have limited time, resources, and cognitive capacities, but remain goal-oriented, they can be characterized as being boundedly rational in their decision-making. The Punctuated Equilibrium approach to policy change posits that because institutions are built by humans, institutional decisions can also be characterized by bounded rationality (Baumgartner and Jones 1993). A microfoundation of bounded rationality implies that attention incurs a cognitive cost and, thus, is scarce. Focusing events or changes in issue definition lead to elevated attention to that issue. Decision changes occur as a result of shifts in attention, not from changes in preferences. Consequentially, policy change takes place during periods of heightened attention to an issue (Baumgartner and Jones 1993). Because this paper is interested in changes in voting behavior, the level of attention paid to a given issue should prove germane to these dynamics. In this paper I conceptualize the degree of attention Congress pays to an issue as the number of roll call votes taken on that issue. 1 The frequency of votes by issue area for the 102 nd through 108 th Congress is presented below: [Insert Table 2] I hypothesize that changes in the voting behavior of women in Congress should be the most pronounced in the issues that receive the most attention. 1 In future versions of this paper, issue attention will be conceptualized in a variety of ways, including Gallup s Most Important Problem responses. Time constraints prevented a more thorough examination of this relationship in the current version. 6

7 A third theoretical consideration is that of Women s Issues. There is, by no means, a consensus about what constitutes a women s issue. Scholastic literature has defined women s issues in a variety of ways, among which are issues that are particularly salient to women or issues with which women have traditionally been concerned in their role as caregivers (Swers 2002), issues that disproportionately affect women (Carroll 1985) or were intended to do so (Wolbrecht 2000). They may be any policies that deal with the relationship between men and women (Sanbonmatsu 2002), policies that deal with individuals in terms of their gender (Sapiro 2003), or policies that are not explicitly about gender but are understood by individuals in gendered ways (Winter 2005). This latter conceptualization refers to those policies that perceive of women in their role as caregivers, such as policies pertaining to Health Care or to children s issues, like Education. Policies that may be particularly salient to women or might disproportionately affect women would be those dealing with sex discrimination, for example. In this paper, I do not enter the debate about what constitutes a women s issue, but instead pay special attention to all issues that might fall in this category according to a variety of understandings. My intent is not to define what a women s issue is, but to be aware of all issues that may prompt distinct behavior so as not to limit my analyses. Consistent with the concept of the scarcity of attention, it may be the case that issues that are particularly salient to women for one reason or another would receive more attention from women than from men. Varied degrees of attention to an issue should in turn yield distinct policy outcomes. Whether due to social identity, socialization, or some other factor, inherent in the notion of a women s issue is that women will interact with this issue differently than will men, possibly resulting in unique voting behavior around this issue. I hypothesize that female members of Congress will 7

8 display distinct voting behavior in policy areas that could be defined as women s issues. DATA and METHODOLOGY The primary aim of this study is to determine how female Republican House members ideological voting has changed over time. Specifically, I examine the direction and magnitude of these changes, to what extent it is distinct from that of male Republican House members, and in what policy areas is their voting changing the most, the least, or not changing at all. To answer these questions, I first develop a model using Poole and Rosenthal s DW-NOMINATE scores for legislators as the dependent variable, 2 and then apply this model to explain W-NOMINATE scores for nineteen broad policy areas. I analyze roll call votes for each of the policy areas separately and by Congress from the 102 nd Congress through the 108 th Congress. DW-NOMINATE scores are a summary measure of roll call votes that take into account all nonconsensual votes in one congress and can be compared directly to the scores of a different Congress, though not across chambers (Poole and Rosenthal 1997). The scores are values between -1 and 1, where -1 represents an extremely liberal voting record and 1 represents an extremely conservative one. 3 In this analysis DW-NOMINATE scores will be my measure of ideology. I have restricted my analysis to the DW-NOMINATE scores of Republican members of the House, and so the more ideological the member is the closer their DW-NOMINATE score will be to 1. The intent is to determine if gender is a significant predictor of ideology and the 2 Data obtained from Poole and Rosenthal at 3 For a more thorough explanation of DW-NOMINATE scores, refer to Poole and Rosenthal

