James M. Vanderleeuw
and
Baodong Liu
Published in
Urban Affairs Review.
ABSTRACT
Borrowing findings from the literature on voter turnout, the authors examine the causes of roll-off in city council contests among black voters in New Orleans, a black empowerment area, between 1965 and 1998. The findings suggest the relevance of institutional power on group political participation. Roll-off among black voters declined after blacks held the majority of city council seats. Moreover, the findings indicate the relevance of election competitiveness. Black voter roll-off was lower in runoff elections than in primaries. Finally, the findings suggest that mobilization by black candidates, particularly by black incumbents, may yield enhanced political participation among black voters in urban elections.
The act of voting is one of the most important forms of citizen political participation in a democratic political system. Accordingly, the decline of turnout in American elections has drawn considerable attention from students of American politics (see Lijpart 1997 for an extensive review of turnout explanations). The traditional literature on turnout asks this central question: Who votes (Abramson, Aldrich, and Rohde 1999,65)? But voting participation is not just about who goes to the voting place. In addition to who votes, we can ask the following: To what extent do those, who turn out to vote, vote? Voters rarely are offered only one election on a ballot. In fact, voters often are offered many elections, from highly visible contests such as the presidential election to contests such as those for school board. If a voter participates in the top-of-the-ballot election and does not vote in other elections on the ballot, participation is not complete. Scholars refer to this failure to complete a ballot as voter roll-off.
The existence of voter roll-off has important consequences. While in recent years turnout in off-year congressional elections has been about 35%, and turnout in local elections around 25%, voter roll-off reduced the level of political participation in elections further down the ballot even further (Ansolabehere and Tyengar 1995, 145-146; Teixeira 1992, 7). To be sure, voter roll-off is relevant both to democratic political theory, that emphasizes citizen participation, and to public policy, where contests for consequential public office and issues often are decided in the electoral arena (Bullock and Dunn 1996).
The research on roll-off, however, has been largely limited in the traditional political participation literature in general and turnout studies in particular. Broadly speaking, traditional participation literature treats those who vote in a given election as the participants in that election, and those who did not turn out as the nonparticipants. However, this dichotomous classification is obviously too restricted because it ignores the fact that some voters do turn out, though they do not complete their ballot. These voters can be regarded as partial participants. Accordingly, it is logical to classify eligible voters into three groups: full participants (who vote in all elections on a ballot), partial participants (who vote in a portion of the elections listed on a ballot), and nonparticipants (who do not turn out to vote).
Although these three groups of eligible voters are conceptually distinct, it can be argued that partial participants, like full participants, have overcome some election costs that might have prevented them from voting, by showing up at the “voting booth” in the first place. Perhaps this is why the traditional turnout literature has tended to treat partial participants the same as full participants, as far as political participation is concerned. Most turnout scholars do not ask whether there is any similarity between nonparticipants and partial participants. It should be clear, nonetheless, that partial participants share at least one characteristic with nonparticipants -- they failed to vote in certain elections listed on the ballot.
Can some factors that have been suggested by the traditional political participation literature as reasons for nonvoting also be causes of partial participation (i.e., voter roll-off)? We address this question in the context of roll-off among black voters in urban elections.
SIMILARITIES BETWEEN NONPARTICIPANTS AND PARTIAL PARTICIPANTS
Brady, Verba and Schlozman (1995, 271) characterize empirical findings on the causes of citizens’ political nonparticipation as falling into three broad categories – citizens fail to participate “because they can’t, because they don’t want to, or because nobody asked.” Like most researchers, Brady et al. use voting as a measure of political participation. Though their analysis is perhaps one of the most thorough studies of political participation, measured by voter turnout, it does not consider the fact that some voters, though they did turn out to vote, failed to complete the ballot. Voting, as stated above, is not just about entering the “voting booth.” A question becomes: Do the three broad categories of nonparticipation suggested by Brady et al also apply to voter roll-off?
