The Role of Personal Attributes in African American Roll-Call Voting Behavior in Congress
Rocca, Michael S., Sanchez, Gabriel R., Nikora, Ron, Political Research Quarterly
In this research note, the authors explore the extent to which personal attributes influence voting behavior for African American members of Congress. The authors test the relationship between legislators' personal attributes and Poole and Rosenthal's DW-DOMINATE scores for black members of the 101st to 108th Congresses. The results suggest that personal attributes matter. They increase the explained variance in the models, and factors such as generational cohort, religion, and military experience are statistically significant. These results suggest that while descriptive members of Congress share much in common, they should not be interpreted as monolithic with respect to congressional voting.
Keywords: descriptive representation; substantive representation; attributes; Latino; African American
The purpose of this research note is to examine the extent to which the personal attributes of black members of Congress (MCs) influence their voting behavior. The motivation for the study stems from two primary sources. First, the 2006 congressional elections led to an unprecedented number of black MCs chairing full committees: Charles B. Rangel (D-NY) of Ways and Means, the late Juanita Millender-McDonald (D-CA) of House Administration, John Conyers Jr. (D-MI) of Judiciary, and Bennie Thompson (D-MS) of Homeland Security. Since black MCs tend to behave differently than white MCs (e.g., Swain 1993; Lublin 1997; Canon 1999; Tate 2003), notwithstanding Swain's (1992) analysis, it is reasonable to expect their newly acquired positions of influence to facilitate substantive changes in policy agendas.
Second, much of the work on black representation in Congress has tended to treat black MCs as a monolithic group (e.g., an MC is coded as either black or not black). But Canon (1999) and Tate (2003) are among those who have found interesting variation within the Congressional Black Caucus. Similarly, (2002) analysis contends that minority and female representatives are not monolithic and that some descriptive representatives are preferable to others. We applaud these efforts to move beyond a monolithic view of descriptive representation and advance this argument by exploring the diversity among black MCs and how it may affect their rollcall vote decisions.
We approach this study from the perspective that personal attributes matter to how MCs vote. We believe representatives' descriptive characteristics are important determinants of their ideology, a factor that students of Congress have long shown to be a key predictor of congressional voting (see Kingdon 1989; Poole and Rosenthal 1997). Our theory is grounded in the work of Easton and Dennis (1969) and Jennings and Niemi (1974), who argued that attitudes are a result of a lifelong process of political socialization and learning. To test our theory, we focus on five general agents of socialization for black MCs: education, religion, generational cohort, military experience, and nativity relative to their district. We also address gender, a factor that has consistently been shown to matter for congressional behavior.
Our results show that even after controlling for institutional and electoral factors, personal attributes indeed matter to how black MCs vote. This suggests that scholars interested in descriptive representation should move beyond simple dichotomous measures for race, ethnicity, or gender when studying minorities' congressional behavior.
Data and Method
We used Amer's (2005) CRS report as well as The Almanac of American Politics (Barone and Ujifusa 1990, 1992, 1994, 1996, 1998, 2000, 2002, 2004) to identify sixty-seven black members of the House during our period of study. The unit of analysis is the legislator-Congress, and the data take a cross-sectional time series form. Because most individuals served more than one term during this period of study, most appear multiple times in the data. To account for the nonindependence across observations, we cluster individual MCs in our ordinary least squares regression model. …