9 DW-NOMINATE score, so I include women as an independent variable. To measure women, I use a dummy variable where 1 indicates that the member is a woman and 0 indicates that the member is a man. Other factors besides a member s gender, however, may affect a members ideological voting behavior. The chief goal of members is reelection (Mayhew 1974), and so the members constituency is one important element that shapes the members roll call votes. A member with a highly ideological district should vote ideologically more often than a member with a moderate district (Canes-Wrone, Brady, and Cogan 2002; Jacobson 1987; Theriault, Hickey, and Blass 2011). The popularity of the president in the members district can also affect members voting behavior (Edwards 1978). Members will not want to vote against the president s preferences if the president enjoys high levels of popularity in their district for fear of electoral retribution (Gronke, Koch and Wilson 2003). To control for constituency ideology as well as presidential popularity I include as an independent variable the percent of the members district that voted for the Republican candidate in the most recent presidential election. 4 The inclusion of this variable ensures that differences in roll call voting behavior are not simply a reflection of a pattern in which women are elected from districts that support her ideology. Members are nearly always running scared with regard to re-election (Jacobson 1987), though some members are more electorally vulnerable than others. In times of heightened political awareness, vulnerable members in particular are likely to vote their constituency s preferences (Theriault, Hickey, Blass 2011). To control for electoral vulnerability I include the members percentage of votes in their most recent district election as an independent variable. Another factor that might affect a member s roll call 4 I am indebted to Patrick Hickey for sharing this dataset. 9

10 voting behavior is the cohort with which the member came to Congress. Different cohorts may display distinctive voting behavior, for example the influx of the ideological Gingrich Senators in the 1990s (Theriault 2008), and increased party polarization over the recent decades is evidence of this effect. To take this effect into account, I include the number of years that a member has served in the House as an independent variable. 5 The model I utilize is presented below: DW-NOMINATE i = β 0 + β 1 gender i + β 2 GOPpercent i + β 3 YearsHouse i + β 4 WinPercent i Using Ordinary Least Squares estimation, I analyze data for the 102 nd through 108 th Congresses, examining each Congress individually, and compare results for each Congress over time. To evaluate the dynamic effect of gender on ideology by policy area, I utilize a similar method. The Policy Agendas Project s dataset on congressional roll call votes has coded each roll call vote by issue content. 6 The consistent coding system sorts votes into eighteen major policy areas and allows for reliable across time comparisons (Baumgartner, Jones, and Wilkerson 2002). Using this coding system and Poole and Rosenthal s roll call votes datasets 7, I compute W-NOMINATE scores for legislators in the statistical program R for each of the eighteen policy areas in each the 102 nd through 108 th Congress. Lopsided votes and legislators who voted fewer than twenty times were excluded. In cases where there were fewer than twenty votes on a given policy area in a 5 Data for Congressional election results and years served in the House graciously provided by Dr. Sean Theriault

11 given Congress, legislators who voted fewer than eight times were excluded. I apply the same model presented above to each policy area for each Congress, using the W- NOMINATE score as the dependent variable in place of the DW-NOMINATE score. Again, I employ ordinary least squares estimation to determine the effect of gender on ideology by policy area for each Congress and compare the results over time. RESULTS and DISCUSSION The results from the OLS estimation of the effect of gender, etc. on DW- NOMINATE scores are presented below: [Insert Table 3] Focusing first on the control variables, these results indicate that, unsurprisingly, constituency factors are extremely good predictors of members voting behavior. Also, the effect of the number of years a member has served in the House is negative which is evidence that the newer a member is to the House, the more ideological their vote. The effect of electoral vulnerability is not a consistent predictor of ideology and would probably be more accurately measured as the spatial distance between the ideological position of the member and that of the district s median voter. For the variable of interest, gender, the direction of all the coefficient estimates is negative. An inverse relationship, therefore, exists between gender and an ideological DW-NOMINATE score. In other words, the negative direction indicates that being a woman decreases the likelihood that a Republican will vote ideologically. Initially Republican women were more moderate than Republican men. A comparison of the coefficients over time, 11

