BLACK MAYORS/WHITE MAYORS
EXPLAINING THEIR APPROVAL
SUSAN E. HOWELL
HUEY L. PERRY
Abstract Performance models have been the norm in research on
national and state executives, but oddly this model has rarely been
extended to the next level of executive office, mayors. The increasing
number of African American mayors suggests that race may complicate
the performance model of approval at the local level. This research tests
a performance model of mayoral approval that takes racial factors into
account. The model is tested in two white cities with white mayors and
two black cities with black mayors. Performance is measured by citizen
evaluations of a variety of urban conditions, some of which citizens can
observe firsthand. Findings indicate that (1) performance matters in
evaluating black and white mayors, and its influence does not seem to
be related to the race of the mayor; (2) performance has more
explanatory power over mayoral approval than race, a pattern that holds
for both black and white mayors; (3) race has more impact on the
approval of black mayors than on white mayors; and (4) blacks in black
cities have lower evaluations of local government performance in their
cities than blacks in white cities. We conclude that the performance
model is generalizable to the local level, even with the complications of
race.
The Performance Model
Most of the existing literature on executive approval is based on a performance
model, which emphasizes citizen evaluations of conditions in the country or
SUSAN E. HOWELL
is professor of political science at the University of New Orleans. HUEY L. PERRY
is professor and Chancellor’s Fellow in the Department of Political Science at Southern University—Baton Rouge. We are grateful to the National Science Foundation for funding this project,
and to Frank P. Scioli Jr. and Marianne C. Stewart for their counsel in preparing the proposal. We
would also like to thank Gary King, Sara Benesh, and Matthew Vile for methodological advice
along the way, and Christine Day, Phillip J. Ardoin, and anonymous reviewers for comments on
an earlier draft. This project could not have been completed without the efforts of the graduate
assistants with the University of New Orleans Survey Research Center, Matthew Vile, Heidi
Unter, Manabu Saeki, and Monica Farris. Address correspondence to Susan E. Howell; e-mail:
[email protected].
Public Opinion Quarterly, Vol. 68 No. 1 Pp. 32–56, © American Association for Public Opinion Research 2004; all rights reserved.
DOI:10.1093 / poq / nfh003
Explaining Mayoral Approval
33
state. The concept, “performance,” is quite broad, and there is no single performance model. However, there are common themes among studies of executive approval. Performance can be conceived of as the quality of social,
economic, and international outcomes experienced by the public (Ostrom and
Simon 1985). The emphasis in this model is on outcomes. It is outcomes and
perceptions of those outcomes—not policy proposals or announcements—that
are crucial influences on executive approval (Brody 1991). Performance can
theoretically cover an infinite number of subjects, but existing research on
presidential approval has emphasized economic outcomes and foreign policy.
There is ample evidence that economic and foreign policy outcomes directly affect
presidential popularity (Edwards, Mitchell, and Welch 1995; Hetherington
and Nelson 2003; Hibbs 1982; Kernell 1978; MacKuen 1983; MacKuen,
Erikson, and Stimson 1989; Nadeau et al. 1999; Ostrom and Simon 1985). At
the state level, where foreign policy is not applicable, evidence from
single state studies (Howell and Vanderleuuw 1990) and from national data
(Niemi, Stanley, and Vogel 1995) indicates that economic conditions impact
gubernatorial popularity and gubernatorial elections in a fashion similar to
that of the presidency.
Beyond economic performance, the research on retrospective voting indicates
that evaluations of past performance in areas of public policy are generally
quite important predictors of voting for an incumbent executive (Fiorina 1981;
Miller and Wattenberg 1985). While this research has focused on voting as a
dependent variable, it is easy to see the applicability to approval of incumbents.
The key theoretical question in our research is whether the performance
model of executive approval applies to the local level. Oddly, this model has
rarely been extended to the next level of executive office, mayors (for an exception see Howell and McLean 2001). Consistent with the research on national
and state executive approval, a city’s economy can be one basis for judging
mayoral performance. However, there are other sentiments about conditions in
the city that could also serve as evaluations of performance. Citizens can evaluate cities in terms of control of crime, conditions of streets, housing, parks,
public transportation, and schools. In addition, these aspects of urban life can
be observed up close in the course of daily life, in contrast to conditions in the
state or the nation. Thus, it is reasonable to expect that the performance model
of approval can apply to mayors as well as to governors and presidents.
Race as a Confounding Factor
There is a major complication to applying the performance model to the local
level. Because of the increasing number of African American mayors and
majority-minority cities, race may be a confounding factor in the operation of
the performance model in racially diverse settings with histories of racial
conflict. If the performance model were applied to the local level without
34
Howell and Perry
considering race and racial factors, it might well suffer from omitted variable
bias. That is, the model would provide an incomplete explanation, and the
estimates of the impact of performance on mayoral approval could be overestimated or underestimated.
There are four ways in which race could affect the performance model in
racially diverse settings, which we define as settings where the minority race
comprises at least 20 percent of the adult population, and where there is a history of racial conflict. First, racial categorization as a representation of group
interests could directly affect mayoral approval. The literature provides generous
evidence of race-based voting in biracial mayoral contests (Adler 2001;
Browning and Marshall 1986; Browning, Marshall, and Tabb 1984; Engstrom
and Kirkland 1995; Huckfeldt and Kohfeld 1989; Kleppner 1985; Vanderleeuw
1990). Given the strong effects demonstrated in these studies, there is
certainly reason to expect that the racial effects in the election will extend to
approval of whomever is elected. The expectation is that race affects mayoral
approval over and above the effect of performance.
Second, race may be antecedent to the performance model and affect citizens’ perceptions of conditions in the locality. If this is so, omission of race
from the performance model provides an incomplete description of how the
model works. Since it is well established that, due to different life experiences, group interests, and socialization, blacks and whites have different perceptions across a variety of domains (Kinder and Sanders 1996), they may
also have differing evaluations of local government services, crime, employment prospects, and other conditions. Two conditions in particular are likely
areas for racial polarization: perceptions of the local economy and evaluations
of the local police. The economic disparity between blacks and whites in most
areas implies that blacks would have lower evaluations of the economy than
whites. Also, the greater distrust of police among blacks has been well
researched (Campbell and Schuman 1968; Decker 1981; Perry 2000; Smith,
Graham, and Adams 1991; U.S. Department of Justice 2000).
The third way in which race affects the performance model is through the
influence of racial attitudes on both the evaluations of performance and on
mayoral approval. White citizens’ evaluations of a black or white mayor in a
racially diverse setting may be partly determined by their racial attitudes. That
is, whites with higher levels of racism will be more negative toward black
mayors and toward conditions in majority black cities. In majority white settings, whites with higher levels of racism will be more positive toward white
mayors and toward conditions in majority white cities. White racial attitudes
have been shown to be predictive of voting behavior in mayoral elections
involving Tom Bradley in Los Angeles (Kinder and Sears 1981; McConahay
and Hough 1976; Sears and Kinder 1971). Also, recent research on white
racial attitudes has emphasized the role of “group conflict”—white resentment
of blacks due to their being competitors for privileges and resources (Bobo
1998, 2000; Bobo and Hutchings 1996). The notion of competition for
Explaining Mayoral Approval
35
resources is quite relevant in an urban setting where racial groups “compete”
for tangible resources.
Blacks’ racial attitudes could also affect the performance model in a similar
fashion. The most crucial black racial attitude is some form of black consciousness (Dawson 1994; Gurin, Hatchett, and Jackson 1989; Shingles 1981;
Verba and Nie 1972). Our expectation is that blacks with higher levels of
black consciousness are more approving of the black mayor and more positive
about conditions in a majority black city, while in a majority white city blacks
with higher levels of black consciousness are less approving of the white
mayor and more negative about conditions in the white city.
