Racial and Partisan Voting in a Tri-Ethnic City: The 1996 Dade County Mayoral Election
Abstract
Which is more important in vote choice: ethnicity or partisanship? Scholars have been asking this important question for decades regarding state, local, and national elections. We directly test this question in a metropolitan local election that constitutes a unique "natural experiment". In 1996, Dade County, Florida held a mayoral election with four major candidates whose partisan and ethnic interactions were not "normal": a Black Republican, two Hispanic Democrats, and one Hispanic Independent. In the runoff, a Black Republican challenged a Cuban-American Democrat in a county where over 80% of registered Black voters are Democrats, and over 60% of registered Hispanic voters are Republicans. As such, this election gives scholars a unique opportunity to untangle the effects of ethnicity and partisanship on vote choice. Using a three-wave survey of Dade County voters in 1996, we find that ethnicity was an overwhelmingly more powerful predictor of vote choice than partisanship. We assess the implications of how this study can be generalized to other multi-ethnic polities.
Racial and Partisan Voting in a Tri-Ethnic City: The 1996 Dade County Mayoral Election
The 1996 Mayoral election in Dade County, Florida dramatically illustrates the power and persistence of ethnic and racial voting in the United States. This election witnessed tens of thousands of Cuban-American Republicans crossing party lines to vote for a well-known Cuban Democrat at the same time that tens of thousands of African-American Democrats crossed party lines to vote for a prominent Black Republican. The significance of the Dade County Mayoral election lies in the fact that it ultimately pitted two candidates who had in the past used their partisan identification to appeal across ethnic-lines against each other, thus setting up a "natural experiment" in voting theory - which is more important in determining vote choice: ethnicity or partisanship? In a tri-ethnic polity such as Dade County, Florida, which factor is the more important determinant of vote choice - ethnicity or partisanship? Because of the reversal of stereotypical partisan-ethnic identity among the major candidates, the Dade County Mayor's race offers an ideal opportunity to answer this research question. Indeed, since more and more large urban areas are becoming multi-ethnic rather than simply black and white, this is a compelling question to answer for political as well as scholarly reasons. Spatial models of voting predict that voters will choose the candidate that is closest to them in some sort of space, whether it be issues (Downs, 1957), partisanship (Campbell et al., 1960), ethnicity, or some other factor. Based on a three-wave poll conducted in February, June, and August 1996 by Florida Scientific Survey, we propose to test the significance of partisanship versus ethnicity in their power to explain vote choice in this very important municipal election.
Both Alex Penelas and Art Teele won their at-large elections to the Dade County Commission in 1990 by creating broad coalitions across Dade's three major ethnic groups. Teele, a former Reagan Administration official, was able to easily beat Barbara Carey, an African-American Democrat, by using his Republican party membership to appeal to Cuban-American Republicans and conservative Anglos. Similarly, Penelas defeated Cuban Republican Jorge Valdez by using his Democratic Party affiliation to win support in the African-American community and among liberal non-Latin white voters.
The fact that neither of these cross-over candidates were able to expand their appeal beyond their own ethnic community in the 1996 Mayoral election serves as one of the clearest examples in recent years of a campaign and election in which ethnic and racial factors predominated over all others. It confirms the trends of ethnic bloc voting which have characterized Dade County elections since the early 1980s (Moreno and Rae, 1992; Moreno and Warren, 1992). If this assessment is correct, then ethnicity will likely remain the most important variable determining electoral outcome in south Florida for some time to come.
Hypotheses
When the candidate pool includes a co-ethnic of the voter, ethnicity is a more powerful predictor of vote choice than partisanship. Specifically:
1A: When the candidate pool includes more than one co-ethnic of the voter, and those candidates have different party affiliations, the voter will support the co-ethnic candidate who also shares the voter's partisanship.
1B: When the candidate pool includes a co-ethnic of the voter, but that candidate has a conflicting party affiliation to the voter's, the voter will still support that co-ethnic candidate.
When the candidate pool does not include a co-ethnic of the voter, he or she casts a vote for a co-partisan candidate.
When the candidate pool contains neither a co-ethnic nor a co-partisan of the voter, he or she votes randomly.
Our hypotheses start with the supposition that voters do cast votes based on a spatial model. Specifically, we assume that all else being equal, residents of Dade County will vote for a candidate who shares their ethnic identity. Further, we assume this hypothesis to be true even when a co-ethnic candidate has a party affiliation that is different from the voter's. For example, hypothesis 1A would predict that a Hispanic Republican, given the choice between a Hispanic Democratic and a Hispanic Republican candidate, would choose the Hispanic Republican candidate. This is a simple prediction, in that the voter has been given two mutually reinforcing vote cues: partisanship and ethnicity. Indeed, this seems an obvious and safe prediction. Still, previous voting research has had a difficult task in empirically verifying this assumption, since candidates of a particular ethnicity are often overwhelmingly from one party affiliation or another. In Dade County, for example, the great majority of Hispanic elected officials are Republicans, while a similar majority of African-American politicians are Democrats. Thus, the task of separating partisan and ethnic influences on voting remains difficult. This is why the 1996 Mayoral election is so compelling as a research subject: the candidate pool included a Black Republican, but no Black Democrats, and two Hispanic Democrats and one Hispanic Independent, but no Hispanic Republicans.
