Current Theories of White Voting Behavior

Studying Whites as a minority certainly can give scholars an important perspective. More than 50 years ago, V. O. Key (1949) convincingly explained why the White minority in the Black belt was the key to understanding Southern politics. Key's Black threat hypothesis, perhaps one of the longest-lived hypotheses in American political science, has profoundly influenced the study of racial politics in America (Taylor 1998; Glaser 1994; Giles and Hertz 1994; Giles and Buckner 1993). Key hypothesized that an increase in Black presence in a community leads to a higher level of Black threat perceived by the community's Whites. Applying this hypothesis to White voting behavior in New Orleans and Memphis, the Black threat thesis predicts that an increase in Black density, measured by the proportion of Black registered voters in an electoral unit, leads to a decrease in White support for Black candidates (Hypothesis 1 in this research).

Key's original theory (1949) was about the “maintenance of White supremacy” in the South, where Black disfranchisement was fairly common (5). The old context in which Key constructed his theory clearly was different from the post-Voting Rights Act era (Lublin 1997, 89). Indeed, Black politicians elected to powerful positions may no longer be just a “threat”; it has become a “reality” in many areas, especially at the local level (Liu 2003). One more issue in the empirical testing of the Black threat hypothesis concerns the proper geographic units that the Black threat theory should be based upon (Liu 2001b; Oliver and Wong 2003). While scholars of Black threat have found support for this hypothesis when Black density is measured at the county or other relatively large election unit level (e.g., Tolbert and Hero 2001), precinct-level or neighborhood analyses tend to provide counter-evidence (e.g., Welch, Sigelman, Bledsoe, and Combs 2001; Carsey 1995). Moreover, some studies have also discovered that the relationship between Black density and White backlash may not be simply linear. It is possible that White backlash against Black interests is the strongest when Black density reaches a critical level of around 20–30%, and Whites may become less discriminatory once they are in the minority (Keech 1968; Liu 2003).

More recently, scholars of racial interaction have provided a competing theory (Welch, Sigelman, Bledsoe, and Combs 2001; Huckfeldt and Sprague 1995; Huckfeldt 1986; Carsey 1995; Oliver and Wong 2003). Here the focus shifts to the positive social consequences of racial interactions. According to this theory, the more White voters communicate or encounter with Blacks, especially at the neighborhood level, the higher the possibility that these Whites will make friends with Blacks or become racially tolerant. With respect to White voting behavior, an increase in Black density (also measured by the proportion of Black voter registration in a voting precinct) may lead to a greater opportunity for racial interaction and, therefore, a higher level of White support for Black candidates (Hypothesis 2).

Critiques of the racial interaction hypothesis point to the possibility of a “self-selection” process that takes place when voters choose to live in a certain neighborhood (e.g., a racially mixed one) because they have already had a certain racial attitude to out-groups (e.g., racial tolerance) before they moved into that neighborhood (Pettigrew 1998; Powers and Ellison 1995; Oliver and Wong 2003). Furthermore, according to a recent study (Liu 2001a), the positive effect of racial contact on White racial tolerance may indeed exist, but this effect can be conditional on other factors, such as the overall group status and political power of Whites in a certain area.

Other than Black density, ideology has also been suggested as a determinant of White voting behavior (Browning, Marshall, and Tabb 2003; Sonenshein 1993). In this vein, the political incorporation theory emphasizes the coalition between White liberals and minorities to defeat White conservatism in urban America. Browning, Marshall, and Tabb (1984) argue that Black political control is vital to Black interests, and the biracial coalition between Blacks and liberal Whites is a key to the success of Black empowerment. More importantly, this coalition must be the “dominant coalition” to protect Black interests (also see Sonenshein and Pinkus 2005 for a case study of the racial coalition between Latinos and White liberals in Los Angeles). Based on this proposition, one may simply hypothesize that White liberals are more likely to support Black candidates in a biracial election than are non-liberal Whites (Hypothesis 3).