12 however, yields interesting results. In the 102 nd Congress, Republican women are less ideological at a statistically significant level. The magnitude and significance of the effect of gender on ideology steadily decreases over time. By the 108 th Congress, gender no longer remains a significant predictor of ideology. As polarization increases, the ideology of Republican women becomes statistically indistinguishable from that of Republican men. Republican women are voting more ideologically with each subsequent Congress; all members of Congress, however, are voting more and more ideologically over time. This is the essence of party polarization. An analysis of the means of the DW- NOMINATE scores across Congresses illustrate that much more is changing. I break down the means of DW-NOMINATE scores for men and women and compare them to each other, along side the overall Republican mean DW-NOMINATE scores for the 102 nd through 110 th Congresses. These results are presented below: [Insert Table 4] [Insert Figure 1] Initially, the mean Republican women s score is notably smaller than the mean score of Republican men, but over time the average women s score becomes nearly indistinguishable from mens. Over the nine congresses, Republican men s mean score changed by.256 while women s changed by.341. This change reveals that Republican women s ideological voting is increasing at a greater rate than that of Republican men, 12

13 suggesting that Republican women are more responsive to party polarization than their male counterparts. Now, I focus on the effect of gender in various policy areas to determine in which policy areas gender has a significant effect and in which areas have changes in Republican women s ideological voting occurred. Some values are missing because there were not enough votes on an issue in a particular Congress to provide reliable results. The results of OLS estimation by policy area are presented below: [Insert Tables 5 through 22] An evaluation of the results reveals the emergence of several patterns. In some issue areas, such as Defense, the effect of gender on W-NOMINATE scores follows the same pattern as the overall effect of gender on DW-NOMINATE scores. That is, the effect is always negative in direction but the magnitude and significance of the effect decreases steadily over time. In other areas, such as Social Welfare, the effect of gender on W- NOMINATE scores is nearly always negative, but also is nearly always insignificant, though decreases in magnitude over time. Further, in other issue areas, such as Agriculture, the effect is nearly always negative, nearly always insignificant, but the magnitude of the effect follows no discernable trend over time. In yet another set of areas, such as Transportation, the effect of gender on W-NOMINATE scores is nearly always negative, follows the same pattern as the overall effect of gender on DW- NOMINATE scores in terms of steadily decreasing significance, but the magnitude of the effect follows no discernable trend over time. In a final set of issues, such as Health, the effect of gender is nearly always negative, but the magnitude of the effect displays no 13

14 temporal trend and the significance of the effect varies dramatically. I group the issue areas by these patterns of the effect of gender over time. The results are displayed below: [Insert Figure 2] The numbers in parentheses indicate the frequency of votes on that issue in the time frame. As the figure indicates, the issue areas in which the effect of gender on W- NOMINATE scores displays the same trend as the overall effect of gender on DW- NOMINATE scores are those that are the most unidimensional and received heightened attention in Congress. Those issue areas in which the effect of gender is consistently insignificant are those that received less attention. The issue areas in which the magnitude of the effect of gender on W-NOMINATE score did not follow any discernable trend are those that display higher dimensionality. There are three issues that fail to fit these patterns: Civil Rights, Health, and Education. These three issues are relatively unidimensional and received heightened attention from Congress, yet the effect of gender on W-NOMINATE scores for these issues varies dramatically in significance and the magnitude of the effect follows no temporal pattern. Of all the issue areas considered, these three can most easily be categorized as Women s Issues. Health and Education clearly evoke women s traditional role as caregivers, and any votes pertaining to gender discrimination are classified as Civil Rights issues. It is not surprising that patterns in voting behavior for issues areas displaying higher dimensionality are not adequately captured with NOMINATE scores. Because I use a measure of ideology that is only one-dimensional, it is expected that the model will lack some explanatory power multidimensional issues. The absence of observable trends 14