The findings of most previous roll-off studies tend to provide answers compatible with the “they can’t” explanation for nonparticipation -- voters roll off because they lack certain skills or resources. Some research findings, for example, suggests that certain ballot formats more than others confuse and/or "fatigue” voters, and therefore cause some voters to cease casting a ballot (Darcy and Schneider 1989; Nichols and Strizek 1995; Walker 1966). Furthermore, demographic characteristics of voters associated with turnout, such as education and political knowledge, have been shown to be associated with roll-off (Clubb and Traugott 1972; Sheffield and Hadley 1984). Some research findings also indicate that voters refrain from casting a ballot when they lack information about a particular election or public office (Bowler, Donavan and Happ 1992).
An additional finding that supports the argument that
voters roll off because “they can’t” fully participate concerns the gap between
white and black roll-off rates. It has been repeatedly reported that in
many elections black roll-off rates are higher than those of whites (see, among
others, Bullock and Dunn 1996;Engstromand Caridas 1991; Vanderleeuw and
Engstrom 1987). It is plausible to argue that blacks are more likely to
roll off due to the lower level of education among black voters, and their lack
of political and economic resources.
It may be the case that even if black voters “can” more fully participate (that is, they have the skills and resources to complete a ballot), they "don't want to." Brady et al. (1995, 271) explained “they don’t want to” as “the absence of psychological engagement with politics-- a lack of interest in politics, minimal concern with public issues, a sense that activity makes no difference, and no consciousness of membership in a group with shared political interests.” With respect to roll-off, particularly in local elections, it is important to determine what enhances black voters’ “psychological engagement” with an often “ignored” electoral office so that black voters may come to “want to” participate. Some analysts have found that the gap between white and black roll-off rates may be reduced when an election contest or a referendum located further down on a ballot is “racially salient” (Engstrom and Caridas 1991,185; Vanderleeuw and Engstrom1987). Racially based campaign rhetoric, a black-on-white election, or a policy issue with racial impact may heighten the level of black political interest so that roll-off among black voters declines.
BLACK VOTER ROLL-OFFAND THE RACIAL CONTEXT OF URBAN ELECTIONS
Over the last several decades the urban political landscape in the United States has undergone a dramatic change. African Americans have been elected to key positions in some of the largest cities, such as Cleveland, Los Angeles, Newark, Atlanta, Detroit, New Orleans, Chicago, New York City, Memphis, Portland and St. Louis. From the perspective of the racial context of modern urban elections, some findings have particular relevance for an analysis of roll-off among black voters in local elections. In their study of Milwaukee, Harris and Zipp (1999) conclude that black electoral victories in major black empowerment areas such as New Orleans and Detroit may influence black voters who reside in majority white cities, such as Milwaukee. To explain this “spillover effect” of black electoral victories, the authors argue that “There now are enough models of blacks holding office – both in cities where blacks have a numerical majority and those where they do not -- that voting for a black candidate is not seen as a wasted vote in Milwaukee” (Harris and Zipp 1999, 496). If the election of blacks to public office in a black empowerment area can have symbolic value for a minority African American electorate that has not yet elected a black candidate, the remaining question is whether the level of black voter roll-off in a black empowerment area itself may be reduced as a result of enhanced political interest and political power among black voters in that area.
In the following pages we will examine the effect of institutional black political power on roll-off among black voters in the City of New Orleans, as one test of the “don’t want to” explanation. We choose New Orleans because this city is a black empowerment area, which is supposed to have influenced blacks elsewhere. In addition to determining the extent to which the “don’t want to” explanation can account for black voter roll-off, New Orleans provides the opportunity to examine the “nobody asked” explanation. Brady et al. explain “nobody asked” as “isolation from the recruitment networks through which citizens are mobilized to politics”(1995 271. See Rosenston and Hanson (1993) for a further discussion of this mobilization model). Vanderleeuw and Utter (1993) discovered that the percentage of black roll-off voters in New Orleans was negatively related to the number of black candidates in elections, which suggests that a mobilization model might explain some of the variation in black roll-off(from the perspective of the “nobody asked” explanation, black candidates were the somebody who “asked”). Vanderleeuw and Utter’s data, however, were based on elections from 1965 to 1986,before blacks controlled the majority of city-council seats; therefore, their analysis did not address the “don’t want to” explanation.