Though ideology has been a focus of many electoral studies, there is no consensus on the effect of a White belief system. Other scholars used the concept of symbolic racism, a much more pessimistic view than that offered in the political incorporation theory, to describe and explain the modern White moralistic and abstract resentments toward Blacks as a group. The subtle sentiments of symbolic racism are suggested as a reflection of the belief that “Blacks violate such traditional American values as individualism and self-reliance, the work ethic, obedience, and discipline” (Kinder and Sears 1981, 416; Kinder and Sanders 1996). Another limitation on the existing theories of White ideology is the difficulty in using them to trace the dynamic impact of racial transition on White political behavior. Arguably, when Whites become a minority in cities such as New Orleans and Memphis, their political power status is structurally changed. As a result, Whites may develop a new approach to Black political power to maximize their own political power, though their ideology may or may not remain the same (for an early theoretical treatment of White reaction to contextual changes, see Eisinger 1974 about White “pragmatic coalition”). Furthermore, a scholar who uses the conceptual framework of White ideology may find it very difficult to distinguish whether Whites' racial hostility is based on real-world concerns or on racial stereotypes (Bobo 1983).

White Voting Patterns in New Orleans and Memphis' Mayoral Elections

This paper focuses on New Orleans and Memphis, two major Southern cities that Key (1949) argued are the key to understanding the racial politics of Louisiana and Tennessee. Indeed, given the racial tensions that have existed historically in these two cities (e.g., see Hirsch 1997; Pohlmann and Kirby 1996), it is reasonable to assume that if a successful biracial coalition can be formed in these two cities, then a same or similar coalition can be formed in other cities where racial tension has not been as high. Both New Orleans and Memphis use a mayor-council form of government and a nonpartisan election system. Figure 1 provides the changes in White support for the leading Black mayoral candidates who won the respective elections in Memphis and New Orleans since 1991. These city-wide estimates were based on the Ecological Inferences (EI) basic model developed by Gary King.¹

Clearly, both cities exhibited the same pattern: White support for Black candidates started at a very low level before increasing substantially to more than 40% in the next 15 years. One may speculate that Whites became more liberal over time in their racial attitude to Black electoral challenges in the two cities. Yet, a closer look at New Orleans' longitudinal White voting data from 1977 to 2002 indicates that 1994 marked a significant drop in White support for Black candidates (see Figure 2). In other words, there has been no consistent growth in White support for Black candidates over time. Therefore, to understand the surges and declines of White support for Black candidates, one has to pay attention to the contexts in which the elections took place.

Mayoral Elections in Memphis

All of the mayoral elections presented in Figures 1 and 2 were biracial elections involving both Black and White candidates. Thus, there was a chance for the White voters to support White candidates in these elections. Indeed, in both cities Whites provided a very low level of support for Black candidates in the early 1990s. In Memphis, the White support for Dr. Willie W. Herenton, the leading Black candidate and former school board superintendent of Memphis, was less than 10%. However, Herenton carried more than 95% of the African-American vote and benefited from a court decision that stopped the use of mayoral runoff election in Memphis to defeat the White incumbent, Dick Hackett, by only 142 votes in the primary, becoming the first elected Black mayor in Memphis history (Pohlmann and Kirby 1996).

Herenton easily won reelection in Memphis with his share of the White vote increasing to more than 30% in 1995. One may quickly attribute this success to the incumbency advantage. Indeed, in 1995 Herenton had no opposition from any Black candidate, and none of Herenton's three White challengers could muster a serious threat to his reelection. The only White candidate with some name recognition was John Baker, who had unsuccessfully run for the U.S. Senate.