This unique nature of the mayoral candidate pool requires that we also specify hypothesis 1B. Given the choice between two candidates, one of whom shares the voter's ethnicity and the other of whom shares the voter's partisanship, we predict that the voter will choose the co-ethnic candidate, based on the powerful predictive effect of ethnicity proposed in hypothesis 1. For example, when a Black Democratic voter is given the choice between a Black Republican candidate and a Hispanic Democratic candidate, the African-American Democratic voter will choose the African-American candidate, even though that candidate represents another party.
Hypothesis 2 concedes that partisanship does have a significant effect on vote choice, but only when ethnic cues are absent for the voter. Here we predict that, when a voter is given a candidate pool that does not contain a co-ethnic, the voter will revert to partisan cues in determining his or her choice for Mayor. This hypothesis can be tested with the data at hand, since there were no Non-Latin-White ("Anglo") candidates in the race. So, we would predict that an Anglo Democrat, given no choice of a co-ethnic candidate, would choose a Democratic candidate for Mayor. For the sake of completeness we include Hypothesis 3, though we note that it is not directly testable here, since the candidate pool included Democratic, Independent, and Republican candidates. Therefore, no one was forced to vote "randomly" without partisan or ethnic cues.
Figure One graphically summarizes our three hypotheses.
Ethnicity and Voting
While a number of scholars have wrestled with the impact of ethnic and racial identification on voting behavior, the more specific body of literature addressing the interaction between partisanship and race and ethnicity (itself a subset of the latter) is severely limited. Indeed, it seems to be held captive to one debate: that between what might be called the 'assimilationists' and the 'mobilizationists' (Wolfinger, 1965). The position of the first, initially articulated by Robert Dahl in his classic Who Governs? (1961), is that as members of an ethnic group acquire economic and political power, the group will begin to fragment causing ethnic and racial ties to be weakened. Over the years, this argument has found its share of supporters and detractors. The latter take the view that ethnic voting persists even after a group has become acculturated(1)
.
One of the first to make this case was Wolfinger (1965), himself a collaborator in Dahl's original study. In his analysis of the Italians and the Irish in New Haven, Wolfinger concludes that the latter's voting behavior has been "relatively impervious to their changed social status." Indeed, Wolfinger's findings suggest that in some cases upward mobility has the effect of heightening rather than diminishing ethnic consciousness. Similarly, Italians continued to be faithful Republicans even though they constituted "the poorest segment of the white population." For Wolfinger, such seemingly non-rational behavior (in economic terms) could only be explained by taking up the issue of ethnicity.
In his seminal article, Wolfinger makes a couple of points in passing that are especially relevant to the present study. First, he notes the positive correlation between geographic proximity (i.e. the concentration of an ethnic group in a particular area) and ethnic consciousness. Secondly, he states that ethnicity is more likely to play a "greater role in non-partisan elections."
These observations ring true in the present context to the extent that, 1) There is a large (the largest in the country) concentration of Cuban-Americans in Dade County, and 2) the Mayor's race was non-partisan. But they must readily be qualified. First, Cubans are dispersed throughout the county. A considerable number live in the cities of Hialeah and Miami. Still others reside in the cities of Coral Gables, South Miami, Westchester, Miami Beach, Key Biscayne and unincorporated Dade County. Secondly, as we note elsewhere, though the election was non-partisan, voters were largely aware of the candidates party affiliation from partisan mailers as well as broadcast advertisements. In fact, the candidates themselves made partisan appeals when they deemed it convenient.
More recent contributions to the debate include the work of Cain and Kiewiet (1984). In their study of Mexican-Americans in California's 30th Congressional district, the authors concur with the prior findings of de la Garza (1977) that ethnic "bloc voting is hardly inevitable," echoing somewhat the assimilationist view. Yet this position continues to be hotly contested. Hero's work (1992), for instance, on the election and re-election of Federico Pena as mayor of Denver in 1983 and 1987 points in the direction of racial and ethnic voting. "[B]oth minorities and nonminorities tend to vote 'for their own'," he tells us.
In what is perhaps one of the latest efforts in this on-going problematic, Lee and Graves (1997) contend that ethnicity does not directly influence vote choice. Indeed, in examining the preferences of Latinos in the 1996 U.S. Senate race in Texas, they find little evidence of party cross over to support a co-ethnic.
The present piece parts company with some of these more recent studies in its attempt to show that Wolfinger's thesis continues to hold. That is, the influence of ethnicity on voting behavior remains a very powerful one.