The fourth and final way in which race may confound the performance
model is through the racial context, that is, the racial composition of the local
setting. Racial context has been found to affect racism, racial policy attitudes,
and party identification among whites (Giles and Hertz 1994; Glaser 1994;
Taylor 1998, 2000; but for contrary evidence see Carsey 1995; Voss 1996).
Racial context also affects political trust and participation among blacks
(Bobo and Gilliam 1990; Howell and Fagan 1988). In this research, racial
context is hypothesized to affect the model primarily through interactions with
other variables. A majority black context is hypothesized to negatively affect
evaluations of conditions in the city, but the effect will be greater among
white citizens than among black citizens. Majority black contexts are often
plagued by problems such as dwindling jobs, poor schools, crime, and deteriorating housing (Adler 2001; Barnes 1994; Kraus and Swanstrom 2001; Nelson
1990; Preston 1990). We hypothesize that these conditions will be met with a
more negative response among whites because of their need to adopt beliefs
and attitudes that support their dominant position in society (Sidanius et al.
2000). Blaming the black majority, which is also the “outgroup,” gives justification for negative perceptions of conditions in the majority-minority city.
Racial context can also interact with race to impact mayoral approval.
Because a majority black setting is the exceptional situation for most Americans, we expect racial polarization in mayoral approval to be greater in those
settings. African American executives bring greater attention to racial matters
simply because they are exceptions to the “norm” of white executives.
In addition to addressing the theoretical question of application of the performance model to the local level, there are implications for governance and
legitimacy. Some cities have had black mayors for more than twenty years,
which should have reduced their novelty. If these mayors are not evaluated in
terms of performance, it is a pessimistic conclusion for urban race relations, as
well as a complication for governance by black mayors who often have to
negotiate with a white business elite who controls a disproportionate amount
of the wealth (Stone 1989). On the other hand, if performance is the yardstick
for evaluating both black and white mayors, it indicates that urban residents
are becoming less sensitive to race in evaluating urban officials (see Moesher
and Silver 1994 and Sonenshein 1989 for evidence of this trend). As prescribed
36
Howell and Perry
by democratic theory, the mayor gains or loses legitimacy biracially based on
citizens’ evaluation of performance.
The performance model of mayoral approval and the effects hypothesized
for racial factors are summarized in the following eight hypotheses. All apply
specifically to racially diverse cities with histories of racial conflict.
H1: Blacks are more negative than whites on evaluations of performance,
particularly evaluations of the local economy and the local police.
H2: In a majority black context, white racism is negatively related to evaluations of performance. In majority white contexts, white racism is
positively related to evaluations of performance.
H3: In a majority black context, black consciousness is positively related
to evaluations of performance. In majority white contexts, black consciousness is negatively related to evaluations of performance.
H4: A black majority context negatively affects evaluations of performance, but the effect is greater among whites than among blacks.
H5: Performance is positively related to approval of both black and white
mayors.
H6: Race affects mayoral approval above and beyond the effects of performance. This effect is expected to be greater for black mayors.
H7: Conservative white racial attitudes are positively related to approval of
white mayors and negatively related to approval of black mayors.
H8: Black consciousness is positively related to approval of black mayors
and negatively related to approval of white mayors.
Data and Methodology
The data for this research are from surveys of black and non-black (hereinafter
called “white” for simplicity)1 registered voters in four cities, Chicago, New
Orleans, Detroit, and Charlotte, NC. The following criteria were used to select
the cities:
1.
Two of the cities have black majorities and black mayors, and the other
two cities have white majorities and white mayors, creating differentiation
on majority race in the city and race of the mayor. A majority is defined
as over 50 percent of the voting age population.
1. Self-identified whites dominate the non-black registered voter samples from each city: New
Orleans—95 percent, Detroit—83 percent, Charlotte—91 percent, Chicago—80 percent. To test
whether the inclusion of Hispanics altered the results, we estimated the Chicago and Detroit models in table 6 including only self-identified whites and blacks. The results were amazingly similar,
with nearly identical adjusted R2’s, standard errors of the estimate, and insignificant differences
between all regression coefficients. Details on this analysis are available from the authors. Therefore, including Hispanics does not reduce racial polarization or change the impact of the performance dimensions on mayoral approval. This may be because we screened for registered voters,
therefore obtaining the more educated and assimilated Hispanics.
Explaining Mayoral Approval
37
2.
Two cities were chosen from the South and two from the non-South in
order to differentiate on region.
3. Those residents not in the majority race had to comprise at least 20 percent of the city’s population over age 18, and the city had to have a history of black/white racial tension.
4. All four mayors had been in office at least three years at the beginning
of data collection. This condition was necessary to allow the mayor’s
performance to affect his approval rating. A newly elected mayor cannot expect to be evaluated on performance.
5. The two black mayors could not be the first black mayors in their
city—in order to reduce the novelty factor.
With four cities the question of generalizability inevitably arises. We make
no claim that these cities represent some larger population. We are, however,
moving beyond the typical single city approach used in local urban research
by purposely varying the majority race of the city and the race of the mayor.
This permits an initial test of the performance model at the local level while
taking into account the confounding effects of race.2
Use of only four cities requires that we set some standards for confirming a
hypothesis. We only draw inferences when the pattern exists either in three of
the four cities, in both black cities but neither white city, or in both white cities
but neither black city. Anything less would mean that we would be drawing a
conclusion based on only one city.
Within each city a nearly equal number of blacks and whites were interviewed. Thus, instead of a cross-sectional representation of each city, we have
maximized variation on the key variable of race (King, Keohane, and Verba
1994).3 Cross-sectional random digit dialing samples of each city’s blacks and
whites were utilized with a screen for registered voters. In order to reach an
equal number of minority race respondents after exhausting the cross-sectional
sample, areas of 40 percent or more minority race concentration were used to
obtain the needed number. Only registered voters within the city limits were
2. There are always questions when certain locations are selected. One alternative considered
was to use more cities and fewer respondents from each city. The problem was the shortage of
black mayors who were not the first black mayor or who had been in office three years. There
were plenty of white mayor locations but far fewer black mayor locations meeting the criteria, so
we settled on basically a four case study design. We also avoided the West so as not to confound
the analysis with large low-income Hispanic populations. This article concentrates on black/white
conflict and leaves a Latino study to future research. The percentage of the population over age 18
that is black in these cities in the 2000 census is as follows: Detroit—79.6 percent, New
Orleans—61.8 percent, Charlotte—29.9 percent, and Chicago—33.8 percent (www.census.gov).
This study examined only registered voters living in the city limits whose approval of the mayor
is politically influential.
3. As a result, we have nearly an equal number of blacks living with a white mayor in a white city
as blacks living with a black mayor in a black city. It is not the location, Chicago or Detroit, that
matters conceptually. It is the race of the mayor and the racial composition of the city that are of
concern to this study. Thus, the samples were designed to equally represent cells of a 2 × 2 table of
respondent’s race and race of the mayor and city. The samples are not designed to be representative
cross sections of each city, only representative samples of blacks and of whites within each city.
38
Howell and Perry
included because they can potentially vote for the mayor. More detail on the
sampling procedure is provided in the appendix. The final number of respondents by race and by city is presented in table 1.