In the following 1999 mayoral election, as many as 15 candidates joined the race—a sign that Herenton's incumbent advantage was challenged. The major White candidate was Pete Sisson, a county commissioner. Herenton managed to maintain his White crossover vote at the 37% level, about the same as his White crossover level in the 1995 race. White support for Herenton in this election, however, was crucial to his reelection because his Black support dropped more than 45%. The division in the Black community was largely due to a strong challenge from City Council Chairman Joe Ford, a Black candidate and former Congressman Harold Ford's brother, who accused Herenton of selling out Black interests to White downtown development projects (Wright 2000). Ford was defeated in the election, but almost half of the Black electorate voted for him.

By 2003, Herenton, now a three-term incumbent, had become increasingly popular in the White community. Herenton received 68% of the votes cast in the 2003 mayoral election to defeat seven other candidates, including John Willingham, a White candidate and a Shelby County commissioner. Herenton's White crossover increased to about 42%. Herenton is currently in his fourth term, longer service than any previous mayor.

Mayoral Elections in New Orleans

In New Orleans, the 1994 mayoral contest was mainly between Donald Mintz and Marc Morial. Mintz was the major White candidate who had entered into a runoff with the Black incumbent mayor, Sidney Barthelemy, in the 1990 race. Marc Morial, the major Black candidate, is a son of Ernest “Dutch” Morial, the first Black mayor in New Orleans. With the local White media alerting the White community that the election might be “the last chance to elect a White mayor” (Engstrom and Kirkland 1995, 8), Morial openly suggested that only his election would stop the White challenge to the 16-year Black control of city hall. He urged the Black community to unite behind him to defeat Mintz. The outcome of the 1994 mayoral race was the most racially divided election in New Orleans' history (Engstrom and Kirkland 1995, 6). Morial won the election, but he received only 3% of the White vote in the primary and 7% in the runoff.

In 1998, Morial successfully won reelection. His two White challengers, lawyer Kathleen Cresson and arts store manager Paul Borrello, were newcomers to New Orleans' local politics and were not a real threat to Morial's reelection (Perry 2003). Despite the weak competition from these White candidates Morial still failed to gather a majority from White voters—his White crossover was only about 38% in 1998. After the election, Morial attempted, but failed, to persuade the city's voters to eliminate or modify the term limit law. Not able to run for a third term, he retired after eight years as mayor.

The 2002 mayoral election quickly attracted 15 candidates including a famous White professor and constitutional scholar from Tulane University. Two Black candidates made it to a runoff. As a surprise to many, C. Ray Nagin, the general manager/vice president of Cox Communications and co-owner of the Brass hockey team, defeated Richard J. Pennington, the popular police superintendent widely regarded as a surrogate candidate for the Morial administration (Perry 2003).²

Nagin's success, based on a deracialized campaign strategy, was largely due to White support. He received 50% of the White vote in the primary and 86% in the runoff. Both were record-breaking White support for a Black candidate in New Orleans history, but his Black support was as low as 12% in the primary and only 39% in the runoff.

To summarize, when White voters rallied behind the major White candidates in the 1991 Memphis mayoral election and the 1994 New Orleans mayoral race, the White candidates were defeated. These defeats seemed to persuade many White voters that continuing to vote for White candidates would further limit their own electoral power. To see the strategic adjustment of White voters in subsequent mayoral elections, it is important to focus on the 1999 Memphis election and 2002 New Orleans race where White voters played an instrumental role in electing Herenton and Nagin and defeating Joe Ford and Richard Pennington, their major Black opponents who were popular in the Black communities.

The pivotal role of the White minority in electing their favorable Black candidates in the two cities' subsequent mayoral elections was intrinsically linked to the racial contexts which structured the overall political power of White voters. In Memphis, based on the voter registration data, the electorate changed from 48% Black at the time of the 1991 mayoral contest to 51% Black in 1995. By 2000, the city's population was 61% Black and 34% White.

