. 3
( 7)


(Sun Reporter 1993). This conclusion was echoed in a recent study of
mayoral voting in Houston (Stein, Ulbig, and Post 2005). Using three
different surveys of voters in the city, the study found that racial con-
siderations faded over the course of Lee Brown™s tenure in that city. As
the authors note: “Racial voting appears to be more in¬‚uential in minor-
ity candidates™ ¬rst electoral bids. In successive elections, voters come to
rely more on their evaluations of the minority incumbent™s job perfor-
mance than their racial-group af¬liation” (Stein, Ulbig, and Post 2005:
177). Though the magnitude of the change should not be overstated, it
seems that white residents became less likely to base their votes on the
race of the candidate and their fear of a black takeover in the incum-
bent elections. Instead when black incumbents ran for reelection, white
residents seemed to more deliberately assess the pluses and minuses of
their candidacies. As Sharon Watson put it in her account of mayoral
bids in eight cities, “In [reelection] campaigns, while race remains a spe-
cial factor, it did not seem to overshadow the campaign, as was true
of the ¬rst elections. Race as an issue appeared neutralized somewhat”
(1984: 172).
The Transformation of the White Vote 63

The Black Incumbent™s Record and the White Vote
The analysis to this point has ignored an important aspect of black incum-
bent elections. If the information model is accurate and white voters
change their minds about black leadership largely because experience with
black incumbents disproves many of their fears, then a black incumbent™s
record in of¬ce should be an important variable shaping the white vote.26
The model predicts that black incumbents whose policies take resources
from the white community to serve the black community or who preside
over cities with faltering economies should do less well than black incum-
bents who resist pro-black policies and govern under robust local eco-
nomic conditions. To analyze the in¬‚uence of black incumbents™ records
on the white vote, I assess a range of factors related both to overall con-
ditions in each city and to the policies that each black incumbent enacted.
Given that the main fears expressed by white residents before the election
of a black mayor were a deteriorating economy, falling housing prices,
and widespread crime, I included measures of each of these three factors
in the model.27 Since residents might logically also gauge black leadership
by local government policy, I assess the impact of local government spend-
ing patterns on the white vote by including a measure of how much a city
shifted resources from developmental spending toward redistributional
functions such as social services, housing, and education during the black
mayor™s ¬rst term. Spending is obviously one of the arenas where black
mayors can affect a large number of white residents, and any emphasis
on redistributive spending is likely to be perceived by white residents as
a strong signal of a black mayor™s underlying preferences for serving the
black community.28

26 Obviously, this is not a relationship that is unique to black incumbents. Voters generally
base their decisions at least in part on the record of the incumbent (Niemi and Weisberg
1993; Krosnick 1988; Downs 1957).
27 Speci¬cally, I include change in per capita income relative to the average change in per
capita income for all metropolitan areas (Bureau of Economic Affairs 2005), change in
median housing prices (Bureau of the Census 1973“2001), and change in the violent crime
rate per 1,000 residents (Federal Bureau of Investigation 1969“2001). Despite the fact
that incumbents are often not responsible for local economic conditions, incumbents who
are lucky enough to serve during periods of relative economic gain tend to be rewarded,
while those who lead during economic downturns are often punished (MacKuen, Erikson,
and Stimson 1992; Lewis-Beck 1988).
28 The exact measure is the change in the proportion of city government spending that
goes toward redistributive policy (housing, education, and social welfare) minus change
in the proportion of spending going to developmental policy (highways, streets, trans-
portation, and airports) (Bureau of the Census 1964“2003). These categories represent
a fairly standard division of spending in the urban literature (see e.g. Peterson 1981).
Changing White Attitudes

table 2.4 Determinants of the White Vote in Black Incumbent Elections

White Support for the
Black Candidate
’0.27 (0.16)
Percent Black of City Population
’1.03 (0.36)——
Racialization of Black Candidate™s Campaign
Candidate Quality
’0.20 (0.08)——
Quality of White Opponent
0.21 (0.08)——
Democratic Party Endorsement
0.21 (0.12)—
Local Newspaper Endorsement
Local Conditions
Change in Per Capita Income 0.40 (0.45)
0.21 (0.09)——
Change in Median Housing Prices
’0.03 (0.08)
Change in Crime Rate
’0.11 (0.44)
Change in Redistributive Spending
Constant 0.01 (0.20)
Adj. R2 0.71
N 25
Note: OLS regression. Figures in parentheses are standard errors.
——— p < 0.01
—— p < 0.05
— p < 0.10

This analysis is displayed in Table 2.4, which presents the results of a
regression explaining the aggregate white vote in black incumbent elec-
tions in the same set of twenty-six cities. With a small number of cases
and eleven independent variables, the model in Table 2.4 stretches the
limits of what regression analysis can do and should therefore be read
with some caution.29 Nevertheless, the results are suggestive.
The ¬rst conclusion is that there are signs of a link between the black
incumbent™s record and the white vote. The clearest evidence of this is that
changes in the local housing market are signi¬cantly related to the white

Since residents may also be concerned about the ¬scal health of a city, in alternate tests
I include a measure of the city™s debt as a percentage of the city™s overall revenue.
29 It is worth noting, however, that the basic results in Table 2.4 remain robust to a range
of different speci¬cations. In a series of alternate tests, I reran the analysis, dropping all
nonsigni¬cant variables or including only one measure for each factor (e.g. including
only one variable for the incumbent™s record). The basic conclusions did not change.
The Transformation of the White Vote 65

vote. If, contrary to white fears, housing prices do not collapse and home-
owners do well under black leadership, white residents will tend to reward
the black incumbent. This ¬nding parallels emerging research on so-called
performance models of mayoral approval (Stein, Ulbig and Post 2005;
Howell and Perry 2004; Howell and McClean 2001). These recent studies
have shown that in a small number of cities for which there are survey data
white approval of incumbent black mayors is related to white evaluations
of local economic conditions and white perceptions of city services.30
However, as Table 2.4 also reveals, for other aspects of the incumbent™s
record, the existence of any relationship to the white vote is less clear.
None of the other factors assessing the incumbent™s record signi¬cantly
predicts the white vote. The most that can be said is that in all three cases
the relationship between the incumbent™s record and the white vote is in
the expected direction.31
Thus, another interpretation is that the relationship between a black
incumbent™s record and the white vote is not nearly as strong as some
might have expected. In only the one case “ housing prices “ is the incum-
bent™s record signi¬cantly related to the white vote, and even here the
magnitude of the effect is not large. For every one point increase in median
housing prices, there is only a one-¬fth of a point gain in white support
for the black incumbent.
Why doesn™t an incumbent™s record matter more? Part of the answer
may be related to the limitations of the empirical model. Too many vari-
ables and too few cases certainly cloud the analysis. The imprecise nature
of the measures used in the analysis may also be a contributing factor.
Whites, for example, may be more sensitive to housing prices and crime
rates in their own neighborhoods than they are to overall changes at the
city level. But a third and perhaps more critical answer here is the fact that
almost all black incumbents exceed expectations. In the majority of the

30 Given that in these surveys local conditions and services are evaluated subjectively by
each individual respondent, there is a possibility of reverse causation “ those who approve
of the mayor tend to rate local conditions and services well. More work will have to be
done to assess this connection.
31 These regression results mirror bivariate analysis. Each of these three factors was indi-
vidually correlated with the white vote in the expected direction, but in no case was the
correlation statistically signi¬cant. The same can be said for the relationship between
af¬rmative action in local government hiring practices and the white vote. In alternate
tests, I included a measure of the increase in the proportion of the public employees who
are African American during the mayor™s term in of¬ce. These tests indicate that there is
a negative but not signi¬cant relationship between af¬rmative action and the white vote
in black incumbent elections.
Changing White Attitudes

cities in the data set, per capita incomes grew compared to the national
average, and in only two cases were gains in per capita income outpaced by
more than 2 percent by gains made at the national level. Median housing
prices rarely fell. And although crime rates did rise in the average city, in
most cases they did not rise at a rate appreciably faster than in the nation
as a whole. Likewise, local government policy under black incumbents
did little to substantiate white fears. The average city did not shift any
resources from developmental projects to such redistributive programs as
welfare, health, and housing, and in only two cases was more than 4 per-
cent of the city budget transferred to redistributive functions. Finally, few
of the cities stood out in terms of af¬rmative action policies. All but one
increased black hiring under the black mayor, but only one city increased
the proportion of blacks in the public sector by more than 5 percent. The
lack of any dramatic change under black incumbents is not surprising,
as these results mirror accounts from a range of existing studies. But it
is important, because it represents a stark contrast with the expectations
and fears of many whites. In essence then, the lesson is the same in almost
every city. By maintaining tolerable or even relatively robust economic
conditions and by choosing not to shift substantial resources away from
the white community, black mayors, in almost all cases, demonstrate that
black leadership does not appreciably hurt the white community. The
bottom line is that black incumbents can help themselves by introduc-
ing policies that bene¬t the city but in the end all they have to do is not
attack the white community. That is often enough to convince some white
residents that they are worth supporting.