The dependent variable, mayoral approval, is a simple question about how
the mayor is “handling his job,” with five response categories ranging from
“strongly disapprove” to “strongly approve”, and with “don’t know” as a middle category. This indicator is similar to those used at the national and state
levels. The indicators of performance are perceptions of various aspects of
urban life, such as evaluations of police protection, traffic control, public
transportation, streets, services for the poor, parks, control of water pollution,
and other municipal services described in the appendix. Respondents were
also asked for their general evaluations of conditions in the city, perceptions
of crime and personal safety, the city’s economic outlook, and public schools,
totaling 27 individual performance items. Thus, we take a more comprehensive approach to performance evaluations than the national studies, which
usually rely on one or two performance areas (Edwards, Mitchell, and Welch
1995; Nadeau et al 1999).4 The large number of performance indicators was
utilized because there are many areas in which city residents can reasonably
evaluate local performance, in contrast to evaluations of national and state
conditions where information and experience are less easy to come by. Furthermore, other than economics, we did not have preconceived notions as to
which aspects of performance were most important.
To facilitate analysis, we reduced 25 of the indicators to 6 dimensions—
General evaluations, evaluations of Police, economic outlook or Economy,
city Services, perceptions of Safety, and Transportation—informed by an
exploratory principal axis factoring procedure. The indicators comprising
each factor were combined into additive indices (not factor scores) and coded
to range from 0 = negative to 1 = positive.5 Two indicators, perceptions of
increasing/decreasing Crime and evaluations of the Schools, appeared to be
both substantively and empirically distinct, so they are used as single indicators, scored from 0 to 1, yielding a total of 8 dimensions of performance.
4. Most of these items can be considered indicators of the quality of life in the city. However, it
should be noted that some items are only relevant to the quality of life of city residents who use
the service. For example, low-income people are most affected by the quality of “services for the
poor” and the quality of “public transportation.”
5. Principal axis factoring with varimax rotation was used to obtain a first look at how the indicators fell into dimensions. There were clear commonalities across cities, but, of course, not every
indicator had its largest loading on the same dimension in every city. We wanted the content of
the performance measures to be identical across the four cities in order to make meaningful comparisons. Therefore, we did not use factor scores, the content of which differed from city to city.
Both the factor analyses and the content of the items were used to select items for additive scales.
Details on the factor analyses are available from the authors. The reliability of the additive scales,
as measured by Cronbach’s alpha, are as follows for New Orleans, Detroit, Charlotte, and Chicago, respectively: Services, .81, .79, .76, .85; Police, .82, .82, .81, .82; Safety, .69, .71, .66, .69;
Economy, .68, .72, .70, .74; Transportation, .53, .57, .49, .62; General, .51, .63, .57, .63.
Explaining Mayoral Approval
39
Whites’ racial attitudes are represented by the concepts of symbolic racism
and group conflict. Symbolic racism (Sears 1988; Sears et al. 1997) is measured by four items selected due to their inclusion in the 1986 through 1994
American National Election Studies. Group conflict is measured by four items
developed by Bobo and Hutchings (1996) measuring whites’ views of racial
competition for resources.6
Our measures of black consciousness are Dawson’s indicator of “linked
fate” (Dawson 1994, p. 77) and four items measuring black solidarity, all of
which have been utilized in one or more of the National Black Election Studies.
Racial Polarization in Mayoral Approval
Mayoral approval is racially polarized in all four cities, but not always in the
expected direction (table 1). In three of the cities the predictable pattern exists,
with whites more likely to approve of the white mayor and blacks more likely
to approve of the black mayor (correlations of −.200 in Chicago, −.244
in Charlotte, and .292 in New Orleans). However, in Detroit, former mayor
Dennis Archer, an African American, was more popular among whites than
among blacks (r = −.219), an anomaly we will return to shortly. In all cities the
difference between whites and blacks is strongest in the categories of
“strongly approve” and “strongly disapprove,” indicating that intensity of support, as well as direction, is affected by race. The large standardized residuals
in the most intense cells also confirm the racial polarization in intensity.
Larger standardized residuals in a cell indicate a greater difference between
the observed and expected values for that cell. This is an indication that performance models of mayoral approval in these settings should include race as
an explanatory factor.
The “reverse” racial polarization in Detroit presents an interesting, and
theoretically important, exception to the “rule” that citizens are more likely to
approve of same-race mayors than are citizens of the other race. While Archer
was approved of by a majority of both whites and blacks, his approval rating
was higher among whites. This case is theoretically important because it demonstrates that racial polarization can exist even in the reverse direction due to the
persistence of different racial agendas. Mayor Archer disappointed black voters
on several fronts. He did not take a strong stand against welfare reform, which
6. To test whether or not these white racial attitude items constitute two dimensions, a principal
components analysis was conducted. The symbolic racism and group conflict items separate into
two factors. When the rotation is oblique, the correlation between the two underlying factors is
only .09. Furthermore, the reliability of a single scale with symbolic racism and group conflict
together is only .43, as measured by Cronbach’s alpha, while the two separate scales have alphas
of .55 (symbolic racism) and .80 (group conflict). The reliability of the symbolic racism scale in
this sample of whites in four cities is somewhat lower than in national samples. Sears et al (1997,
p. 26) report .78 (1986 National Election Studies, eight items), .76 (1992 National Election Studies, six items), and .65 (1994 GSS, four items).
Table 1.
New Orleans
(Black City)
Mayor Approval by Race
Strongly Disapprove
Disapprove
Don’t Know
Approve
Strongly Approve
Pearson’s r = .292***
Detroit
(Black City)
Strongly Disapprove
Disapprove
Don’t Know
Approve
Strongly Approve
Pearson’s r = −.219***
Charlotte
(White City)
Strongly Disapprove
Disapprove
Don’t Know
Approve
Strongly Approve
Pearson’s r = −.244***
Chicago
(White City)
Strongly Disapprove
Disapprove
Don’t Know
Approve
Strongly Approve
Pearson’s r = −.200***
NOTE.—Standardized residuals in parentheses.
*** p < .001.
40
White
Black
23%
(3.1)
12
(1.2)
7
(.3)
34
(1.1)
24
(−3.7)
N = 251
8%
(−3.1)
7
(−1.2)
6
(−.3)
26
(−1.1)
53
(3.7)
N = 253
12%
(−2.0)
6
(−2.1)
5
(−.3)
34
(.2)
43
(2.6)
N = 250
23%
(1.9)
14
(1.9)
6
(.3)
32
(−.2)
25
(−2.4)
N = 282
3%
(−2.2)
3
(−1.6)
7
(−1.9)
37
(.3)
50
(2.2)
N = 273
11%
(2.3)
7
(1.7)
15
(2.0)
35
(−.3)
32
(−2.3)
N = 255
6%
(−1.9)
6
(−1.5)
3
(−.7)
29
(−.4)
56
(2.1)
N = 250
13%
(1.9)
12
(1.5)
5
(.7)
32
(.4)
38
(−2.0)
N = 262
Explaining Mayoral Approval
41
was supported by Michigan’s Republican governor, and, according to some
observers, the economic revitalization of the city benefited whites much more
than blacks (Farley, Danziger, and Holzer 2000). Furthermore, Archer was simply was not Coleman Young, the charismatic leader of the civil rights movement in Detroit, who was the first African American mayor in that city and
served for twenty years. Mayor Archer’s style was more technocratic and legalistic. The reverse racial polarization in Detroit confirms other analyses suggesting that personal style has implications for the mayor’s relationships with white
and black constituents (Perry 2000, 1990). Thus, for both instrumental and symbolic reasons, Mayor Archer was more popular among whites than blacks.
How Racial Factors Influence Perceptions of Performance
RACE
Table 2 summarizes ordinary least squares models for each city with the performance dimensions as dependent variables, and race and a series of common
demographic controls as explanatory variables. The first conclusion one can
draw is that the effects of race on perceptions of performance depend on
whether one is considering a majority white city or a majority black city. As
hypothesis 1 predicted, blacks are more negative than whites on the economy
and the police. They are also more negative about their safety. However, all
three of these effects of being black exist only in the majority white cities. For
example, in Charlotte, being black is associated with being 9 percent lower on
the police scale, 7 percent lower on the economy scale, and 4 percent lower on
the safety scale. Chicago blacks are 13 percent lower on the police scale, 7
percent lower on the economy scale, and 5 percent lower on the safety scale.