The following section provides four hypotheses based on factors related to the “don’t want to” and “nobody asked” explanations for political nonparticipation as applied to black voter roll-off in the city of New Orleans.
HYPOTHESES
As stated above, one focus of our investigation is the effect of institutional black political power. This focus is designed to ascertain conditions under which black voters change from “don’t want to” to “want to” complete their ballot. The emphasis on institutional black political power is inspired by a fundamental assumption derived from Key's (1949) conflict-based explanation of racial politics. This assumption is that group-based political behavior is a reflection of a cohort's relative political status. Office-holding in a governing institution provides to group members symbolic status. When members of a group traditionally deprived access to seats on a legislative body come to hold a majority of such seats, that cohort derives enhanced status. This enhanced status becomes manifest in heightened electoral participation (Bledsoe 1986; Bobo and Gilliam 1990; Peterson and Somit 1997). Our first hypothesis therefore is that institutional black political power, defined as a black majority on the city council, will reduce the level of black roll-off. Whether or not black voters “don’t want to” fully participate in elections may also be related to the competitiveness of an election. Runoff elections eliminate all but the two strongest candidates. Voters motivated to participate in atop-of-the-ballot runoff election are more likely to participate in an election further down the ballot, by comparison to voters in a primary election. Our second hypothesis therefore is that black voter roll-off will be lower in runoff elections than in primary elections.
To ascertain whether or not roll-off among black voters is related to “nobody asked” explanation, we examine the extent to which changes in the level of black voter roll-off can be accounted for by a mobilization model. Here we investigate two additional hypotheses. We first investigate how the presence of black candidates affects roll-off. Black candidates, on average possessing the most extensive social and political connection to black communities, have the greatest ability to mobilize black voters (Vanderleeuw and Utter1993). Accordingly, our third hypothesis is that black voter roll-off will decline as the number of black candidates increase. Mobilization requires financial and organizational bases. In this regard, we expect that black incumbents have more advantages than on-incumbents. Office-holding along with greater campaign resources better positions incumbents to inform citizens about relevant issues and positions. Our fourth hypothesis therefore is that elections with a black incumbent will reduce roll-off among black voters.
THE SETTING
New Orleans figures prominently in the study of southern politics, political empowerment, and race and voting behavior. From the time of V.O. Key (1949) scholars have utilized New Orleans as a laboratory -- its long history of racial politics making it an appropriate setting for a variety of social and political inquiries centered around race.
Both the population and political leadership of New Orleans have undergone pronounced racial change, which makes this city a valuable setting for our present investigation. In1960 the city’s population was 37 percent black. By 1990 the population was 62 percent black. At the time of the passage of the1965 Voting Rights Act, 17percent of registered voters were black. By 1986 a majority of the city's registered voters were black; by 1998blacks comprised 63 percent of registered voters.
Over the last several decades the racial composition of New Orleans' elected leadership has shifted from majority white to majority black. Until the mid 1970s this city had an all-white council. The first black city council member was appointed in 1976;blacks were elected to the city council for the first time in 1977. In 1986 voters elected the city's first majority-black council; blacks won four of the seven city council seats. The city council has been majority-black since that time. In 1994 the number of council seats held by blacks increased to five. In 1977 voters elected the city's first black mayor. Since that time no white has won the mayoralty.