Individual White Residents and Learning under Black Incumbents
Although the changes in the vote outlined in Tables 2.2, 2.3, and 2.4
are important and certainly could be interpreted as showing that white
residents learn from their experiences under black incumbents, we need to
look at the sentiments expressed by individual white residents for direct
evidence of white learning. These views reinforce the basic conclusions I
have drawn from my analysis of the aggregate vote. At least among those
who were willing to speak out, there was a strong sense of relief and a
clear change in perceptions of black leadership. The comments of one
Memphis resident are illustrative: “I get a feeling he is keeping the city
from being split asunder by racial divisions. I thought there would be a
lot of pressure on him from blacks to give in to all of their demands. I
think he has been a mayor for all of the city” (A. Davis 1995). A voter
The Transformation of the White Vote 67

in Los Angeles put it this way: “A lot of people were very suspicious
and fearful before Bradley got in. But they never say anything now. I™m
sure they have changed their opinions. . . . Most important, he is a good
person. Whether he is black or white is immaterial” (US News & World
Report 1975). In Atlanta, the head of a local civic organization also noted
a change in white sentiments in that city: “A lot of people who thought
they couldn™t live with a black administration have found they can do so
quite well” (Eisinger 1980: 77). Many whites did not change their minds
about black leadership at all, and in some cities the change was more
dramatic than in others. But for the most part, the sentiments of white
voters in black incumbent elections are a far cry from the expressions of
fear and uncertainty that surround black challenger elections.
Black candidates also made note of the learning process. Thirman
Milner™s comments are typical: “I ran up against racial issues in my ¬rst
election with all the talk that crime would go up, the city would go down
and I would only address black issues. But people began to see there are no
racial distinctions when it comes to operating as a mayor” (Hagstrom and
Guskind 1983). His views were echoed by Carl Stokes, Cleveland™s ¬rst
black mayor: “There are so many prejudices that impact upon voters. But
once they™ve seen an individual in action and what they™ve been able to
do, they discard some of those biases and fears and look at that candidate
the same way they look at any other public of¬cial” (Vickers 1997: 23).
Another sign of learning that cannot be seen in the regressions in
Tables 2.2, 2.3, and 2.4 is a dramatic shift in tone between the typical
black challenger election and the typical black incumbent election. The
black challenger elections tended to be intense, racially polarized affairs
that were characterized by the media with phrases like “black versus
white” (in the Flint election, 1983), “race-dominated” (in Chicago, 1983),
and “highly polarized” (in Memphis, 1991). They often included highly
provocative campaigns that played on white uncertainty and racial fears.
But after a period of only a few years, these bitter, polarized elections
often gave way to more ordinary affairs. In sharp contrast to black chal-
lenger elections, electoral contests with black incumbents were generally
described with phrases that emphasized their lack of racial animosity:
“low-key, almost digni¬ed politicking” (in the Atlantic City election,
1986), “a colorless campaign” (in Atlanta, 1977), “ho-hum voters” (in
Memphis, 1995), “a lusterless campaign” (in Cleveland, 1969), and “a
humdrum affair” (in Denver, 1999). In Newark, for example, a black chal-
lenger election that had included numerous incidents of racial heckling,
bomb threats, and even violence was followed four years later by a black
Changing White Attitudes

incumbent election the New York Times characterized as “surprisingly
uneventful” (May 12, 1974). Similarly, in Birmingham, four years after
Richard Arrington™s election bid “brought racial animosities in the city to
their highest pitch since the civil rights demonstrations of 1963” (Raines
1979), his reelection run ended “with a biracial coalition and the largest
victory margin in city history” (Russakof 1983). Incumbent elections in
some cities, including Chicago, New York, and New Haven, did remain
racialized and as such did not ¬t this pattern, but in the vast majority of
cases there was a marked change in the nature of the election campaign.

Do White Residents Simply Give Up?
One possible confounding factor in the analysis to this point is “white
¬‚ight.” If a signi¬cant number of whites perceived black victory as cer-
tain and effectively gave up, choosing not to vote or to leave their city of
residence altogether, the changes that we see in the white vote might be
more apparent than real. To see if changes in the white vote are at least in
part a function of white ¬‚ight or selective nonvoting by whites, in alter-
nate analysis I added two measures to the regression model in Table 2.4:
voter turnout in each election and change in the city™s white population
between the challenger and the incumbent election. The regression results
indicate that the white vote was not related to declining turnout or to
white ¬‚ight.32 Other available evidence supports this conclusion. First, in
cities where public opinion polls are available, the data show that white
voters were not substantially more likely than non-voters to express sup-
port for black incumbents.33 This strongly suggests that anti-black voters

32 The model assesses the relationship between overall turnout and white support for the
black candidate, but it seems likely that if turnout numbers for whites were available for
all of these cities, white turnout would also prove to be unrelated to white support for
the black candidate. White turnout is correlated with overall turnout at 0.97 for cities
in which turnout by race is available. There is little evidence, moreover, of a relationship
between the vulnerability of the black incumbent and turnout. White residents are no
more apt to fail to vote in cities where blacks are the majority and could presumably
determine the outcome of the election than they are elsewhere. White turnout actually
drops less in majority black cities (5.5 point drop) than it does in minority black cities
(8.7 point drop). Moreover, even if the ¬rst election was close and whites presumably
had a real chance of reversing the outcome, white turnout dropped at about the same
rate in these cities as it did elsewhere.
33 Similarly, in cities where data are available by precinct or district, turnout falls at about
the same rate across neighborhoods. White neighborhoods that were more opposed to
the black challenger generally lost voters at about the same rate as white neighborhoods
that had been more supportive of the black challenger (Sonenshein 1993).
The Transformation of the White Vote 69

were not selectively dropping out. Second, not that many white residents
actually left their cities in the short period between the challenger and
incumbent elections. The white proportion of the population declined by
2 percent on average across the cities “ not enough to account for the
changes in the white vote that are seen in Tables 2.2, 2.3, and 2.4 (Bureau
of the Census 1994, 1990, 1978).34 And those whites who did leave the
cities were probably not especially anti-black in their thinking. Existing
research indicates that whites who leave cities tend to be younger, wealth-
ier, and better educated (Deane 1990). The poor, older, and less educated
who are left behind because the costs of moving are too high are the
people who are most likely to be racially intolerant (Bobo 1983; Rieder
1985). Thus, if anything, white out-migration should have led to a more
rather than a less racialized white vote.35
Changes in the vote, changes in the tone of campaigns, and changes
in the sentiments expressed by many white residents from challenger to
incumbent elections all point to the conclusion that the information pro-
vided by black representation erodes the fears of many white residents
and makes them more willing to consider supporting black leadership.

changes over time
If black leadership is to represent a real turning point in American race
relations, the effects that appear in the data set of twenty-six black mayors
must extend to other candidates in other elections. Experience with one
black incumbent, if it merely convinces white residents that a particular
black incumbent is trustworthy, will only have a temporary and isolated
effect. If, however, learning is more extensive and experience under black
incumbents leads whites to fundamentally reevaluate the broader threat
posed by black leadership and the degree to which black interests and
white interests are in con¬‚ict, these elections could signal a new era in
racial politics. Part of the answer to this question appears in Chapter 3 in
a more detailed examination of white views under black representation,
but a cursory look at the white vote for evidence of a gradual increase in