None of these effects are large in magnitude, but they exist independent of
socioeconomic status and other demographics.
Another effect of race is the greater negative evaluation of schools by
whites in both of the majority black cities. Again, the magnitude of the effect
is small. While not originally hypothesized, this may be due to the fact that the
public schools in these cities are overwhelmingly black, underfunded, abandoned by the white middle class, and socially constructed as largely inferior
by the local media (Schneider and Ingram 1993). Race has no relationship to
evaluations of the schools in the majority white cities.
Because the impact of race differs from white cities to black cities, the analysis now shifts to the racial context. In table 3 the four cities are pooled and
the contextual variable, Race × BlackCity, is added in order to test the statistical significance of the effects of race from white cities to black cities.7 The
7. This is essentially the same analysis technique used by Bobo and Gilliam (1990) when they entered
the contextual variable, Empowerment, to their regression models of individual level behavior.
42
—
—
—
—
—
—
.12***
—
Services
Police
Safety
Economy
Transportation
General Evaluations
Schools
Crime
—
.05
.05
—
—
—
.11
.06
R2
—
—
.09***
—
−.05**
—
.05*
—
Race
(Black)
Detroit
(Black City)
—
.03
.09
.05
.02
—
.02
.04
R2
—
−.09***
−.04*
−.07***
—
—
—
—
Race
(Black)
Charlotte
(White City)
—
.14
.16
.19
—
—
.04
.05
R2
R2
.10
.12
.08
.17
.02
.07
—
.07
Race
(Black)
−.09***
−.13***
−.05*
−.07***
−.06**
−.16***
—
−.13***
Chicago
(White City)
NOTE.—Cell entries are unstandardized regression coefficients for the Race variable and the adjusted R2’s for the models. Cells without entries indicate an insignificant model based on the F statistic, or an insignificant relationship between Race and the dependent variable. Race is coded as a dichotomous variable;
0 = nonblack, 1 = black. All models include controls for gender, income, education, age, and race of interviewer, but coefficients are not presented for the sake of
brevity. Full equations are presented in Supplementary Material Online.
* p < .05.
** p < .01.
*** p < .001
Race
(Black)
New Orleans
(Black City)
Race and Dimensions of Performance
Dependent Variable
Table 2.
Explaining Mayoral Approval
Table 3.
43
Interactions between Race and Racial Context
Dependent Variable
Race
Black City
Race × BlackCity
R2
Services
Police
Safety
Economy
Transportation
General Evaluations
Schools
Crime
−.06***
−.11***
−.05**
−.08***
—
—
—
−.07**
−.16***
−.13***
−.11***
−.19***
−.09***
—
−.20***
.07**
.06***
.09***
.09***
.08***
—
—
.07**
.08*
.19
.12
.10
.19
.06
—
.13
.06
NOTE.—Cell entries are unstandardized regression coefficients and the adjusted R2’s for the
models. Cells without entries indicate an insignificant model based on the F statistic, or an
insignificant coefficient. Race is coded as a dichotomous variable; 0 = nonblack, 1 = black. All
models include controls for gender, income, education, age, and race of interviewer, but coefficients are not presented for the sake of brevity. Full equations are presented in Supplementary
Material Online.
* p < .05.
** p < .01.
*** p < .001.
coefficients associated with the variable, Race, represent the impact of being
black in the white cities, and the coefficients associated with the interaction
term, Race × BlackCity, represent how a black racial context affects those
Race coefficients. Furthermore, in keeping with our rules for drawing conclusions, we are only interested in the interaction coefficients associated with
Police, Safety, Economy, and Schools, where the patterns across cities warrant inferences and are not being driven by one city.8
The interactions of interest, predicting evaluations of Police, Safety, Economy, and Schools, are all significant, and the substantive interpretation is that,
with the exception of Schools, the negative effect of being black in the white
cities (the Race coefficient) is virtually eliminated in the black cities (the Race
coefficient plus the Race × BlackCity coefficient). For example, the impact of
being black on Police evaluations in the white cities is −.11, but in the black
cities it is −.02 (−.11 + .09). In the case of Schools, the effect of being black in
the white cities is insignificant, but in the black cities, the effect of being black
on evaluations of public schools is positive. Thus, it seems that race affects the
performance model by having a modest influence on some of the performance
evaluations themselves, and this influence depends on racial context.
The final conclusion from tables 2 and 3 is the low explanatory power of
most of the models, suggesting that citizens’ evaluations of these performance
factors are a result of their observations and experiences with actual services
8. For example, the interaction coefficients associated with Services (−.06) and Crime (−.07) are
being driven solely by Chicago, the only city where being black means lower evaluations of Services and Crime. Therefore, we will not draw conclusions about any of the pooled coefficients
associated with Services or Crime.
44
Howell and Perry
and conditions, information obtained from the media, friends, and family, or
other unknown factors.
RACIAL ATTITUDES
Whites’ and blacks’ racial attitudes are also expected to affect performance
evaluations differently in white versus black contexts (hypotheses 2 and 3).
With the exception of evaluations of Police, this expectation is not borne out.
In an analysis not shown, the four city models are first estimated separately.
Whites’ racial attitudes have effects on some performance dimensions in addition to Police, but the effects are so sporadic that we cannot draw a conclusion
about racial context. Furthermore, the effects of black racial attitudes, in addition to being unrelated to the racial composition of the city, are extremely
small.
Police evaluations are a special case where white racial attitudes are clearly
influential (again, data not shown here). In the two black cities, whites who
perceive higher levels of Group Conflict have more negative opinions about
the police than whites with lower levels of Group Conflict. In the white cities,
the effects of conservative white racial attitudes on police evaluations are
either insignificant or positive. Based on these four cities, we suggest that living in a majority black city might incline whites to view the police, who are
also majority black, partially through the lens of their racial attitudes.
Interesting results emerge when racial context is added to both black and
white models of all performance measures (tables 4 and 5). The first conclusion is that living in one of the black contexts has a negative effect on most of
the eight performance dimensions among both blacks and whites, regardless
of racial attitudes. With the exception of Crime, all of the coefficients associated with Black City are negative or insignificant. Table 5 shows the negative
impact that living in the black cities has on white residents’ evaluations of
conditions in their city. In these black cities, whites are 16 percent of the scale
lower in their evaluations of municipal services than they are in the white cities, 12 percent lower in their evaluations of the police, 10 percent lower in
perceptions of safety, and so on.9 Furthermore, in the four performance areas
singled out from table 2, Police, Safety, Economy, and Schools, the negative
impact of living in the black cities is greater for whites than for blacks, partially confirming hypothesis 4.10
9. While it appears in table 5 that whites with conservative racial attitudes have more negative
evaluations of most of the performance measures, these pooled coefficients are being driven by
relationships in one city or in one white and one black city. In accordance with the standards we
set earlier for drawing conclusions, this is not sufficient to draw a general conclusion.
10. The statistical significance of these racial differences lies back in table 3. The interaction
term, Race*BlackCity, in table 3 is simply another way of interpreting the effects of Black City
on the performance evaluations of blacks versus whites in tables 4 and 5. Six coefficients from
table 3 are statistically significant, but only four, Police, Safety, Economy, and Schools, are
supported by evidence from three of the four cities (see table 2).
Explaining Mayoral Approval
45
Table 4. Racial Attitudes, Racial Context, and Dimensions of Performance
for Blacks
Dependent Variable
Black Solidarity
Linked Fate
Black City
R2
Services
Police
Safety
Economy
Transportation
General Evaluations
(Eq. not sig.)