DATA
Though they provide much insight, previous studies of voter roll-off offer relatively little regarding the effect of the racial context of urban elections, due at least in part to the paucity of longitudinal data. In the present study we utilize election data on 46 contests for New Orleans city council from 1965 through 1998. Five seats on the New Orleans city council are elected from single-member district and two are elected at-large. Our analysis employs both primary and runoff election contests for district seats that were accompanied by a top-of-the-ballot mayoral election. The municipal election years covered are 1965, 1969, 1977, 1982, 1986, 1990, 1994 and 1998. [1]
To calculate black voter roll-off we utilize aggregate, precinct-level data. We first calculate a measure of overall (i.e., nonracial) roll-off: the number of votes cast in a city council contest expressed as a percentage of the number of votes cast in the top-of-the-ballot mayoral election. We then estimate the percentage of black voters in the mayoral election who rolled-off in the city council contests by regressing overall roll-off onto the percentage of registered voters who are black in each relevant precinct (for black voter roll-off, Mean=14.8, SD=8.6, Range=1.1-36.0). We use ecological regression because aggregate racial composition and election figures only are available for the time-frame under investigation. The value here from our perspective is that we can investigate patterns in roll-off across a 33 year period. [2]
The two variables used to measure the “don’t want to” explanation are institutional black political power and election type. We define institutional black political power as a majority-black city council. Contests that take place when blacks comprise a council majority are coded 1 (N=18) and all others are coded 0 (N=28). Election type is coded 1 for a runoff (N=11) and 0 for a primary (N=35). The two variables used to measure the “nobody asked” explanation are voter mobilization by black candidates and by black incumbents. The influence of voter mobilization by black candidates is measured by the number of black candidates competing in a council district election contest (Mean=2.1, SD=2.6, Range=0-13). Mobilization by black incumbents is measured by black incumbency status. Council district elections that involve a black incumbent are coded 1 (N=12) and all others are coded 0 (N=34).
As indicated above, New Orleans has undergone racial transition both in its population and in its political leadership. New Orleans is considered a black empowerment area because of the increasing power of African Americans in City Hall. This pattern of racial and political change is consistent with the minority incorporation theory suggested by Browning et al. (1984, 1997), who argue that the size of a minority population affects the formation as well as the strength of urban minority political organization. In the following analysis, therefore, we account for the potential effect of the racial composition of the New Orleans electorate. The percentage of registered voters in a district who are black is our measure of the electorate's racial composition (Mean=48.1, SD=19.7, Range=16.0-82.7).[3] We will also test for the effect of time in order to determine the extent to which black voters exhibited an increasing interest in their local elections over time (suggested by Harris and Zipp 1999). For the time variable, contests in the earliest election year, 1965, are coded 0, and subsequent elections are assigned a value that reflects the number of years beyond this base-point.
FINDINGS
Table 1 provides summaries of our four hypothesis-related variables based on descriptive findings. These findings suggest the relevance of the “don’t want to” as well as the “nobody asked” explanations for roll-off among black voters.
The mean level of roll-off among black voters was almost 11 percentage points lower for elections held in the context of a majority-black city council, by comparison to those held in the context of a majority-white council. Mean black roll-off was about nine percentage points lower in runoff elections than in primaries. The level of black voter roll-off also may be associated with mobilization effects of black candidates. The mean level of black roll-off was about nine percentage points lower in elections where least five black candidates competed, by comparison to elections with no black candidate. Mean roll-off was almost five percentage points lower when a council district election included a black incumbent.
Though the findings in Table 1 suggest the applicability of the “don’t want to” and “nobody asked” explanations, descriptive findings also suggest that various electoral characteristics may be highly collinear. Table 2 reports descriptive statistics on our variables for each New Orleans council district election contest under investigation, ordered by election year.
The municipal elections of 1977 represent an important turning point in New Orleans’ politics. In 1977 city voters elected New Orleans’ first black mayor, Ernest Morial. Blacks competed for city council seats in substantial numbers, and for the first time African Americans were elected to the council. Prior to1977, blacks competed in only two council district contests, the 1969 primaries for district seats A and B. Only 4.7% of candidacies across these 12 contests held before 1977 were black candidacies (3 out of 39), and there were no black incumbents. The electorate’s black share exceeded 40% in only the 1969 elections for the District B seat, and elections took place within the context of a majority-white city council. The average level of roll-off among black voters across these 12 election contests is estimated at 17.9% (SD=8.0).