34 This ¬ts with other analysis, which indicates that the rapid out-migration of whites was
largely stemmed by the time most of these cities had elected their ¬rst black mayors
(Brown 1997).
35 It is also worth noting that while race is often a factor in motivating moves from neigh-
borhood to neighborhood, studies of inter-metropolitan migration suggest that racial
motivations seldom explain moves into and out of larger cities (Koven and Shelley 1989;
Long 1988; McHugh 1985; South and Deanne 1993; Stahura 1988).
Changing White Attitudes

table 2.5 Changes in Voting Behavior Over Time: Black
Challenger Elections

Average White Support
Decade (N) for Black Challenger (%) Average Turnout (%)
1960s (2) 17 76
1970s (5) 21 67
1980s (13) 33 54
1990s (6) 34 41

support for black candidates over time can also tell us a little about how
wide-ranging the effects of black leadership are.
In Table 2.5, I begin to test this proposition by examining white support
for black challengers by decade in the twenty-six elections.36 The analysis
reported in the table suggests that white support for black candidates is in
fact on the rise. Support for black candidates who run as challengers has
increased markedly over time. The number of cases in each decade may be
small, but the change is clear. Black challengers in the 1960s faced almost
unanimous white opposition (83 percent of white votes). In the 1970s,
white support for black challengers was, on average, 4 percent higher.
In the 1980s, it grew by another 12 percent. By the 1990s, the average
black challenger received the support of 34 percent of all white voters “
certainly not an indication that race is now irrelevant, but nevertheless a
big step from the fear and anti-black mobilization of the late 1960s.
Two other changes point to a similar transformation of the white vote
over the same period. As the second column of Table 2.5 illustrates,
turnout dramatically declined in black challenger elections over time.
Overall, turnout dropped from an average of 76 percent in the 1960s
to 41 percent in the 1990s. For the cities in which the data on turnout
can be broken down by race, it appears that white turnout declined at
roughly the same rate as overall turnout. It may well be that the election
of a black candidate to of¬ce does not create nearly as much uncertainty
and fear in the 1990s as it did in the 1960s.
An equally important change appears to have occurred in the tone and
content of the white candidates™ campaigns. Early black challengers faced

36 A much stronger test of this proposition would assess white voter support for a random
set of black challengers over time. Unfortunately, no such data set is available. Generally,
only major contests in large cities have exit polls or pre-election polls, and smaller cities
rarely have data on the racial and ethnic makeup of precincts, making it nearly impossible
to obtain the vote by race in most local contests.
The Transformation of the White Vote 71

white opponents who tried to play on white uncertainty about black lead-
ership. Since whites had little past experience with black leadership and
no idea what to expect, white candidates could campaign effectively by
playing on uncertainty. So, for example, in Los Angeles Sam Yorty could
run against Tom Bradley in 1973 by stating: “You know what kind of city
we™ve got. We don™t know what we might get. So we™d be taking quite
a chance with this particular kind of candidate.” Similarly, in Chicago
mayoral candidate Eddie Vrdolyak played on white concerns about the
consequences of black leadership: “It™s a racial thing. Don™t kid yourself.
I am calling on you to save your city, to save your precinct. We™re ¬ghting
to keep the city the way it is” (quoted in Rivlin, 1992: 155). In Newark,
the white police chief announced dramatically in 1970: “Whether we
survive or cease to exist depends on what you do on [election day]”
(quoted in Eisinger 1980: 15). In Atlanta, when Maynard Jackson ran in
1973, the slogan was, “Atlanta is too young to die.”37 But by the 1990s,
whites had had the chance to evaluate the effects of black leadership in
a range of major cities. Since Atlanta, Los Angeles, and other cities had
done relatively well under black leadership, white candidates could no
longer credibly claim that black leadership would mean the demise of the
white community or that the city™s police force would quit en masse if an
African American were elected mayor. The old scare tactics faded out of
There are other factors at work over this time period “ increasingly
strong norms against overt racism, and declining racial tension as the riots
of the 1960s and 1970s faded into memory “ so it is hard to know how
much of these changes are actually due to white learning. But the end result
is that more and more black challengers are winning of¬ce in majority
white cities. In the 1960s, there were no black mayors of majority white
cities with more than 50,000 residents. In the 1970s, 11 black candidates
won election in these majority white cities. In the 1980s, the number
increased to 15, while in the 1990s, 18 blacks won elections in majority
white cities. Eighteen may not seem to be a large number, but it actually
represents a sizeable portion of all big-city black mayors “ 57 percent

37 This phenomenon is not con¬ned to black mayoral candidates, of course. When Harvey
Gantt ran for the Senate in 1990, Jessie Helms countered with an advertisement depicting
a white worker losing a job to a less quali¬ed black man. Analysts felt that the advertise-
ment was the turning point of the campaign. “The ad scared some whites, convincing
them that Gantt was a threat to their future and . . . propelled Helms to victory” (Frisby
1991: 14). Years earlier, George Wallace was much more direct: “If I don™t win, them
niggers are going to control this state” (quoted in Black and Black 1973: 736).
Changing White Attitudes

in 2001. When blacks were being elected in that year, white voters were
critical to their success most of the time.
Changes in the white vote, in the kinds of campaigns white opponents
run, and in the success rates of black candidates in minority black locales
all hint at a sea change in the views and perceptions of a large segment
of the white community. Though it is impossible on the basis of the data
presented here to assign this change de¬nitively to the effects of white
learning from black leadership, they certainly leave open the possibility
that experience with black leaders is fundamentally altering the nature of
biracial politics in this country. To more directly assess changes in white
attitudes and perceptions, in Chapter 3 I turn to a closer inspection of
white racial views before, during, and after the election of blacks to the
But before we proceed, it is worth noting two quali¬cations. First,
as Table A.1 in Appendix A makes clear, there is considerable variation
across the cities in the data set. Although change in most cities led to
greater acceptance of black candidates and less racialized political con-
tests, black mayors in Atlantic City, Baltimore, Chicago, Durham, Flint,
New Haven, New Orleans, New York, Philadelphia, and San Francisco
faced more white opposition when they ran as incumbents than they did
when they sought the mayoralty in the ¬rst place. Some of these cases can
by explained by pointing to a poor local economy, a strong white chal-
lenger, or to more idiosyncratic events, such as scandals, but others can-
not. Chapter 4 will delve more deeply into differences across cities and
contexts and will try to link much of the variation in white reactions to
differences in the local information environment. Nevertheless, the lack
of positive change in many cities should give us some pause.
Second, even in the cities that did experience positive change, there
is evidence that race continues to play a critical role even in black
incumbent elections. And despite the changes noted here, blacks are still
greatly underrepresented at the mayoral level. The ICMA reports that
only 2.1 percent of all mayors in the nation are African American “ well
below the 12 percent of the national population that is African American
(MacManus and Bullock 1993). Clearly, the nation has a long way to go
before race no longer plays a role in mayoral politics. Expectations about
the dawn of an era of interracial collaboration should be tempered.

The Transformation of White Attitudes

The last chapter revealed important changes in the white vote. Experience
with black incumbency increased white support for black candidates and
altered the nature of the white vote in biracial elections. These changes
imply that whites are learning from their experience with black leadership,
but the fact that some white voters begin to choose black candidates over
white candidates does not tell us exactly what whites learn “ how their
views toward blacks change. The goal of this chapter is to focus directly on
white views to try to understand what white voters learn during and after
the transition from white to black leadership. The information that black
incumbency provides appears to have a broad impact on white views.
The tests analyzed here suggest that experience with black incumbents
leads to positive changes in how many whites perceive black leadership
and the larger black community. The changes are not always dramatic,
but they are real. The results also show that everyone does not learn
the same things from black leadership: among whites, Republicans and
Democrats, in particular, appear to learn differently, and the result is a
growing partisan divide on matters of race.