Schools
Crime
—
—
—
—
—
—
−.04**
−.05***
−.05**
—
—
—
−.11***
−.04** a
—a
−.11*** a
−.09***
—
.12
.05
.07
.10
.06
—
—
—
—
—
−.13*** a
.16***
.08
.07
NOTE.—Cell entries are unstandardized regression coefficients for the racial attitude variables
and the adjusted R2’s for the models. All models include controls for gender, income, education,
age, ideology and race of interviewer. Full equations are presented in Supplementary Material
online.
a
These are the four performance measures where we can say, based on statistical significance
and our criteria for confirmation of a hypothesis, that the effects of Black City are greater for
whites than for blacks (see footnote 10).
* p < .05.
** p < .01.
*** p < .001.
Table 5.
for Whites
Racial Attitudes, Racial Context, and Dimensions of Performance
Dependent Variable
Symbolic Racism
Group Conflict
Black City
R2
Services
Police
Safety
Economy
Transportation
General Evaluations
Schools
Crime
—
—
−.08*
—
−.07*
−.16**
—
−.25***
−.05*
−.15***
—
−.17***
−.06*
−.10*
—
−.11*
−.16s***
−.12*** a
−.10*** a
−.18*** a
−.08***
—
−.20*** a
.08**
.26
.18
.14
.28
.07
.03
.15
.06
NOTE.—Cell entries are unstandardized regression coefficients for the racial attitude variables
and the adjusted R2’s for the models. All models include controls for gender, income, education,
age, ideology and race of interviewer. Full equations are presented in Supplementary Material
online.
a
These are the four performance measures where we can say, based on statistical significance
and our criteria for confirmation of a hypothesis, that the effects of Black City are greater for
whites than for blacks (see footnote 10).
* p < .05.
** p < .01.
*** p < .001.
46
Howell and Perry
Even more interesting are the negative effects of living in the black cities on
blacks themselves. Previous research has demonstrated the gains in sociopolitical attitudes and political participation that accrue to African Americans from
living in cities with black mayors, with blacks being more trusting of
government, more efficacious, and more knowledgeable about politics, and as
a result, more politically active (Bobo and Gilliam, 1990). These gains in civic
attitudes and participation do not extend to evaluations of concrete aspects of
the quality of life.11 Perhaps this is to be expected, given the myriad of
problems facing majority-minority cities.
In sum, perceptions of performance are inextricably tied to racial factors in
racially diverse settings. Being black negatively affects perceptions of the quality of life in the majority white cities, and living in the black majority cities has
a negative effect on both blacks’ and whites’ perceptions of the quality of life.
These are the underpinnings of a performance model of mayoral approval.
The Performance Model of Mayoral Approval
The final series of models directly tests the performance model by estimating
the effects of the performance evaluations, race, and the control variables on
mayoral approval (table 6). Before automatically including all of the performance dimensions in the model, we must consider the problem of reciprocal
causation between the performance measures and mayoral approval. That is,
people who approve of the mayor may respond more positively when asked
about conditions in the city. This is a particularly sticky problem in cross-sectional studies where there is no hard evidence as to which attitude comes first.
However, some performance measures are arguably more prone to this endogeneity problem than others. Specifically, we contend that reciprocal causation is
more of a problem for general performance measures than for specific ones.
That is, whether one likes or dislikes the mayor is much more likely to affect
how respondents answer general questions about how things are going in the
city, than it is to affect specific questions about the streets, the crime trend,
opportunities for employment, excessive use of force by police, and most of the
items comprising the performance dimensions. The specific items are less
susceptible to psychological adjustments. This contention is supported by
MacKuen, Erikson, and Stimson (1989) who studied presidential approval
using the University of Michigan Index of Consumer Sentiment, in which
three of the five items are specific, and concluded that the direction of causality
went from the consumer attitudes to presidential approval. In our study the
General Evaluations dimension of performance is most subject to reciprocal
11. The one exception to greater black negativism is African Americans being more likely to
think that crime is decreasing in the black cities as compared to the white cities. This could be due
to specific crime reduction programs introduced in Detroit and New Orleans that targeted housing
projects and other low-income areas.
47
2.54***
.92*
−.34
.87 *
.16
.49
.33*
.30***
445
.07***
.24***
2.0***
1.00**
.14
.65
.31
.41
.32*
.25***
428
.09***
.22***
.24***
2.01***
1.14**
−.09
.22
.46
.58**
−.08
.26***
421
.07***
−.40**
Charlotte
(White City)
.31***
1.08*
.94**
.41
1.32***
.60
.48*
.34*
.32***
443
.07***
−.31**
Chicago
(White City)
NOTE.—Cell entries are unstandardized regression coefficients for the performance dimensions, which are coded 0 to 1. All models include controls for gender,
income, education, age, party identification, and race of interviewer. Full equations are presented in Supplementary Material online. OLS = Ordinary least squares.
* p < .05.
** p < .01.
*** p < .001.
−.61***
Detroit
(Black City)
.63***
New Orleans
(Black City)
Race, Performance, and Mayoral Approval
OLS regression models
Race (black)
Performance:
Services
Police
Safety
Economy
Transportation
Schools
Crime
Adj. R2
N
Explanatory power excluding
performance measures
Explanatory power
excluding race
Table 6.
48
Howell and Perry
effects from mayoral approval because it asks broad questions about the direction of the city. Thus, taking a conservative approach, we will exclude General
Evaluations from the final model of mayoral approval.12
The first conclusion to be drawn is that performance matters in evaluating
both the black mayors and the white mayors, confirming hypothesis 5. Two
performance dimensions influence mayoral approval in all four cities, Services and Police; and perceptions of crime influences approval in three of the
four cities. Perceptions of government services have the largest impact in all
cities. Recall that this scale is composed of nine items where citizens rate
services ranging from parks to pollution control. Apparently, delivery of these
city services is an important determinant of mayoral approval. For example,
the interpretation of the Detroit coefficient is that movement from being negative on all of the municipal services (score = 0) to being positive on all of the
municipal services (score = 1) would result in a change of 2.54 points on the
5-point mayoral approval scale.
The influence of the Police and Crime dimensions on mayoral approval
indicates the importance of crime control in these urban areas. Evaluations of
police have more impact on mayoral approval than perceptions of whether
crime is increasing or decreasing. Perhaps this is due to the visibility of police
and to the high level of responsibility that urban citizens give to their police.
Alternatively, it could be due to less measurement error in the Police scale,
which consists of five indicators, compared to the single indicator used to
measure crime.
With multiple measures of performance it is difficult to grasp the influence
of performance as a whole. One method is simply to manipulate coefficients
associated with the significant performance measures and observe the predicted effect on mayoral approval. Taking Charlotte as an example, movement
from the lowest to the highest score on Police, Services, and Schools would
produce a 3.73 point (2.01 + 1.14 + .58 = 3.73) movement on the 5-point mayoral approval scale. However, this scenario overestimates the impact of performance because it is unlikely that pairs of people exist who are uniformly
negative versus uniformly positive on all three performance measures.
A more conservative approach to estimating performance effects would be
to manipulate certain performance items by arbitrary amounts and add the
12. Another approach would be to create an instrumented variable of General Evaluations that
would be purged of the reciprocal effects of mayoral approval and use two-stage least squares.
However, creating an instrument requires finding variables in the data that affect General Evaluations without directly affecting mayoral approval. We tried using various combinations of demographics, but the explanatory power was so weak that an adequate instrumented General
Evaluations could not be constructed. We also rejected using the seven specific dimensions to create an instrumented General Evaluations, because the specifics are hypothesized to directly affect
mayoral approval, and, again, the explanatory power was low, .17. There is probably a component
of General Evaluations that is not affected by mayoral approval, as the General Evaluations questions came before the mayoral approval question in the interviews. However, because General
Evaluations cannot be purged of reciprocal effects in these cross-sectional surveys, it was decided
to exclude General Evaluations from the final model.