Starting with the municipal elections of 1977there is a marked increase in the number of black candidates -- blacks competed for council seats in 29 of the 34 elections held between 1977 and 1998; 64.4%of all candidacies were black candidacies (94 out of 147). Elections increasingly involved a black incumbent (in 12 cases), and in a majority of cases the black share of the electorate exceeded 50% (in 24 elections). The average level of black voter roll-off among these 34 contests is estimated at 13.7% (SD=8.7).
Beginning with the 1990 elections, contests for council district seats were held within the context of a majority-black city council (a majority-black council is elected in 1986). In this setting the candidate field was dominated by black candidates. Among the 16 contested city council elections held from 1990 through 1998, the candidate field in nine cases was all-black; in two others, the 1994 elections for district seats B and E, the election was characterized by a large field of black candidates with a sole white entry. In most of the council district seat contests from 1990 through 1998, blacks comprised the majority of registered voters. The level of roll-off among black voters averaged 7.8% (SD=2.9). By 1998, all candidates in these contested council district elections were black, as were all incumbents. In each case, the black share of the electorate exceeded 68%. Roll-off among black voters averaged 6.8%.
To avoid difficulties in the subsequent multivariate analyses we investigated for collinearity among our explanatory variables. We detected no serious mutlicollinearity among our four hypothesis-related variables – majority black council, election type, number of black candidates, and black incumbency. The Phi value among sets of our three dichotomous variables -- majority black council, election type and black incumbency – did not exceed .218; in fact, none of these three bivariate associations was statistically significant at the .05 level. The bivariate correlation (r value) between the number of black candidates and each of these three dichotomous variables in no case exceed .358; the association between black candidates and black incumbency was not statistically significant at the .05 level. The time and black city council variables, however, were correlated at .783, and the racial composition variable was correlated at or greater than .600 with the black city council variable, with the number of black candidates, and with the time variable (relationships significant at .05). These variables therefore are entered into separate equations in the following multivariate analyses.
Equation 1 of Table 3 supports all four of our hypotheses. The percentage of black roll-off voters declined when black voters “wanted to” more fully participate. The independent effect of institutional black political power was to reduce the level of roll-off among black voters by an estimated eight percentage points. The type of election within which voters had the opportunity to participate produced a somewhat stronger effect; black voter roll-off declined almost12 percentage points in a runoff election. The percentage of black roll-off voters was also reduced when black voters were “asked to” more fully participate. Though the strength of the mobilization variables is somewhat less pronounced by comparison to the strength of institutional black political power and election type (comparing Beta coefficients), each mobilization variable has a statistically significant independent effect. For any additional black candidate, the level of black voter roll-off declined by almost one percentage point; the presence of a black incumbent reduced roll-off by about five percentage points. These four variables explain nearly two-thirds of the variation in the percentage of black voters who rolled-off across these 46 New Orleans council district election contests.[4]
In Equation 2 of Table 3, the effect of time is taken into account. The majority black council variable was excluded because of collinearity with the time variable, as noted above. The remaining variables – election type, the number of black candidates, and the presence of a black incumbent – continue to have a statistically significant independent effect when time is entered. Though the level of black voter roll-off diminishes over the years, the independent influence of time per se is not statistically reliable.
The independent influence of the electorate’s racial composition is taken into account in the model reported in Equation 3 of Table 3. Because the electorate’s racial makeup correlated highly with both institutional black political power and the number of black candidates, we excluded these two variables. Growth in the black share of the electorate reduced the level of roll-off, and this effect was somewhat more pronounced than that of election type. With the influence of racial composition controlled, black incumbency’s effect is statistically unreliable. This model, however, accounts for 48% of the variation in black voter roll-off (Adj. R2=.483), which is almost 18 percentage points less than the amount of variation accounted for by the model reported in Equation 1 (Adj. R2=.655). Given this differential, the four-variable model reported in Equation 1 therefore offers the best explanation for roll-off among black voters in contests for New Orleans’ council district seats.
