Descriptive representation and evaluations of government

Abstract

A great deal of scholarship has looked at the effect that descriptive representation has on policy-making at the congressional level. Much less work has examined the effect of descriptive representation on individual citizens, and of the work that exists, most of it has focused on turnout. In this research, we look at the linkage between descriptive representation and political evaluations among the public. The theory of descriptive representation suggests that having a descriptive representative at the congressional level should positively affect an individual's attitude toward his or her US House legislator, and may extend to government institutions. We examine support for these arguments using data from the 2008 American National Election Study, which has an oversample of African-Americans and Latinos. We find that having a descriptive representative boosts evaluations of one's US House legislator for both African-Americans and Latinos, but has differential effects on attitudes toward government institutions.

Keywords:

• African-American politics
• Latino/a politics
• political behavior
• minority representation
• representation

Notes

1. There are some scholars who have challenged the efficacy of descriptive representation for African-Americans on the local and state levels (Critzer 1998; DeWeever 2000; Layman 1993).

2. Other scholars do not find that as a general rule voter turnout increases among African-Americans when they know their Congress member's race (Gay 2001; Griffin and Keane 2006; Tate 2003).

3. According to the last census, African-Americans make up 12.2% of the population, while Latinos make up 16.5% of the population. The disparity in representation in Congress is not as stark for Latinos if you take into account the voting age population: Latinos account for 8.4% of the voting age population.

4. Nunnally acknowledges that a number of scholars have found that instances of historical discrimination of blacks “have become the foci of explanations for their contemporary low-level political and social trust compared to other racial groups” (Abramson 1977; Brehm and Rahn 1997; Putnam 1995, 2000; Rahn and Tansue 1998; Walton 1985); but in her analysis she finds that there is no significant difference between African-Americans and Latinos when it comes to trusting the government at all levels—federal, state, and local (Nunnally 2012).

5. We determined this from the summary party identification measure available in the American National Election study. We count someone as a Democrat (Republican) if they are a strong/weak Democrat (Republican) or independent-leaning Democrat (Republican).

6. There was some concern that this measure may be highly correlated with our measures of descriptive representation, but that does not appear to be the case. The correlation between Black incumbent and Black majority district is positive as expected, but not that high, .58. The correlation between Latino incumbent and Latino majority district is even lower, .39. Furthermore, we estimated the variance inflation factor after all of our models and it never exceeded three, which suggests that collinearity is not a problem. Finally, if we run Model 1 without a control for Black majority district, the findings for having a Black representative remain the same for five out of the six dependent variables. If we do the same in the Latino models, none of the results for having a descriptive representative change.

7. This effect was calculated using CLARIFY (Tomz, Wittenberg, and King 2003) with all dummy variables set to their mode and the remaining independent variables set to the mean.

8. The results also remain the same if we add a linked fate measure to the Latino model. The linked fate measure is only significant for feelings toward the legislator and serves to boost evaluations by 6.49 units.

9. Furthermore, they argue that African-Americans with lower education and political knowledge are also more likely to have elevated perceptions of descriptive representation in Congress, and black majority districts are typically lower on these dimensions.

10. If we add a linked fate measure to our analyses, it is never significant.