15 in the effect of gender on W-NOMINATE scores for issues with greater dimensionality confirms my first hypothesis. Regarding issue attention, we can conclude that gender has a negligible effect on issue that receive low levels of attention. Furthermore, consistent with my second hypothesis, the issues for which the effect of gender changed the most over time are those that received the most attention. The overall trends in the effect of gender on ideology are most pronounced in those issues that are the most unidimensional and receive the most attention, save for women s issues. Consistent with my third hypothesis, women s issues seem to be most resistant to the trend of the decreasing effect of gender. For these issues, gender continues to have a significant effect on Republican women s ideology and voting behavior. CONCLUSIONS In this study, I have found that the effect of gender on ideology diminishes in magnitude and significance over time. That is, Republican women s voting behavior is becoming more and more like that of Republican men. Republican women are no longer the most liberal members of their party. They are becoming more ideological over time, and at a faster rate than Republican men. This suggests that Republican women are more responsive to partisan polarization than their male counterparts. These overall trends are most pronounced in issue areas that are the most unidimensional and receive the most attention. Gender seems to have no effect on voting behavior in issues that receive low attention in Congress. In women s issues, however, Republican women seem resistant to the influence of party polarization and remain distinctly more moderate in this set of 15

16 votes. For Education, Health, and Civil Rights votes, gender continues to have a significant effect on Republican women s ideology. REFERENCES Aldrich, John H Why Parties?: The Origin and Transformation of Political Parties in America. Chicago, IL: University of Chicago Press. Baumgartner, Frank R., and Bryan D. Jones Agendas and Instability in American Politics. Chicago: University of Chicago Press. Baumgartner, Frank R., Bryan D. Jones, and John Wilkerson Policy Dynamics. Chicago: University of Chicago Press. Box-Steffensmeier, Janet, Suzanna De Boef, Tse-Min Lin. The Dynamics of the Partisan Gender Gap. American Political Science Review, Vol. 98, No. 3 (August 2004), pp Burns, Nancy, Kay Lehman Schlozman, Sidney Verba The Private Roots of Public Action. Cambridge, Massachusetts: Harvard University Press. Canes-Wrone, Brandice, David W. Brady, John F. Cogan. Out of Step, out of Office: Electoral Accountability and House Members' Voting The American Political Science Review, Vol. 96, No. 1 (Mar., 2002), pp Carroll, SJ Women as Candidates in American Politics. Bloomington: Indiana University Press. Edwards, George C. III Presidential Electoral Performance as a Source of Presidential Power. American Journal of Political Science. 22.1: Epstein, Michael J., Richard G. Niemi, and Lynda W. Powell Do Women and Men State Legislators Differ? Fridkin, Kim, and Gina Serignese Woodall Different Portraits, Different Leaders? Gender Differences in U.S. Senators Presentation of Self. Gronke, Paul, Jeffrey Koch, and J. Matthew Wilson Follow the Leader? Presidential Approval, Presidential Support, and Representatives Electoral Fortunes. The Journal of Politics. 65.3: Jacobson, Gary C The Marginals Never Vanished: Incumbency and Competition in Elections to the U.S. House of Representatives, American Journal of Political Science, Vol. 31, No. 1 (Feb., 1987), pp Jochim, Ashley E., and Bryan D. Jones. Forthcoming. The Evolving Structure of Legislative Choice: Roll Call Voting in the House of Representatives. Kathlene, Lyn In A Different Voice. Lawless, Jennifer L. and Richard L. Fox It Takes a Candidate: Why Women Don t Run for Office. New York: Cambridge University Press. Mayhew, David R Congress: The Electoral Connection. New Haven: Yale University Press. Poole, Keith T., and Howard Rosenthal A Spatial Model for Legislative Roll Call Analysis. American Journal of Political Science 29(2): Poole, Keith T., and Howard Rosenthal Congress: A Political-Economic History 16

17 of Roll Call Voting. New York: Oxford University Press. Poole, Keith T., Nolan McCarty, and Howard Rosenthal Polarized America: The Dance of Ideology and Unequal Riches. Cambridge: MIT Press. Riker, William Heresthetic and Rhetoric in the Spatial Model. In Advances in the Spatial Modeling of Voting. New York: Cambridge University Press. Sanbonmatsu, Kira Democrats, Republicans, and the Politics of Women s Place. Ann Arbor: University of Michigan Press. Sapiro V Theorizing Gender in Political Psychology Research. Oxford Handbook of Political Psychology. New York: Oxford University Press. Sinclair, Barbara Party Wars. Norman: University of Oklahoma Press. Swers, Michele L The Difference Women Make. Chicago: University of Chicago Press. Theriault, Sean Party Polarization in Congress Cambridge University Press. Theriault, Sean, Patrick Hickey, and Abby Blass Roll Call Votes in The Oxford Handbook of the American Congress. New York: Oxford University Press. Theriault, Sean, Jennifer Lawless. Will She Stay or Will She Go? Career Ceilings and Women s Retirement from the U.S. Congress. Legislative Studies Quarterly, November : Volden, Craig, Alan Wiseman, Dana Wittmer. Forthcoming. The Legislative Effectiveness of Women In Congress. Winter, Nicholas J Framing gender: Political rhetoric, Gender Schemas, and Public Opinion on US Health Care Reform. Politics and Gender 1(3): Wolbrecht, C The Politics of Women s Rights: Parties, Positions, and Change. Princeton, NJ: Princeton University Press. 17