Explaining Mayoral Approval
49
effects. For example, continuing with the Charlotte example, movement of
one-fourth of the range of the Services, Police, and Schools scales (e.g., from
0 to .25 or from .50 to .75) would yield a .93 point shift in the 5-point mayoral
approval scale, holding other variables constant. This result is still a notable
shift in approval.
A second method of assessing the performance model is to examine its
explanatory power. We use the method recommended by Gujarati (1995, pp.
456–57), which involves dropping sets of explanatory variables and observing
the effects on explained variance. Beginning with the full model, the explanatory power ranges from .25 to .32. However, when all of the performance
measures are excluded, the explanatory power drops to between .07 and .09.
Obviously, performance accounts for much of the variance that is explained in
these models of mayoral approval.
It is important to note that, regardless of the method used, performance
effects are not clearly greater for the black mayors or the white mayors, nor is
any performance measure more influential in black or white cities.13 The coefficients of determination and the results of the explanatory power analysis are
remarkably similar across these four cities. But the importance of specific performance dimensions varies from city to city in ways unrelated to the race of
the mayor and city. The impact of the specific aspects of performance probably
varies over time and across locales depending on circumstances and events.
How does race add to our understanding of the performance model in these
racially diverse cities? Race has considerable impact on mayoral approval in
New Orleans and Detroit, and a moderate effect in Charlotte and Chicago, even
with controls for the seven performance measures, party identification, and
other control variables. Consistent with hypothesis 6, race has more impact on
approval of the black mayors than on approval of the white mayors. To confirm this finding, the data were pooled and an interaction between race and
black city was added to the model. The interaction coefficient was positive and
significant at p < .01. The conditioning effect of racial context was expected
because a black mayor and majority black context are the exceptional situations
for most Americans, which draw attention to race and elevate its importance.
Turning to the explanatory power of race, it is clear that dropping race from
the models reduces explanatory power, and more so for the black mayors than
for the white mayors. In New Orleans and Detroit, explanatory power is reduced
by 3 and 6 percent, whereas in Charlotte and Chicago the reduction is only 2 and
1 percent. Thus, race has both more impact and more explanatory power when
evaluating the black mayors than when evaluating the white mayors.
The main conclusion to be drawn from table 6 is that performance matters
in evaluating both the black and white mayors. In spite of the racial polarization in mayoral approval, many citizens are observing conditions in their city
13. The coefficients for Schools are significant only in the white cities, but pooling the data and
adding an interaction term, Schools*BlackCity, indicates that the difference between black and
white cities is insignificant.
50
Howell and Perry
and giving the mayor credit or blame. Successful performance can and does
overcome race as a predictor of approval. All four of these mayors had at least
57 percent approval from both races. Thus, based on the evidence from these
four cities, we tentatively conclude that the performance model, which is commonly used to explain presidential and gubernatorial approval, can be generalized to mayors.
Hypotheses 7 and 8 predict that racial attitudes should be added to mayoral
approval models. However, only two of sixteen possible coefficients associated with racial attitudes were even statistically significant. We conclude,
somewhat to our surprise, that racial attitudes, as measured by symbolic racism, group conflict, black solidarity, and linked fate, do not add to our understanding of approval of mayors in these racially diverse settings.14
Conclusion and Discussion
Performance models have been the norm in research on national and state executives, but this model has not been tested systematically beyond one city at the
local level. When Niemi and his colleagues argued for a test of state economics
on gubernatorial voting (Niemi, Stanley, and Vogel 1995, p. 938), they cited
the visibility of a single governor, the election by citizens statewide, the fact of
four-year terms, and the practice of holding state elections in different years
than the presidential election. All of these factors apply to most mayors as well.
However, the performance model, whether based on economics or something
else, faces a complication at the local level, and that is the presence of black
executives. The vast literature documenting racial polarization in the elections
of black mayors strongly suggests that racial factors should be added to this
model when examining the approval of mayors in racially diverse settings.
Our test of the performance model on two black mayors and two white
mayors indicates that performance does indeed affect mayoral approval, and
both its impact and explanatory power are greater than that of race for both the
black and the white mayors. Performance measures add between 16 and 25
percent to the explanatory power of the models of mayoral approval beyond
race, racial attitudes, and important demographic variables. In contrast, race
adds only 1 to 6 percent to explanatory power. Furthermore, the overall effect
of performance is not related to the race of the mayor. All of these are reassuring findings given the bivariate racial polarization in mayoral approval and
14. Why is race itself influential while racial attitudes are so weak? In the case of blacks, one
possibility is that limitations in the attitude scales result in a failure to capture the full richness of
black identity or black consciousness. The concept, Linked Fate, is measured by only one indicator, and Black Solidarity taps support for collective action, only one form of black consciousness.
In the case of whites, the racial attitudes, Symbolic Racism and Group Conflict, do not purport to
be measures of white racial identity; they are best described as measures of antiblack feelings and
competition with blacks. So, for both blacks and whites, it is likely that, in this research, much of
the concept, racial identity, is still captured by the simple demographic, race.
Explaining Mayoral Approval
51
suggest that we are making progress toward a “normalization” of black executives in some locations (see also Hajnal 2001).
One of the unique aspects of this study is the comprehensive measure of
performance. Conceptually, the local level is more than just the third level
below the national and state levels. The meaning of “performance” is qualitatively different at the local level, in that citizens can evaluate many aspects of
performance firsthand. In contrast, most information about state, and certainly
national, conditions is obtained from the media. This distinction gives “performance” at the local level a richness and complexity not possible at the two
higher levels of government.
As we would expect, racial polarization remains in all four cities, and race
has more impact on approval of the black mayors than the white mayors.
Other black mayors would certainly not be surprised to hear this. The mayors
in our two majority black cities must also face a citizenry that is more dissatisfied with a variety of municipal services than citizens in the white cities. This
greater dissatisfaction is not limited to white citizens, but exists among their
black constituents as well.
White mayors in the majority white cities, on the other hand, face an electorate that is more racially polarized with regard to performance. That is, the
gap between black and white evaluations of a variety of aspects of local government performance is greater than in the black cities. Race has direct effects
on the approval of the white mayors, as well as indirect effects through the
performance evaluations.
It should be noted that approval does not automatically translate to support at
the polls. Given the history of racial polarization in elections, it is safe to say
that race influences vote more than it influences approval. Approval is an “easy”
response to a survey question, and it is usually given without the charged atmosphere of an election campaign. In contrast, voting is behavior, which is more
likely to activate basic predispositions and identities. Nonetheless, between
elections, approval ratings have political impact for mayors as well as for presidents and thus deserve explanation from public opinion researchers.
The performance model should be examined in a greater variety of urban settings. Future research should examine majority white cities with black mayors,
majority black cities with white mayors, and cities with Latino mayors. Such
research will more firmly settle the question of the generalizability of the performance model to the local level and further clarify the role of race and ethnicity.