assessing white attitudes
To assess changes in white racial attitudes and policy preferences under
black mayors, I compared the attitudes of a representative sample of white
respondents in cities with a white mayor to the attitudes of a similar sam-
ple of white respondents in cities with a black mayor. The least compli-
cated and most direct test of white learning would examine the attitudes of
a representative group of individuals in a city over time as they experienced
Changing White Attitudes

the transition from white to black leadership. Unfortunately, this type of
panel data is unavailable. In lieu of it, I utilized responses from the Amer-
ican National Election Study (ANES). The ANES contains an array of
questions designed to gauge white racial attitudes and political orienta-
tion and is therefore a valuable tool for my purposes. Although the data
are cross-sectional and thus do not follow the same individuals over time,
I engage in a variety of tests to try to show that black leadership does, in
fact, cause a change in white attitudes. Combined, these tests hopefully
present a clear picture of change.
By pooling samples from the years 1984 to 1992, I obtained an ample
number of white responses from a wide variety of cities with black and
white mayors.1 Details on the sampling, survey instruments, and other
methodology concerning the survey can be found in Miller, and National
Election Studies (1994). Data on the race of mayors and council members
come from the National Roster of Black Elected Of¬cials. Other data on
city-level characteristics, such as racial demographics and median income,
were drawn from the relevant census publications (Bureau of the Census
1964“2003; 1994; 1990).
I looked for change in white views on two basic sets of measures: per-
ceptions of black leadership and feelings about the black community.2 I
chose these two areas because each addresses a critical aspect of the infor-
mation model. If white residents learn anything from their experiences
with black incumbents, it should be re¬‚ected ¬rst in their views of black
leadership. When black incumbents do not target the white community
but instead serve the interests of many members of the white community,
whites should recognize this and the perception that black leaders pose
a threat should be reduced. To see if perceptions of the threat posed by
black leadership changed under black incumbents, I examined responses

1 In this pooled sample, I included respondents from cities or primary areas with popula-
tions greater than 25,000. Within the sample, there are 1,605 white respondents living
in eighteen cities with black mayors. These cities are 10 to 76% black. Across the whole
pooled sample, there are 6,543 respondents from seventy cities represented in the data.
2 Another important step in the learning process is for white residents to recognize the
economic effects of black leadership. The ANES has one question that taps into this
kind of learning: “How much real change do you think there has been in the posi-
tion of black people in the past few years?” When I included it in supplementary tests,
these tests appeared to show that the longer a city had experienced black leadership, the
more likely white residents were to believe that the pace of racial change had slowed.
Since answers to this question could be interpreted less as an objective measure of racial
change and more as a liberal indictment of American society, the conclusions to be drawn
about white learning from this result are not clear.
The Transformation of White Attitudes 75

to the following question: “Some say that the civil rights people have been
trying to push too fast. Others feel they haven™t pushed fast enough. How
about you: Do you think that civil rights leaders are trying to push too
fast, are going too slowly, or are moving about the right speed?”3 This
particular question in some ways represents a dif¬cult ¬rst test. Since the
question does not refer to local leaders and local politics, whites would
have to generalize from their experiences with local black leadership to
all black leadership in order for there to be a change on this measure.4
The advantage of this particular question, however, is that it also taps
into broader aspects of race relations. Bobo (1983) has argued that this
question is a key indicator of the state of racial con¬‚ict, and tests indi-
cate that responses to the question are signi¬cantly correlated with recent
or imminent changes in local racial policy making (Bobo 1983). Thus,
the measure provides a reasonable assessment of race relations, and any
improvement in white views on it should be considered signi¬cant.
A second important question about white learning is whether white
views about the black community as a whole change under black lead-
ership. Whites may learn that black leadership is less threatening but,
nevertheless, they may not change their views about the larger black com-
munity. If white residents do not believe that local black leaders represent
the interests of the black community, then they may feel that the black
community still represents a signi¬cant threat. To assess this larger change
in white attitudes, I examined white responses to all other ANES ques-
tions in the pooled sample that tapped views about the black community
without getting into speci¬c policy debates. Of these questions, the most
general measure of white views toward blacks was a black feeling ther-
mometer. For the feeling thermometer, white respondents were asked how
“warmly” or “favorably” they felt toward blacks as a group on a scale
from 0 to 100. I reversed the scale to get a measure of anti-black affect.
Since different respondents can assign very different meaning to the same

3 Answers to the question were coded as follows: (1) too fast, (0.5) about the right speed,
(0) too slowly.
4 The task is made even tougher by the fact that this question, and the others, are embedded
in the ANES, a survey that focuses on national politics. However, given the extensive media
coverage of black mayors and the intense fears that many white respondents seem to have
about black leadership at the local level, there is reason to believe that white views even
on more national or general questions will be affected by their experiences with a black
mayor. If this assumption is incorrect, and white responses to these questions do not re¬‚ect
the knowledge that white residents have obtained from living under black mayors, then
we should expect to ¬nd no relationship between experience under a black mayor and
white attitudes.
Changing White Attitudes

value on the scale (that is, what does a value of 60 mean?), I controlled
for each respondent™s feelings toward whites on a similar reversed feel-
ing thermometer.5 The result is a measure of how negatively respondents
felt toward blacks as a group relative to their feelings toward whites “ a
measure that is akin to in-group favoritism (Tajfel 1981).6
A second measure of whites™ overall views of the black community
comes from the four remaining race questions in the pooled sample:
(1) “It™s really a matter of some people not trying hard enough; if
blacks would only try harder they could be just as well off as whites.”
(2) “Irish, Italian, Jewish and many other minorities overcame prejudice
and worked their way up. Blacks should do the same without special
favors.” (3) “Over the past few years, blacks have gotten less than they
deserve.” (4) “Generations of slavery and discrimination have created
conditions that make it dif¬cult for blacks to work their way out of the
lower class.” In each case, respondents were asked whether they agreed
or disagreed with the statement and how strongly.7
On the surface, each of these four questions addresses a different issue.
Yet each focuses on one central element of race relations: the extent to
which blacks face barriers in American society. The four questions are
not only linked conceptually, they are also linked empirically. Answers
to the four questions are highly correlated, with inter-item correlations
ranging from a low of 0.39 to a high of 0.59. Thus, I was able to create
what I call a racial resentment scale using these four questions.8 If black
representation does change white views about blacks as a group, then it
should be re¬‚ected in both the measure of anti-black affect and the racial
resentment scale.9

5 Since some respondents tend to be high raters and others low raters, in alternate tests I
standardized responses by controlling for each respondent™s mean score across an array
of six feeling thermometer measures. The following results are almost identical with or
without feelings toward whites controlled, and with or without standardizing by the mean.
The scale has a mean response of 0.55 (std. dev. = 0.10).
6 This scale can also be viewed as a measure of prejudice (Hurwitz and Pef¬‚ey 1998; Allport
1954). The more whites prefer whites as a group over blacks as a group, the more they
can be seen as prejudiced.
7 For each question, responses were (1) disagree strongly, (2) disagree somewhat, (3) neither
agree nor disagree, (4) agree somewhat, and (5) agree strongly.
8 This scale is almost identical to a racial resentment scale developed by Kinder and Sanders
9 To create the scale, responses to each question were ordered from least to most sympathetic
to blacks. Individual responses were then added together, and the scale was normalized to
a 0“1 range. The reliability of the scale is high, with a Cronbach™s alpha of 0.77. The scale
The Transformation of White Attitudes 77

To try to ensure that any differences in white attitudes between black-
run and white-run cities are really a function of black leadership, I con-
trolled for demographic and socioeconomic factors known or suspected
to affect white racial attitudes. The model includes the following control
variables: (1) personal socioeconomic characteristics “ age, education,
income, gender, employment status, home ownership, and the number of
years the respondent had lived in the city; and (2) contextual variables “
portion of the city that is black, level of urbanism, year of the interview,
and residence in the South.10 All independent variables were coded 0 to
1 for ease of interpretation.11 Coding and descriptive statistics for these
variables are included in Appendix B.