18 !!"#$%&'(!! Means of House Republicans' DW-NOMINATE Scores Congress GOP Men's mean score GOP overall mean score GOP Women's mean score! Table 1 Mean Extracted Dimensions for Eighteen Issue Areas (Jochim & Jones) Issue Area Science 2.67 Trade 2.53 Agricultre 2.44 Public Lands 2.10 Transportation 2.05 Energy 2.00 Foreign Affairs 1.85 Business 1.88 Crime and Law 1.65 Defense 1.63 Environment 1.53 Education 1.53 Civil Rights 1.42 Social Welfare 1.41 Health Care 1.29 Housing 1.29 Economics 1.28 Labor 1.16 Mean Dimensions 18

19 Table 2 Frequency of Roll Call Votes by Issue Area, 102 nd to 108 th Congress Issue Area Defense 704 Foreign Affairs 636 Public Lands 575 Crime and Law 498 Economics 495 Business 351 Health 328 Labor 299 Education 264 Transportation 259 Civil Rights 242 Environment 210 Science 207 Agriculture 174 Trade 162 Welfare 150 Energy 129 Housing 128 Number of Votes 19

20 Table 3 20

21 Figure 2 21

22 Table 5 The Effect of Gender, Constituency Ideology, Length of Time in the House, and Election Win Percentage on House Republicans' W-NOMINATE Scores for Civil Rights and Minority Issues Votes Congress Gender *** -.146* -.293*** -.128* (.195) (.075) (.070) (.061) (.063) (.049) (.050) District Pres. GOP *** 2.058*** 2.118*** 1.107*** 1.019*** 1.119*** % (.656) (.265) (.321) (.320) (.243) (.176) (.221) Years in House * (.006) (.003) (.003) (.003) (.002) (.002) (.002) Win % ** -.589*** (.302) (.167) (.168) (.230) (.111) (.100) (.108) Constant *** -.487***.295**.176 (.414) (.147) (.145) (.276) (.126) (.097) (.133) Adjusted R N Note: Dependent variable is Republican House members' DW-NOMINATE score. Figures represent unstandardized regression coefficient estimates and standard error estimates, using OLS estimation. *Statistically significant at.05; **statistically significant at.01; ***statistically significant at.001. Table 6 The Effect of Gender, Constituency Ideology, Length of Time in the House, and Election Win Percentage on House Republicans' W-NOMINATE Scores for Health Votes Congress Gender -.194* -.248*** ** -.108* -.076** (.084) (.066) (.068) (.084) (.037) (.051) (.022) District 1.124*** 1.301*** 1.915*** 1.508***.946***.734***.600*** Pres. GOP % (.300) (.240) (.312) (.314) (.142) (.186) (.091) Years in * -.007*** -.002* House (.003) (.002) (.003) (.003) (.001) (.002) (.001) Win % * -.477* * (.139) (.151) (.163) (.220) (.064) (.106) (.045) Constant -.621* **.303**.617*** (.192) (.132) (.141) (.161) (.073) (.104) (.054) Adjusted R N Note: Dependent variable is Republican House members' DW-NOMINATE score. Figures represent unstandardized regression coefficient estimates and standard error estimates, using OLS estimation. *Statistically significant at.05; **statistically significant at.01; ***statistically significant at

23 23

24 24

25 25

26 26

27 27

28 28

29 lkjlakjdfkljasfdl 29

30 30

31 31