Appendix
THE SURVEYS
The Mayoral Approval Surveys were conducted between May and August 2000 in four
cities, New Orleans, Detroit, Chicago, and Charlotte, NC by the University of New
52
Howell and Perry
Orleans Survey Research Center. This project was funded by the National Science
Foundation (Award #9986165). All surveys included only registered voters. Nearly
equal numbers of blacks and whites were interviewed in each city in order to maximize
variance on race, a key independent variable. Thus, the samples do not represent cross
sections of each city’s registered voters, but they do represent a cross section of whites
or blacks in each city. In order to obtain the desired number of respondents from the
minority race, first the cross-sectional sample was exhausted. Then, areas of 40 percent
or more minority race concentration were used to obtain the needed number. This was
necessary due to the practical difficulties and expense of finding respondents from the
minority race (whether white or black) in a cross-sectional sample. For example,
Detroit is only 20 percent white, so a cross-sectional sample yielding 250 blacks would
only include 112 whites. The more targeted sampling allowed us to obtain the required
number of whites without extraordinary costs. However, this method does oversample
minorities (white or black) who live among 40 percent or more minorities. Samples,
drawn by Survey Sampling of Fairfield, CT, used a random digit dialing procedure, to
which the survey center applied a screen for registered voters.
Response Rate #3 in AAPOR’s Standard Definitions: Final Disposition of Case
Codes and Outcome Rates and RDD Telephone Surveys and In-Person Household Surveys was utilized, and the response rates in the four cities were as follows: New
Orleans—63 percent, Chicago—55 percent, Charlotte—58 percent, and Detroit—53
percent. These response rates are respectable considering that half of the sample
is urban minorities, a largely low-income group who are more difficult to include in
telephone surveys.
Interviewers for all four surveys were university students. The interviews were conducted from a central phone bank with constant supervision by a graduate student. The
survey center used a CATI system to manage the sample, and as many as fifteen
attempts were made to reach respondents. The final number of blacks and whites interviewed in each city is presented in Table 1.
QUESTION WORDING
Mayoral Approval: In general, do you approve or disapprove of the job Mayor XX is
doing? (PAUSE) Is that strongly or not very strongly?
Services: I would like to ask you about government and government services in the
city. Is each of the following aspects of government very good, good, fair, poor, or
very poor? Overall level of government services? Drainage and flood control? Control
of abandoned housing? Health services? Services for the poor? Quality of housing?
Control of litter and trash? Parks and recreation? Control of pollution?
Police: I would like to ask you about government and government services in the
city. Is each of the following aspects of government very good, good, fair, poor, or
very poor? Police protection. How would you rate the [city] police on the following?
The ability of the [city] police to respond quickly to calls for help and assistance?
The ability of the police to find the criminals after a crime has been committed?
Being courteous, helpful, and friendly? Avoiding the use of unnecessary or excessive
force?
Safety: In general, how safe do you feel from crime walking in your neighborhood
during the day? (very safe, safe, not very safe, not at all safe). During the night?
Explaining Mayoral Approval
53
Walking downtown at night? Turning to crime and safety, would you say that the
amount of crime in [city] has increased, decreased, or remained about the same over
the last several years?
Economy: Now I’d like to ask you some questions about the economy in [city]. Is
each of the following very good, good, fair, poor, or very poor? Opportunities for
employment? Likelihood of new jobs and industry coming into the city? Likelihood of
your family increasing its income in the next several years?
Transportation: Now I would like to ask you about transportation in [city]. Are each
of the following very good, good, fair, poor, or very poor? Conditions of your local
streets and roads? Availability of public transportation within the city? Control of traffic congestion?
General Evaluations: Thinking back over the last five years, would you say that
[city] has become a better or worse place to live, or hasn’t there been any change? And
thinking ahead over the next five years, do you think that [city] will become a better or
worse place to live, or won’t there be much of a change?
Schools: Now thinking about education, how would you rate the quality of public
schools in [city], very good, good, fair, poor, or very poor?
Symbolic Racism: (1) Over the past few years blacks have gotten less than they
deserve (agree strongly to disagree strongly). (2) Irish, Italians, Jews, and other
minorities overcame prejudice and worked their way up. Blacks should do the same
without any special favors. (3) In past studies, we have asked people why they think
white people seem to get more of the good things in life in America—such as better
jobs and more money—than black people do. These are some of the reasons given by
both blacks and whites. It’s really just a matter of some people not trying hard enough.
If blacks would only try harder, they could be just as well off as whites. (4) Generations of slavery and discrimination have created conditions that make it difficult for
blacks to work their way out of the lower class.
Group Conflict: (1) More good jobs for blacks means fewer good jobs for members
of other groups (agree strongly to disagree strongly). (2) The more influence blacks
have in local politics the less influence members of other groups will have in politics.
(3) As more good housing and neighborhoods go to blacks, the fewer good houses and
neighborhoods there will be for members of other groups. (4) Many blacks have been
trying to get ahead at the expense of other groups.
Black Linked Fate: Do you think what happens generally to black people in this
country will have something to do with what happens in your life? (if Yes): Will it
affect you a lot, some, or not very much?
Black Solidarity: How important is it for blacks to vote for black candidates when
they run for office—extremely important, very important, or only somewhat important? (not very important: volunteered). To participate in black-only organizations
whenever possible? To have control over the government in mostly black communities? To have control over the economy in mostly black communities?
References
Adler, Jeffrey S. 2001. “Introduction.” In African American Mayors: Race, Politics, and the
American City, ed. D. R. Colburn and Jeffrey S. Adler, pp. 1–22. Urbana: University of
Illinois Press.
54
Howell and Perry
Barnes, Claude W., Jr. 1994. “Atlanta’s Post–Civil Rights Political Economy: Demobilization
and Economic Dependency.” Paper presented at the annual meeting of the Southern Political
Science Association, Atlanta.
Bobo, Lawrence D. 1998. “Race, Interests, and Beliefs about Affirmative Action: Unanswered
Questions and New Directions. American Behavioral Scientist 41 (April): 985–1003.
———. 2000. “Race and Beliefs About Affirmation Action: Assessing the Effects of Interests,
Group Threat, Ideology, and Racism.” In Racialized Politics: The Debate about Racism in
America, ed. David O. Sears, James Sidanius, and Lawrence D. Bobo, pp. 137–64. Chicago:
University of Chicago Press.
Bobo, Lawrence D., and Franklin D. Gilliam, Jr. 1990. “Race, Sociopolitical Participation, and
Black Empowerment.” American Political Science Review 84:377–93.
Bobo, Lawrence D., and Vincent L. Hutchings. 1996. “Perceptions of Racial Group Competition:
Extending Blumer’s Theory of Group Position to a Multiracial Social Context.” American
Sociological Review 61:951–72.
Brody, Richard A. 1991. Assessing the President: The Media, Elite Opinion, and Public Support.
Stanford CA: Stanford University Press.
Browning, Rufus P., Dale Rogers Marshall, and David H. Tabb. 1984. Protest Is Not Enough.
Berkeley: University of California Press.
Browning, Rufus P., and Dale Rogers Marshall. 1986. “Black and Hispanic Power in City Politics: A Forum.” PS: Political Science 19:573–640.
Campbell, A., and H. Schuman. 1968. Racial Attitudes in Fifteen American Cities. Ann Arbor,
MI: Survey Research Center, Institute for Social Research.
Carsey, Thomas M. 1995. “The Contextual Effects of Race on White Voter Behavior: The 1989
New York City Mayoral Election.” Journal of Politics 57:221–28.
Dawson, Michael C. 1994. Behind the Mule: Race and Class in African-American Politics. Princeton,
NJ: Princeton University Press.
Decker, Scott. 1981. “Citizen Attitudes toward the Police: A Review of Past Findings and Suggestions for Future Policy.” Journal of Police Science and Administration 9(1):80–87.
Edwards, George C., III, William Mitchell, and Reed Welch. 1995. “Explaining Presidential
Approval: The Significance of Issue Salience.” American Journal of Political Science
39:108–34.
Engstrom, Richard L., and Willie Kirkland. 1995. “The 1994 New Orleans Mayoral Election:
Racial Divisions Continue.” Urban News 9 (Spring): 6–9.
Farley, Reynolds, Sheldon Danziger, and Harry J. Holzer. 2000. Detroit Divided. New York: Russell Sage Foundation.
Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New Haven, CT:
Yale University Press.
Giles, Michael W., and Kaenan Hertz. 1994. “Racial Threat and Partisan Identification.” American Political Science Review 88:317–26.
Glaser, James M. 1994. “Back to the Black Belt: Racial Environment and White Racial Attitudes
in the South.” Journal of Politics 56:21–41.
Gujarati, Damodar N. 1995. Basic Econometrics. New York: McGraw-Hill.
Gurin, Patricia, Shirley Hatchett, and James S. Jackson. 1989. Hope and Independence: Blacks’
Response to Electoral and Party Politics. New York: Russell Sage Foundation.
Hajnal, Zoltan. 2001. “White Residents, Black Incumbents, and a Declining Racial Divide.”
American Political Science Review 96:603–17.
Hetherington, Marc J., and Michael Nelson. 2003. “Anatomy of a Rally Effect: George W. Bush
and the War on Terrorism.” PS: Political Science and Politics 36:37–50.
Hibbs, Douglas A. 1982. “The Dynamics of Political Support for American Presidents Among
Occupational and Partisan Groups.” American Journal of Political Science 26:312–32.
Howell, Susan E., and Deborah Fagan. 1988. “Race and Trust in Government: Testing the Political Reality Model.” Public Opinion Quarterly 52:343–50.
Howell, Susan E., and William McLean. 2001. “Performance Models and Minority Mayors.”
Public Opinion Quarterly 65:321–43.
Howell, Susan. E., and James M. Vanderleeuw. 1990. “Economic Effects on State Governors.”
American Politics Quarterly 18:158–68.
Huckfeldt, Robert, and Carol Weitzel Kohfeld. 1989. Race and the Decline of Class in American
Politics. Urbana: University of Illinois Press.
Explaining Mayoral Approval
55
Kernell, Samuel. 1978. “Ethics: Presidential Popularity.” American Political Science Review
72:506–22.
Kinder, Donald R., and Lynn M. Sanders. 1996. Divided By Color: Racial Politics and Democratic Ideals. Chicago: University of Chicago Press.
Kinder, Donald R., and David O. Sears. 1981. “Prejudice and Politics: Symbolic Racism Versus
Racial Threats to the Good Life.” Journal of Personality and Social Psychology 40:414–31.
King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific
Inference in Qualitative Research. Princeton, NJ: Princeton University Press.
Kleppner, Paul. 1985. Chicago Divided: The Making of a Black Mayor. DeKalb: Northern Illinois
University Press.
Kraus, Neil, and Todd Swanstrom. 2001. “Minority Mayors and the Hollow-Prize Problem.” PS:
Political Science and Politics 34:99–105.
MacKuen, Michael B. 1983. “Political Drama, Economic Conditions, and the Dynamics of Presidential Popularity.” American Journal of Political Science 27:165–92.
MacKuen, Michael B., Robert S. Erikson, and James A. Stimson. 1989. “Macropartisanship.”
American Political Science Review 83:1125–42.
McConahay, John B., and J. C. Hough. 1976. “Symbolic Racism.” Journal of Social Issues
32:23–45.
Miller, Arthur H., and Martin P. Wattenberg. 1985. “Throwing the Rascals Out: Policy and Performance Evaluations of Presidential Candidates, 1952–80.” American Political Science
Review 79:359–72.
Moesher, John, and C. Silver. 1994. “Race, Social Stratification, and Politics.” Virginia Magazine
of History and Biography 102 (October):519–50.
Nadeau, Richard, Richard G. Niemi, David P. Fan, and Timothy Amato. 1999. “Elite Economic
Forecasts, Economic News, Mass Economic Judgments, and Presidential Aproval.” Journal of
Politics 61:109–35.
Nelson, William. 1990. “Black Mayoral Leadership.” National Political Science Review 2:188–95.
Niemi, Richard G., Harold W. Stanley, and Ronald J. Vogel. 1995. “State Economies and State
Taxes: Do Voters Hold Governors Accountable?” American Journal of Political Science
39:936–57.
Ostrom, Charles W., Jr., and Dennis M. Simon. 1985. “Promise and Performance: A Dynamic
Model of Presidential Popularity.” American Political Science Review 79:334–58.
Perry, Huey L. 1990. “The Reelection of Sidney Barthelemy as Mayor of New Orleans.” PS:
Political Science and Politics 23:156–57.
———. 2000. “Mayor Richard Arrington, Jr., and Police-Community Relations in Birmingham,
Alabama.” In Governing Middle-Sized Cities: Studies in Mayoral Leadership, ed. James R.
Bowers and Wilbur C. Rich, pp. 103–18. Boulder, CO: Lynne Rienner Publishers.
Preston, B. Michael. 1990. “Symposium—Big City Black Mayors: Have They Made a Difference?” National Political Science Review 2:129–95.
Schneider, Anne, and Helen Ingram. 1993. “Social Construction of Target Populations: Implications for Politics and Policy.” American Political Science Review 87:334–47.
Sears, David O. 1988. “Symbolic Racism.” In Eliminating Racism: Profiles in Controversy, ed.
P. A. Katz and D. A. Taylor, pp. 53–84. New York: Plenum.
Sears, David O., and Donald R. Kinder. 1971. “Racial Tensions and Voting in Los Angeles.” In
Los Angeles: Viability and Prospects for Metropolitan Leadership, ed. Werner Z. Hirsch, pp.
51–88. New York: Praeger.
Sears, David O, Colette Van Laar, Mary Carrillo, and Rick Kosterman. 1997. “Is It Really Racism? The Origins of White Americans’ Opposition to Race-Targeted Policies.” Public Opinion
Quarterly 61:16–53.
Shingles, Richard. 1981. “Black Consciousness and Political Participation: The Missing Link.”
American Political Science Review 75:76–91.
Sidanius, James, Pam Singh, John J. Hetts, and Chris Federico. 2000. “It’s Not Affirmative
Action, It’s the Blacks.” In Racialized Politics: The Debate About Racism in America, ed.
David Sears, James Sidanius, and Lawrence D. Bobo, pp. 191–235. Chicago: University of
Chicago Press.
Smith, Douglas, Nannette Graham, and Bonnie Adams. 1991. “Minorities and the Police: Attitudinal and Behavioral Questions.” In Race and Criminal Justice, ed. Michael Lynch and E. Britt
Patterson, pp. 22–35. New York: Harrow and Heston.
56
Howell and Perry
Sonenshein, Raphael J. 1989. “Can Black Candidates Win Statewide Elections?” Political Science Quarterly 105:219–41.
Stone, Clarence N. 1989. Regime Politics: Governing Atlanta, 1946–1988. Lawrence, KS: University of Kansas Press.
Taylor, Marylee. 1998. “The Effect of Racial Composition on Racial Attitudes of Whites.” American
Sociological Review 65:512–35.
Taylor, Marylee C. 2000. “The Significance of Racial Context.” In Racialized Politics: The
Debate about Racism in America, ed. David O. Sears, James Sidanius, and Lawrence D. Bobo,
pp. 118–36. Chicago: University of Chicago Press.
U.S. Department of Justice. 2000. Sourcebook of Criminal Justice Statistics, 1999. NCJ-183727.
Washington DC: Bureau of Justice Statistics.
Vanderleeuw, James M. 1990. “A City in Transition: The Impact of Changing Racial Composition on Voting Behavior.” Social Science Quarterly 71 (June): 326–38.
Verba, Sidney, and Norman H. Nie. 1972. Participation in America: Political Democracy and
Social Equality. New York: Harper & Row.
Voss, D. Stephen. 1996. “Beyond Racial Threat: Failure of an Old Hypothesis in the New South.”
Journal of Politics 58:1156–70.