is centered on a value of 0.60, indicating that most white respondents are slightly more
resentful than they are sympathetic on these questions. Scores are distributed normally.
Standard deviation is 0.24. I tried two alternate tests to ensure the robustness of
my results. First, I repeated the following analysis with each individual question, rather
than the whole scale. Although statistical signi¬cance usually declines, there are few
substantive changes to the results. Second, I used maximum likelihood estimation
con¬rmatory factor analysis to develop a latent factor representing the main theme of
these four questions. When I substituted the latent factor into the following analysis, the
results were almost identical.
10 I also include in each model a dummy variable for cities that are within twelve months of
the transition to the ¬rst black mayor of a city. I do so for two reasons. First, there is likely
to be heightened racial tension during the transition year. As a result, white respondents
may be less rather than more racially tolerant in the period surrounding the election of
a city™s ¬rst black mayor. Second, white residents in cities that have just been taken over
by black mayors have had little experience with black leadership and hence little time to
learn what the consequences of black leadership are.
11 One possible confounding factor is white ¬‚ight. The rapid outmigration of a large num-
ber of anti-black white residents under black mayors could clearly affect the remaining
mix of white racial attitudes under black incumbents. As noted in Chapter 2, there are
several factors that suggest this is not a primary factor here. First, the rapid outmigra-
tion of whites was largely stemmed by the time these surveys were taken (Frey 1980).
Between 1980 and 1990 in the cities included in this survey, the proportion of each
city™s population that was black increased by 2%. Moreover, the proportion of the pop-
ulation that was black increased only marginally faster in cities that had black mayors
during this period (2.3% increase). Second, in the analysis, I control for the number
of years a respondent has been living in a community. If changes in white attitudes
were due to racially conservative whites moving out, one would expect that longer stays
would be associated with more liberal racial attitudes. This is decidedly not the case.
If anything, whites who live longer in their communities are less racially tolerant and
more resentful. This seems to con¬rm research on migration patterns, which indicates
that because moving is expensive, whites who leave cities tend to be younger, wealth-
ier, and better educated (Deane 1990). The poor, older, and less educated who are left
behind are the people who are most likely to be racially intolerant (Bobo 1983; Rieder
Changing White Attitudes

Finally, to help assuage fears that the changes we will see in white atti-
tudes under black mayors are not the result of especially racially tolerant
white communities electing black mayors, I compared the racial attitudes
of white residents in cities prior to the election of a black mayor with
the attitudes of whites in cities that did not elect black mayors. Across
the attitudinal measures looked at in this study, there were no consistent
or substantial differences between the two groups of white residents. In
other words, whites in cities that elected black mayors were not especially
racially tolerant before the arrival of a black mayor.12 It is only after years
of experience under a black incumbent that differences emerge. Given
that the main factors driving the election of black mayors are the size and
resources of the black community (Karnig and Welch 1980), it probably
should not be surprising to ¬nd that white racial views in cities electing
black mayors are not unlike white racial views elsewhere.

improved attitudes under black mayors
Whites do appear to learn from their experiences under black represen-
tation. As illustrated by Table 3.1, whites who live in cities governed by
black mayors have signi¬cantly more positive racial attitudes than whites
who live under white mayors. First, under black mayors, as one might
expect, there were clear changes in how whites perceived black leader-
ship. White residents who experienced black mayoral leadership were
less apt to believe that black leaders were pushing too hard. Given that
this particular question can also be viewed as a measure of perceived
racial group con¬‚ict, there is at least some evidence here that the elec-
tion of blacks to of¬ce diminishes racial tension, though the change is not
Importantly, it is not just whites™ views of black leadership that
changed under black mayors. Feelings toward the black community as

12 It should, however, be noted that whites in cities that were just about to elect black mayors
were slightly more liberal or Democratic leaning than whites in other cities (analysis not
13 Part of the reason why differences between the two types of cities are not very pro-
nounced may be that whites who live in cities with white mayors are observing black
mayoral leadership from afar and are themselves learning from those observations.
Although it is unlikely that this distance learning will have the same effect as actually
living under a black mayor “ witnessing ¬rst hand that one™s own community did well
under black leadership is probably very different from reading about economic trends in
another city in the newspaper “ it is quite possible that learning is occurring in a range of
The Transformation of White Attitudes 79

table 3.1 The Impact of Black Representation on White Racial Attitudes

Views of Black
Leadership Views of the Black Community
Blacks Pushing Racial
Too Hardb 1
Anti-black Affect
’0.21 (0.12)— ’0.02 (0.01)——— ’0.04 (0.01)———
Black Mayor (1 = yes
0 = no)
’1.5 (0.16)——— ’0.05 (0.01)——— ’0.21 (0.02)———
’0.00 (0.01) ’0.01 (0.02)
Income 0.04 (0.17)
0.04 (0.01)——— ’0.01 (0.02)
Age 1.2 (0.21)
0.23 (0.08)——— 0.02 (0.01)———
Gender (1 = male) 0.00 (0.01)
’2.1 (0.21)——— ’0.06 (0.01)——— ’0.27 (0.02)———
Ideology (1 = liberal)
’0.55 (0.14)———
Party ID (1 = Democrat) ’0.02 (0.01)
0.01 (0.01)
’0.62 (0.23)——— ’0.00 (0.01)
Employment Status 0.01 (0.02)
(1 = unemployed)
0.03 (0.01)——
Years Living in City 0.11 (0.13) 0.01 (0.01)
0.66 (0.28)—— 0.04 (0.01)——— 0.13 (0.03)———
Percent Black in City
’0.07 (0.02)———
Level of Urbanism 0.05 (0.16) 0.01 (0.01)
0.30 (0.10)——— 0.01 (0.00)——— 0.03 (0.01)——
South (1 = yes)
’0.24 (0.15)
’0.30 (0.14)—— 0.02 (0.01)——— 0.04 (0.01)———
’0.04 (0.01)———
’0.24 (0.16)
’0.38 (0.13)——— ’0.02 (0.01)———
1992 0.01 (0.01)
’0.13 (0.19)
First Year of Black 0.01 (0.01) 0.01 (0.02)
Mayoralty (1 = yes)
0.59 (0.01)——— 0.88 (0.02)———
’4.1 (0.25)———
Intercept 1
’0.64 (0.23)———
Intercept 2
Adj. R2 /pseudo R2 0.15 0.09 0.21
χ2 424
N 2597 2914 2461
Note: Figures are unstandardized coef¬cients with their standard errors.
——— p < 0.01
—— p < 0.05
— p < 0.10
a Ordinary least squares.
b Ordered logit.

a whole also appear to have shifted. Whites who lived under black may-
ors were signi¬cantly more likely to feel warmly toward the black com-
munity. For many whites, experience with black leadership also seems
to have led to a diminution of racial resentment. White residents in
cities with black mayors were substantially more sympathetic to black
Changing White Attitudes

interests and more understanding of the barriers facing the black com-

A Causal Connection?
Thus far, my analysis of Table 3.1 has overlooked an important issue.
Given the cross-sectional nature of the data, it is possible that the causal
arrow is reversed: cities with more tolerant white residents could be more
likely to elect a black mayor rather than the opposite. Although the avail-
able evidence seems to suggest that the presence of a black mayor in a
city depends much more on the size and resources of the black com-
munity than on the characteristics of the white community (Karnig and
Welch 1980), I nevertheless conducted several different tests on this point.
In each case, the results were the same: each model suggests that black
mayoral leadership does lead to less negative views of blacks and black
The ¬rst test is a two-stage least-squares model that uses instrumental
variables to try to address the question of causality. It is described in detail
in Appendix B, and the results of the model are displayed in Tables B.1
and B.2 in the appendix. This alternate test strongly suggests that white
views on race are signi¬cantly more positive under a black mayor than
they are under a white mayor even after the possibility of reverse causality
is taken into consideration. As an additional check on the link between a
black mayoralty and white racial attitudes, I reanalyzed the data with a
different sample of respondents. In this second alternate model, I included
only white respondents from cities that have had a black mayor at some
point in their history. Since all of the cities in this new sample elected black
mayors at some point, any positive change in white attitudes under black
leadership is less likely to be the especially racial tolerant cities electing
black mayors and is more likely to be directly related to experience with
an incumbent black mayor. The results, which are displayed in Table B.3

14 One might object that the model in Table 3.1 does not take into account all of the
potentially relevant factors in the local environment. Perhaps white attitudes depend on
the class makeup of the black population or the city-level mix of partisanship. However,
alternate test revealed no such links. Although both the class makeup of the black com-
munity and the percentage of Republicans in the city were related to the presence of a
black mayor (see Table B.3), their inclusion in these models did little to affect the overall
results. I could ¬nd no city-level contextual factor that eliminated the link between a
black mayoralty and white attitudes.
The Transformation of White Attitudes 81

in Appendix B, corroborate the ¬ndings in Table 3.1. In this more select
group of cities, the analysis reveals once again that under black mayors
anti-black affect declines, racial resentment wanes, and whites perceive
less racial group con¬‚ict.