EXPLANATIONS FOR BLACK VOTER ROLL-OFF
“They can’t,” “they don’t want to,” and “nobody asked” have been offered as explanations for the low levels of voter turnout commonly seen in the United States. This study applied two of these explanations –“they don’t want to” and “nobody asked” – to partial participation in voting, i.e., voter roll-off. The specific setting was roll-off among black voters in the city of New Orleans, a black empowerment area. The findings are enlightening.
With regard to whether black voters roll-off because “they don’t want to,” our findings indicate that roll-off among black voters is sensitive to the broader institutional context within which elections take place. A major concern of ours was the effect of institutional black political power. The percentage of black voters in the mayoral election who rolled-off in city council contests declined significantly when blacks held the majority of city council seats. This supports the argument that institutional political power is relevant to understanding a group’s political behavior, and suggests that black urban political empowerment enhances the level of political efficacy among African Americans, which yields more complete voting. Also with regard to the “they don’t want to” explanation, our findings indicate that the level of black voter roll-off is responsive to the specific election context, confirming the hypothesized relationship between black voter roll-off and the type of election in which voters can participate. The level of black voter roll-off in New Orleans city council contests was significantly lower in highly competitive runoff elections than in primaries. This suggests the relevance of election salience in producing a higher level of ballot participation among voters.
With regard to whether black voters rolled-off because “nobody asked,” our findings strongly suggest the importance of candidate mobilization efforts. The level of black voter roll-off is related to the presence of black candidates. The percentage of black roll-off voters declined as the number of black candidates increased. That roll-off also declined when an election involved a black incumbent supports our expectation that black incumbents have advantages that act to reduce the level of black voter roll-off that black non-incumbents lack.
Though the level of black voter roll-off diminished over time, time itself had no independent effect. The decline in the level of black voter roll-off over time had to do with changes in the context of these elections – the change from a majority-white to a majority-black city council, the increased number of black candidates, and the increased frequency of black incumbents – that occurred across several decades. More importantly, because time had no effect on black voter roll-off, a potential implication is that just because African Americans hold the majority of seats in a governing institution, such as a city council, we would not necessarily expect to witness a further decrease in the level of black voter roll-off in the future. We believe it is also worth noting that the concentration of black voters in the electorate had a positive, independent influence on greater ballot completion. This suggests that voters may be willing to more fully complete ballots when they believe that their racial group has sufficient numbers to influence the outcome of an election.
These findings provide a solid basis for future inquiry. The aggregate-level data that we employed provided us a valuable and somewhat rare opportunity – to identify patterns in the level of black voter roll-off across 46 election contests for city office over a 33 year period. We have endeavored to offer plausible theoretical explanations to support the relationships that were demonstrated by our findings. In offering these explanations, we have posited the presence of certain linkage variables: political efficacy as the link between institutional office-holding and black voter roll-off; election salience as the link between election type and roll-off. A future line of inquiry that can verify these theoretical explanations involves the use of individual-level data. Similarly, individual-level data may allow for a more direct demonstration of the influence of black candidates on black voters than is allowed by aggregate data.
We focused on roll-off among black voters because our research questions concerned the effect of black political empowerment on black political participation. An additional line of inquiry concerns roll-off among white voters. We did examine the possible impact of our explanatory variables on white voter roll-off. The results revealed that none of the models employed in our analysis explained white voter roll-off. In fact, compared to the level of roll-off among black voters, the level of white voter roll-off was smaller and less variable across these same council district election contests. [5] This may suggest the particular importance of a black urban political empowerment on influencing the level of political participation among blacks in local elections. Future research might investigate this issue more thoroughly. The determinants of white voter roll-off certainly deserve attention.