In New Orleans, the electorate was 54% Black in 1990 and 63% Black in 1998. By 2000, the city's population was 67% Black and 27% White.³

Clearly, the racial transition has provided a better chance for Black candidates to win in both cities, while White candidates faced increasing difficulty to win as a minority in the cities' elections. Thus, continuing to vote for a White candidate was likely to produce White defeats. It became rational for White voters to strategically vote for a “friendly” Black candidate to at least help defeat “threatening” Black candidates. The following sections use both aggregate and individual-level data to test the three hypotheses (the Black threat (H1), racial interaction (H2), and political incorporation (H3) hypotheses) based on the existing theories of White voting behavior.

Findings

Table 1 reports three equations that use precinct-level estimates of White crossover voting for Herenton in the 1999 election and Nagin in the 2002 runoff as the dependent variables. Equations 1 and 2 are simple bivariate regressions with precinct-level Black density as the independent variable. The regression coefficients are not statistically significant, which suggests that, unlike what the Black threat and racial interaction hypotheses proposed, there is no relationship between Black density and White crossover voting at the precinct level. Thus, there is no empirical support for Hypotheses 1 and 2 based on the aggregate-level data in New Orleans and Memphis.

According to the political incorporation theory (H3), there should be a higher level of support for the Black candidate of the biracial coalition from White liberal Democrats than from White Republicans. Equation 3 therefore adds one more independent variable, the percentage of registered voters who are White Republicans, into the New Orleans model. The R-squared increased from .008 to .201, indicating that the new model explains about 20% more variation in the dependent variable. Thus, Nagin's White support was from White Republicans, certainly not from White liberals.⁴

Moreover, the Black density variable also becomes statistically significant, which shows that with the level of White Republican voters held constant, White support for Nagin increases as Black density increases.

To further test if Black density and party identification are statistically significant once other factors are controlled, we turn to our logistical regression models based on the individual-level survey data collected by the University of New Orleans Survey Research Center. The dichotomous dependent variable is based on the question, “[a]s things stand today, who are you most likely to vote for in the election for Mayor of New Orleans, Ray Nagin or Richard Pennington?” (Recoded as “1” if the answer is Nagin, “0” if the answer is Pennington.) The logistic regression models are employed to examine the voting decision of White and Black voters.

The racial context variable is used to test H1 and H2, which suggest that the Black density of a neighborhood may affect the probability of Whites voting for Nagin. However, according to Equation 1 of Table 2, racial context, or Black density at the neighborhood level, did not affect White voters' decision to vote for Nagin. Thus, survey data do not provide empirical support for H1 and H2 (see Appendix for survey question wording and coding).

Party affiliation, on the other hand, is a predictor of Whites' choices. Democrats, contrary to political incorporation theory (H3), were less likely to vote for Nagin. This seems to support an earlier observation reported in Equation 3 of Table 1. It is also in line with Perry's (2003) argument that Nagin led a conservative coalition in the 2002 mayoral election. However, based on the logistic regression model in Equation 1, which includes two dummy variables to measure party affiliations, Nagin's White support was actually neither from Democrats nor from Republicans. In fact, those Whites who identified themselves as “independent” were most likely to support Nagin.

Other control variables also reveal very important findings: Nagin's White supporters were those who emphasized the importance of his previous business experience; gave negative evaluations of the Morial administration; and knew that the major White-controlled local newspaper, the Times-Picayune, had endorsed Nagin. Moreover, these White voters were, comparatively speaking, older, and were not affected by the negative TV commercials attacking candidates.

For Black respondents, Equation 2 shows that only two independent variables are statistically significant. Similar to White voters, Black respondents also indicated that Nagin's previous business experience positively influenced their support for him in the mayor election. This finding suggests that both White and Black groups stress the importance of their city's future economic success as the reason to support Nagin. The main difference between Black and White voters regarding their support for Nagin is that only Black voters who had high levels of family income were likely to support Nagin in the runoff. In other words, social class was an indictor of Nagin's coalition membership only for Black voters. For White voters, family wealth was not a factor.⁵