When Do Whites™ Attitudes Change?
One last way to try to ensure that the changes we see under black mayors
are real is to look at changes over the course of black mayoral leader-
ship in a city. If changes in white racial attitudes really are a function
of information garnered from the actions of black incumbents, then we
should see whites beginning to feel less and less threatened and express-
ing increasingly positive racial attitudes as the years go by under a black
mayor. In Table 3.2, I test this proposition. Here we are only including
white respondents from cities that currently have a black mayor. The
dependent variables are the same: I consider attitudes toward black lead-
ership and views of the black community (anti-black affect and the racial
resentment scale). The independent variable in this case is the length of
time a city has had a black mayor. In order to ensure that the results are
not skewed by racially tolerant respondents from one or two cities that
have had decades of black mayoral leadership, I normalized the years of
black leadership in each city.15 This allowed me to compare the views
of respondents reporting at the beginning of a black mayor™s tenure with
respondents reporting later in the tenure of a black leader.16 It is worth
noting that across the entire white American population there has been
no overall improvement on these particular racial attitudes, and on some
questions whites in America actually feel more negatively toward blacks
than they did decades ago (Schuman et al. 1997). Thus, positive changes
under black mayors are not the result of a general improvement in racial
attitudes over time.

15 Thus, length of time under black leadership is measured on a scale from 0 to 1, with 0
being the ¬rst year a black mayor was elected and 1 being the last year the city had a
black mayor. To create the scale, I simply add up the number of years the city has had a
black mayor in of¬ce (excluding any interim years in which a white mayor held of¬ce).
I then calculate the number of years the city has had a black mayor at the time of the
survey for each respondent and divide the latter number by the former.
16 As a secondary test, I repeated the analysis with a nonnormalized scale (the number of
years under black leadership). The results were almost identical, but the level of sig-
ni¬cance of the effects of black leadership on white racial attitudes declined in some
Changing White Attitudes

table 3.2 How Time Under Black Leadership Affects White Racial Views

Views of Black
Leadership Views of the Black Community
Blacks Pushing Anti-black Racial
Too Hard2 Affectb Resentmenta
’0.05 (0.02)——— ’0.15 (0.04)———
0.75 (0.52)
Years Under a Black
Mayoralty (1 = last year,
0 = ¬rst year)
’1.8 (0.33)——— ’0.05 (0.01)——— ’0.22 (0.03)———
’0.05 (0.39) ’0.00 (0.03)
Income 0.00 (0.01)
1.4 (0.48)——— 0.04 (0.02)—— ’0.00 (0.04)
Gender (1 = male) 0.16 (0.17) 0.00 (0.01) 0.02 (0.02)
’2.5 (0.47)——— ’0.06 (0.02)——— ’0.28 (0.04)———
Ideology (1 = liberal)
’0.61 (0.30)——
Party ID (1 = Democrat) ’0.02 (0.03)
0.01 (0.01)
’0.00 (0.02)
Employment Status 0.05 (0.59) 0.02 (0.04)
(1 = unemployed)
’0.03 (0.31)
Years Living in City 0.01 (0.01) 0.03 (0.02)
0.08 (0.04)——
Percent Black in City 0.77 (0.52) 0.02 (0.02)
’0.07 (0.03)——
’0.01 (0.01)
Level of Urbanism 0.38 (0.40)
0.39 (0.22)—
South (1 = yes) 0.01 (0.01) 0.03 (0.02)
’0.34 (0.39)
’0.73 (0.38)— ’0.02 (0.01)—— 0.05 (0.02)——
’0.94 (0.44)—— ’0.02 (0.02)
’1.6 (0.47)——— 0.08 (0.03)———
’0.01 (0.01)
0.62 (0.02)——— 0.96 (0.04)———
’4.3 (0.65)———
Intercept 1
’0.61 (0.61)
Intercept 2
Adj. R2 /pseudo R2 0.20 0.08 0.21
χ2 138
N 622 800 800
Note: Figures are unstandardized coef¬cients with their standard errors.
——— p < 0.01
—— p < 0.05
— p < 0.10
a Ordinary least squares.
b Ordered logit.

The results here are not totally uniform, but there does appear to
be a clear link between the length of time a white respondent has lived
under black leadership and at least some of his or her views. In particular,
Table 3.2 reveals a signi¬cant positive relationship between time under
a black mayor and white views of the black community. As time passes
The Transformation of White Attitudes 83

and white residents™ experiences with black leadership grow, whites feel
less resentful and more warmly toward the black community. Alternate
tests also show that the longer a city has experienced black leadership, the
more likely whites are to believe that the pace of racial change is slowing.
In other words, as time goes by, white residents are more apt to recog-
nize that black leadership does not mean racial upheaval. At the same
time, there is no clear link in Table 3.2 between time under black mayoral
leadership and white perceptions of black leadership. As the years went
by, whites did not become more likely to say that black leaders were not
pushing hard. Whether this is because this particular question did not
tap into local politics effectively or because there is some other interven-
ing variable is unclear. Without panel data de¬nitive answers are hard to
come by.
One thing we can do is look at events in these cities over time to see
if they mirror the pattern of positive change that we see in this public
opinion analysis. Accounts from the bulk of the cities do suggest that
time and experience with black mayoral leadership are important vari-
ables. Atlanta is a typical case: Maynard Jackson™s mayoral victory was
followed by a period of bitter racial confrontation. Only with the passage
of time did the true impact of a black mayoralty become clear. Speci¬-
cally, it was only after Jackson played a pivotal role in breaking a strike
of low-paid, mostly black garbage workers that he began to receive more
support from the white community. According to one advisor, Jackson™s
actions helped make whites “less paranoid” (Scott 1977). Similarly, the
beginning of Mayor Coleman Young™s tenure in Detroit was marked by
considerable racial strife (Rich 1987). Only after Young proved to white
residents that he could govern the city fairly did racial tensions decline.
After Young prepared an austerity budget that cut city services and laid
off nearly 4,000 workers, his electoral support grew, and by the time of
his reelection, his two white primary opponents could muster only a com-
bined 19 percent of the white vote (Scott 1977). This anecdotal evidence
con¬rms that it takes time for white residents to assess the impact of
black leadership, just as it takes time for a black incumbent™s actions to
reduce white uncertainty enough to alter white attitudes toward the black
A review of Tables 3.1 and 3.2 reveals two other interesting ¬ndings.
First, the results suggest that the size of the black population is negatively
related to white views of blacks and black leadership. The higher the
percentage of blacks in a city, the more racial con¬‚ict whites perceive and
the more racially resentful they are. This ¬nding ¬ts into a long line of
Changing White Attitudes

research that has shown that white attitudes are closely shaped by the
size of the local black population.17 In most of these studies, as the black
population grows, race relations deteriorate.
This relationship does, however, raise an important question. Why
is there a negative relationship between a larger black presence and
white attitudes on one hand and a positive relationship between the pres-
ence of black leadership and white attitudes on the other? Why do the
actions of blacks in positions of authority seem to quell white fears, while
the presence of a large black population seems to increase racial tensions?
I argue that the critical difference between the positive effects of black
leadership and the negative effects of a larger black population is the
control that black leaders have over their actions and the white commu-
nity. What may make black political leaders different from black neigh-
bors or co-workers and consequently why black representation should
provide more credible information that alters white views is the fact
that when blacks assume political of¬ce, it marks one of the ¬rst times
that blacks have authority to enact policies or make changes that could
harm the white community. This perceived authority or control not only
makes black political leadership especially threatening, it also makes the
actions of black leaders especially informative. If whites believe that a
mayor has considerable power to in¬‚uence events (as surveys show they
do), then the fact that black leadership does not hurt white interests can
be seen as credible evidence that blacks are not out to get the white com-
munity. In contrast, the behavior of blacks in most other situations can
be dismissed because whites can surmise that black behavior was con-
If my interpretation is accurate, then these results help us to understand
the dynamics of interracial contact, suggesting that whites change their