ENDNOTES
1. In 1975 the state of Louisiana adopted what is termed a bipartisan
primary. Regardless of party affiliation, all candidates for the same
office compete on the same ballot. If no candidate receives a vote
majority in the bipartisan primary, a runoff is held one month later between
the top two vote-getters. Elections prior to 1977 are Democratic Party
primaries that also necessitated a runoff if no candidate received a vote
majority. The mayoral contest is a true top-of-the-ballot election. City
elections are held on a date distinct from elections to state or federal
office. Council members hold a four-year, concurrent term of office.
Our analysis excludes uncontested district elections, district elections that were held without an accompanying top-of-the-ballot mayoral election, and at-large council elections. It is unlikely that uncontested council elections generated much of the vote, simply by virtue of being uncontested. We could not test for this, however. Data for the uncontested elections were unavailable. The following were the uncontested council district elections: District B, 1965; District D, 1977; District A, 1990; districts A and C, 1998. Due to a redistricting dispute and to special elections to fill vacancies, council elections without an accompanying top-of-the-ballot mayoral contest were held in 1976, 1980and 1981. Elections for the at-large council seats, that took place in every election year under investigation, were excluded because both at-large seats are up for election at the same time and voters can cast up to two ballots. Because of this, we cannot determine with sufficient confidence using our data the level of roll-off in the at-large contests.
2. The estimate of black voter roll-off is calculated as follows: the
value of the unstandardized regression coefficient multiplied by 100,added to
the value of the intercept. New Orleans’ large number of racially
homogeneous precincts (where one racial cohort comprises at least 90% of the
electorate) allows for confidence in our estimates. In 1965(the earliest
election year in our study) 29 were homogeneous black and 246were homogeneous
white (out of 397); in 1998 (the last year in our study) 130 were homogeneous
black and 76 were homogeneous white (out of 445).
3. Voters register by racial category, “black” or “white.” Since
1986 voters have had the option of designating “other.” The percentage of
registrants who chose this designation is small, typically no greater than five
percent of total registrants. Voter registration figures were obtained
from the Orleans Parish Registrar of Voters. Election returns were
obtained from the Orleans Parish Board of Supervisors of Elections. The
racial status of candidates was obtained from candidate profiles and election
coverage in the Times-Picayune, New Orleans’ local newspaper.
4. It could be asserted that the total number of candidates, and not
simply the number of black candidates, influences the level of black voter
roll-off. Similarly, it could be argued that the level of roll-off among
black voters is sensitive to incumbency per se, not solely to black
incumbency. The total number of candidates was highly correlated with the
number of black candidates (r=.781). When we replaced the number of black
candidates with the total number of candidates, the results were virtually
identical with those reported in Equation 1 of Table 3. In an attempt to
isolate the effect of candidates’ race, we entered into the model reported in
Equation 1 of Table 3 the number of white (i.e., non-black) candidates (the
bivariate association between the number of black, and the number of white
candidates was -.125, p=.408). The independent influence of the number of
white candidates was not statistically reliable (p=.259),while the independent influence
of each of the four hypothesis-related variables remained statistically
significant at the .05 level. These findings support our black candidate
mobilization explanation of black voter roll-off.
A variable measuring white incumbency (the 12 council contests that included a white incumbent were coded 1, the remainder were coded 0) was entered into the model reported in Equation 1of Table 1 (the bivariate association between black incumbency and white incumbency based on the Phi value was -.353, p=.017). The independent influence of white incumbency was not statistically reliable (p=.570), while the independent influence of each of the four hypothesis-related variables remained statistically significant at the .05 level. These findings support black incumbency as an explanation for black voter roll-off.
5. When the measure of overall roll-off is regressed on to the percentage
of registered voters who are black in each precinct, the regression intercept
is the estimate of roll-off among white voters (for white voter roll-off,
Mean=7.3, SD=6.7, Range=0.0-26.0).