17 In particular, a higher percentage of blacks in the local community has been linked to
increased white on black violence (Corzine, Creech, and Corzine 1983), an increased
sense of threat among whites (M. Taylor 1998; Fossett and Kiecolt 1989; Giles and
Evans 1986), higher levels of white bloc voting (Murray and Vedlitz 1978), greater white
support of racist candidates (Giles and Buckner 1993; Black and Black 1973), more con-
servative racial policy preferences (Glaser 1994), greater support for segregation (Petti-
grew and Campbell 1960), suppression of black voting (Matthews and Prothro 1963),
and decreased interracial friendship (Shaw 1973). Judging by these studies, white Ameri-
cans seem to be keenly aware of and very threatened by the presence of minorities in their
neighborhoods and cities. However, it is important to note that others have found either
no relationship between the size of the black community and whites attitudes (Bledsoe
et al. 1995) or a positive relationship (Welch et al. 2001; Kinder and Mendelberg 1995;
Carsey 1995).
The Transformation of White Attitudes 85

views of black leaders only when they receive information that they per-
ceive as credible. In particular, these ¬ndings corroborate recent research
in social psychology, which is increasingly ¬nding that how group mem-
bers receive information about out-groups is more critical than what infor-
mation they receive (Scarberry, Ratcliff, and Lord 1997; Bar-Tal 1997).18
The second interesting ¬nding to emerge from Tables 3.1 and 3.2 con-
cerns the link between white racial views, political ideology, and partisan-
ship. Both tables report a fairly strong relationship between identifying
as a Republican and holding conservative ideological views and holding
more negative views of black leaders and the black community. This sug-
gests that there may be important differences in attitudes toward blacks
across different segments of the white community. Given that all whites
do not feel the same way about the black community, we should probably
not expect that all whites will respond the same way to black leadership.
In particular, some whites may be less affected by the information black
incumbents provide.

Who Changes Their Mind?
Given the clearly divergent racial agendas of the Republican and Demo-
cratic parties and the link between Republicanism and more negative
views of the black community illustrated in Tables 3.1 and 3.2, one might
predict that white Republicans will be more resistant to the informa-
tion that black representation provides (Carmines and Stimson 1989;
Huckfeldt and Kohfeld 1989; Kinder and Sanders 1996; Kinder and Sears
1981). White Democrats, on the other hand, may be much more receptive
to having a positive view of blacks and black leadership, particularly since
such an attitude ¬ts into their political views more broadly. If this were
true, one should ¬nd that most if not all of the positive change in racial
attitudes and voting behavior that occurs under black mayoral leadership
is con¬ned to white Democrats.
In Table 3.3, I test this proposition by separating out the responses of
white Democrats, white Republicans, and white Independents to black

18 As I mentioned in the Introduction, the exceptional effects of black mayoral leadership on
white attitudes echo a series of studies in experimental psychology that suggests that one
of the few times whites really change their attitudes about blacks is when they know that
blacks are free to choose their actions (Wilder, Simon, and Faith 1996). The link between
mayoral authority and attitude change also ¬ts with formal models in political science
that have shown that actions by individuals are uninformative unless the individual is in
control and has the power to choose a different course of action (Lupia and McCubbins
Changing White Attitudes

table 3.3 How Time Under Black Leadership Affects the Views of
Democrats, Republicans, and Independents

Views of Black
Leadership Views of the Black Community
Blacks Pushing Anti-black Racial
Too Hardb Affecta Resentmenta
’0.08 (0.02)——— ’0.18 (0.04)———
Years Black Mayor 0.57 (0.52)
Years Black Mayor— ’0.08 (0.02)——— ’0.07 (0.04)——
0.01 (0.37)
Democrat View
Years Black Mayor— 0.12 (0.02)——— 0.14 (0.03)———
’0.42 (0.36)
Republican View
’1.7 (0.31)——— ’0.06 (0.01)——— ’0.22 (0.03)———
’0.04 (0.38)
Income 0.01 (0.02) 0.01 (0.03)
1.5 (0.46)——— 0.04 (0.02)——
Age 0.01 (0.04)
Gender (1 = male) 0.10 (0.17) 0.01 (0.01) 0.02 (0.02)
’0.06 (0.02)——— ’0.28 (0.04)———
Ideology (1 = liberal) ’2.6 (0.46)
’0.95 (0.58)— 0.15 (0.02)——— 0.14 (0.05)———
Party (1 = Democrat)
’0.00 (0.02)
Employment Status 0.20 (0.59) 0.01 (0.04)
(1 = unemployed)
Years Living in City 0.13 (0.29) 0.01 (0.01) 0.02 (0.02)
0.09 (0.04)——
Percent Black 0.01 (0.01) 0.03 (0.02)
in City
’0.05 (0.03)—
Level of Urbanism 0.41 (0.39) 0.01 (0.01)
South (1 = yes) 0.38 (0.21) 0.01 (0.01) 0.03 (0.02)
’0.46 (0.38)
’0.80 (0.37)—— 0.02 (0.01)— 0.05 (0.02)——
’0.94 (0.42)—— ’0.02 (0.02)
’1.6 (0.4)——— 0.09 (0.02)———
1992 0.02 (0.01)
0.54 (0.02)——— 0.88 (0.05)———
’4.7 (0.65)———
Intercept 1
’0.95 (0.61)
Intercept 2
Adj. R2 0.21 0.14 0.22
χ2 154
N 658 800 800
Note: Figures are unstandardized coef¬cients with their standard errors.
——— p < 0.01
—— p < 0.05
— p < 0.10
a Ordinary least squares.
b Ordered logit.
The Transformation of White Attitudes 87

mayoral leadership.19 The dependent variables are the same racial atti-
tude questions examined earlier. The only change to the model is that
in Table 3.3 I interact the years under a black mayor variable with
dummy variables for Democratic and Republican Party identi¬cation.
Independents become the baseline group. To gauge how white Indepen-
dents respond to time under black leadership, one need only look at the
coef¬cient for “Years Black Mayor.” To determine how Democrats (or
Republicans) respond to time under black leadership, one has to add
the coef¬cient for the “Years Black Mayor “ Democrat” (or Republican)
interaction term and the coef¬cient for “Years Black Mayor” together.
Table 3.3 shows that, as expected, white Democrats, white Republi-
cans, and white Independents respond differently to black representation.
First, the more time the white Independents in the sample spent under
black mayoral leadership, the more positive their racial views became:
as the years under a black mayor went by, white Independents felt less
anti-black affect and expressed less racial resentment. Changes were even
more pronounced and positive for white Democrats: in two of the three
cases, white Democrats™ views improved even more rapidly than those
of white Independents. Republicans, however, responded less positively
to time under a black mayor. On two of the three attitudinal measures,
Republican views improved signi¬cantly less than those of Independents.
Adding the Republican interaction term with the “Years Black Mayor”
term it appears, in fact, that there was little to no change in the level
of racial resentment and anti-black affect among white Republicans. The
same pattern emerges if one does not look at change over time under black
leadership but instead simply compares whites in cities with a black mayor
to whites in cities with a white mayor, as was done in Table 3.1. White
Democrats who lived in cities with a black mayor had signi¬cantly more
positive views of blacks than white Democrats who lived in cities with a
white mayor. By contrast, white Republicans™ racial views changed only
marginally. This suggests that some white Republicans either ignore or dis-
count the words and actions of black incumbents. Given this resistance
to the information that black leadership provides, it seems reasonable to
argue that a racial prejudice or racial stereotype model of political behav-
ior may more accurately describe the views of some Republicans than an
information model. In short, the information model may not apply to all
whites equally.

19 Independents are self-identi¬ed Independents or respondents who listed no party
Changing White Attitudes

The same pattern is also evident when one looks at the actual vote in
mayoral elections involving black incumbents. As noted in Chapter 2, in
the few cases where voting data are broken down by both race and party
identi¬cation, the evidence suggests that the bulk of the increase in white
support for black incumbents came from white Democrats (Sonenshein
1993; Pettigrew 1976). White Republicans, in contrast, tended to vote
against black candidates whether they were challengers or incumbents.
This suggests that racial considerations are most consequential for white
Democrats (see Hurwitz and Pef¬‚ey 1998 on this point). Whether white
Democrats support black leadership seems to be greatly affected by their
views of blacks, while white Republicans may have a range of reasons for
opposing black candidates that may or may not include racial considera-
The end result of these divergent patterns is that black representa-
tion has a polarizing effect on the white community. Democrats become
more and more racially liberal, while Republicans™ views stay largely
the same. In fact, the racial attitudes gap between white Democrats and
white Republicans more than doubles in size under black mayoral lead-
ership. Table 3.4 illustrates this growing gap by presenting the difference
between mean Democratic views and mean Republican views across the
same series of racial questions. As column one of Table 3.4 reveals, the
gap between white Democrats and white Republicans in cities with a
white mayor was rather small. On one of the three questions, the dif-
ference is not even statistically signi¬cant (p < .05). But in cities with a
black mayor, the gap between white Democrats and white Republicans
on each indicator doubled or more than doubled. On matters of race,

20 Although one could argue that this pattern is the result of the defection of racist white
Democrats from the party, there are two factors that make this highly unlikely. First,
since most Democratic defectors become Independents rather than Republicans, one
would expect any negative change to be among Independents, not Republicans. Second,
if this pattern is due to racial realignment, one would expect that the pattern would be
more pronounced in the South, where racial realignment was itself most pronounced.
Additional tests, however, reveal no clear difference in the pattern between Southern
and non-Southern cities. Third, by all accounts, racial realignment was a slow, gradual
process. Thus, it is a stretch to argue that realignment can account for changes in the
Democratic and Republican votes that occur in the same city over a four-year period. Not
enough Democrats defect in a four-year period to change the Democratic vote as much
as it changes in many cities. Finally, it seems unlikely that the gradual defection of whites
from the Democratic Party over time could account simultaneously for the differences in
the views of different white partisans who live under black mayors and for differential
changes in the views of different white partisans as the years under black leadership
go by.
The Transformation of White Attitudes 89

table 3.4 Black Representation and the Polarization
of the White Community

Difference Between Mean
Democratic and Mean
Republican View
Cities Without a Cities With a
Black Mayor Black Mayor
’0.06—— ’0.11——
Sense of Black Threata
Anti-black Affecta
’0.04—— ’0.11——
Racial Resentmenta
Note: Indicates that difference between Democratic and Republican
mean is signi¬cant.
— p < 0.10
—— p < 0.05
——— p < 0.01
a All dependent variables coded 0“1.

black representation means even greater division between Democrats and

The tests in this chapter indicate that experience with black mayors can
lead to improved white views of black leadership and signi¬cant positive
changes in white attitudes toward the black community at large. These
changes are all consistent with an information model of white behavior.
When blacks are perceived to have the power to hurt white interests and
when by all accounts they do not do so, many whites may interpret that
experience as a sign that blacks and black leadership are not as bad as
they may have feared and their attitudes appear to change accordingly.
All of this ¬ts well the information model. None of it is predicted by a
racial prejudice or a social dominance view. We thus have a fairly clear
test and fairly compelling evidence in favor of the information model. We
also have fairly strong evidence that many whites in American cities are
far from prejudiced. They are open to change and willing to look at the
black community in a more positive light.
At the same time, the data also indicate that black representation has
signi¬cant limitations. Not everyone seems to “learn” from black leader-
ship. The change that occurs within the white community, when consid-
ered in the aggregate, is apparently largely to be con¬ned to Democrats
Changing White Attitudes

and Independents. It seems as if part of America remains prejudiced and
is unwilling or unable to change. Moreover, the limited nature of change
to date is evident in accounts of policy divides from many of the cities
included in the data set. Despite years and often decades of black mayoral
leadership, whites and blacks continue to differ over minority contract
set-asides, police behavior, public schools, and downtown versus neigh-
borhood development, among other things (Rivlin 1992; Stone 1997). At
least over the short term, black representation has only a limited impact on
the “extraordinary racial divide” that separates white and black America
(Kinder and Sanders 1996: 28).

Learning Across Different Cities

To say that black representation generally leads to signi¬cant positive
change in white racial attitudes and white voting behavior is not to say
that white Americans in all contexts are equally likely to be affected by
black leadership. Indeed, if the information model is accurate, then white
reactions should vary sharply depending on the local information envi-
ronment. In this chapter, I begin to consider how context affects both
the information that whites receive from black leadership and the nature
of their response to that leadership. White reactions to black leadership
depend greatly on the amount of control black leadership is perceived to
exercise over local policies and conditions. The higher the level of per-
ceived control, the more information whites obtain from black leadership
and the more positively they respond to black incumbents.
A brief account of events in two different cities, Memphis and New
York, highlights the enormous variation in white reactions to black lead-
ership and raises important questions about why white residents respond
so differently in different contexts. Memphis and New York are parti-
cularly interesting because the two cities followed similar patterns before
the election of a black mayor and only diverged after that mayor entered
of¬ce. In both cities, black mayoral leadership came only recently and
only after overcoming signi¬cant white opposition.
In Memphis, the ¬rst black mayor was elected in 1991. The late arrival
of a black mayoralty in Memphis was not the result of a lack of effort
on the part of the black community. In every election between 1971 and
1991, black candidates vied for the mayoralty, with widespread support
from black leaders and black voters. But every time a black candidate ran
for election, the white community in Memphis responded with almost
Changing White Attitudes

unanimous support for the white candidate. The most successful of the
black challengers, Otis Higgs, managed to win a paltry 10 percent of the
white vote in 1975. In each election, whites won by trumping African
American turnout (Wright 2000). Given this history, it is no surprise that
the white vote was once again squarely behind the white candidate on
October 4, 1991, when another black challenger, Willie Herenton, sought
the mayoralty. Despite the fact that Herenton™s campaign tried to avoid
racial issues, and despite the fact that there were few major policy differ-
ences between Herenton and his white opponent, white voters in Memphis
once again refused to support the black challenger. In what has been
described as “one of the most racially polarized mayoral elections in urban
American history,” Herenton was opposed by 98 percent of white voters,
who turned out in record numbers to try to defeat the black candidate
(Pohlman and Kirby 1996: xv; Wright 1996). Fortunately for Herenton
and his black supporters, the outcome of this election was different. An
African American ¬nally took over the mayoralty.
What makes the Memphis story so remarkable is what happened
after Willie Herenton assumed power: decades of heated white opposi-
tion faded away, and support for black leadership grew dramatically. By
Herenton™s fourth year in of¬ce, a citywide poll revealed that three-
quarters of all whites in the city approved of the job that he had done and
a slim majority thought he should be reelected (A. Davis 1995; Polhman
and Kirby 1996). Later that year, Herenton, running as an incumbent,
received the support of nearly 40 percent of all white voters (Hobbs 1995).
Four years later, in 1999, he won again with half of the white vote, prompt-
ing the Commercial Appeal, the city™s main newspaper, to highlight the
“unprecedented unity” surrounding the election (Goad 1999). In his third
reelection bid, in 2003, Herenton faced no serious opposition “ white or
This turnaround in white political behavior in Memphis seems all
the more noteworthy when Memphis is contrasted with New York City.
African Americans in New York City have also long had a dif¬cult time
getting elected. Historically, black representation on the city council and
other elected of¬ces in the city has been well below parity (Logan and
Mollenkopf 2003; Mollenkopf 1986). As John Mollenkopf has noted,
“New York City has not incorporated minorities and . . . has not pro-
duced policies that are especially aimed toward minorities” (1986: 591).
Thus, when David Dinkins ran to become the city™s ¬rst black mayor in
1989, most observers were not surprised to ¬nd the city racially polarized
Learning Across Different Cities 93

in response. Despite Dinkins™s repeated attempts to appeal to the city™s
white voters by stressing nonthreatening themes, such as ¬scal prudence
and law and order, he managed to garner only 28 percent of the white
vote (Kaufman 1998; Arian et al. 1991). Nevertheless, Dinkins became
the city™s ¬rst black mayor.
In sharp contrast to events in Memphis, after Dinkins was elected, black
leadership in New York was met with anything but growing acceptance.
In fact, race relations in the city appeared to get worse rather than better
with Dinkins in of¬ce (Sleeper 1993). The next four years were marred by a
series of high pro¬le racial incidents. And when Dinkins ran for reelection,
racial polarization actually increased. In the 1993 election, 78 percent of
white voters opposed Dinkins “ an increase that was just enough to give
his white opponent, Rudy Giuliani, the victory (Kaufman 1998). Black
leadership in New York City was over almost before it began.1
These examples suggest that in some cases black leadership leads to
racial conciliation, while in other cases it leads to continued, if not height-


. 3
( 7)