The Context
The creation of the position of "Executive Mayor" was a directed response to the establishment of single member districts for the Dade County Commission. Before 1992 a commission of eight members governed Dade County along with a weak county Mayor, all elected countywide in nonpartisan elections for four-year terms, although they were required to reside within separate geographical districts. The Mayor's role within the commission was largely ceremonial apart from chairing Commission meetings. Executive authority over the county administration resided with the County Manager who was appointed by the commission.
The Dade Commission has sole authority over the unincorporated areas of the county, which due to rapid suburban development contains the majority of county residents. However, Dade still has thirty cities that control their own police forces, city property tax rates, and zoning. The cities may also exceed Metro's minimum standards in service provision in their own domain. The County Commission performs all the above services for unincorporated Dade, and in addition controls mass transit, the airport, the seaport, public health, and parks and recreation for the entire county. As mentioned above, the county also sets minimum standards that the municipalities have to meet in service provision. County government went through a major reform in 1992 when Dade was force to change from "at large" to single-member district elections. Due to a successful federal court action mounted by an alliance of Black and Latino leaders, the County Commission was compelled to change its mode of election. Judge Donald Graham also abolished the ceremonial position of county Mayor in his ruling in Meek v. Dade (Filkins, 1992). The change from "at-large" to "single-member" districts created a vacuum in Dade politics, as no commissioner represented the whole county. Alex Penelas, the youngest member of the County Commission, proposed to fill this vacuum by establishing the position of executive Mayor. Penelas argued that reforming Dade government to a "strong Mayor" system would serve two functions. First, it would give voters direct control over the executive branch of county government replacing a county manger responsible only to the County Commission with an executive Mayor directly responsible to all voters. Second, the strong Mayor would represent the whole county, not just a single district or a particular ethnic group. Penelas argued that "there is a perfect check and balance here; Commissioners will be more aware of local needs and we need to balance that with a Mayor's countywide perspective' (Strouse, 1992). Penelas was able to convince a skeptical commission to place his "strong Mayor" plan on the September, 1992 primary ballot. The proposal attracted the support of reformers who argued that if Dade planned to follow the example of New York and Chicago, then the county Mayor needed to be a full time executive. Two citizens' task forces recommended in favor of adding to the Mayor's power. One of them, headed by lobbyist Greg Borgognoni and including a dozen community activists including Commissioner Arthur Teele, recommended a plan virtually identical to Penelas'. Penelas was also supported by many Non-Latin White politicians who though that the "strong Mayor" position would probably go to an Anglo, since non-Latin Whites were still the largest voting bloc in Dade County. Then Dade Mayor Steve Clark, who was first elected to the job in 1970 and had served continuously since 1974, backed Penelas' plan for strong Mayor. He claimed that the job of running county government had become more complicated since the 1957 charter (Strouse, 1992) created the present Mayor's job.
The proposal for executive Mayor passed with little fanfare on October 1, 1992. Most Dade voters were too preoccupied with the devastation wreaked by Hurricane Andrew to pay much attention to county politics. Turnout in the October election was only 13 percent with the strong Mayor referendum passing overwhelmingly 57 percent to 43 percent. (Filkins, 1992a). The referendum gave the Mayor the power to veto ordinances passed by the County Commission, appoint the chair of the County Commission and all committee chairs, submit the budget, appoint the County Manager with the approval of the commission, and the day to day responsibility of running the huge county bureaucracy. The referendum also established that the office of Mayor would come into existence in October 1996. The Dade County Commission revisited the strong Mayor issue in the fall of 1995. Commissioners fearful of losing many of their prerogatives to the new position attempted to reduce mayoral power before the new office was even established. To gain political support for this they appointed a blue-ribbon committee of leading citizens on "the power of the executive Mayor." However, the only major reduction of Mayoral power recommended by the committee was to give the County Commission the prerogative to elect its own chair, the committee even recommending giving the Mayor more power over the appointment and firing of the county manager by requiring the manager to resign upon the election of a new Mayor.
Nevertheless, the Dade County Commission went ahead with three ballot referendums designed to significantly reduce mayoral power. Dade voters, in a direct slap to the commission, rejected in the March 12, 1996 election all initiatives that reduced the power of the Mayor. In fact, the only measure that passed was the one that increased mayoral power over the hiring and firing of the county manager. Dade voters made it clear that they wanted Dade's executive Mayor to be the most powerful elected office in the county.
The Pre-Campaign (November 1995 - June 1996)
Political observers believed that they were four probable candidates for Dade Mayor: County Commissioner Maurice Ferre who served as Mayor of Miami from 1978-1986, County Commissioner Alex Penelas, the Chair of the County Commission and Dade's most prominent African-American politician, Arthur Teele, and former city of Miami Mayor Xavier Suarez. Suarez, who was Mayor of Dade's largest city from 1986-1994, announced his candidacy for Mayor in November 1995, almost a year before the election. Taking advantage of his early announcement he held the largest fund-raiser in Dade history raising over $200,000 from 2500 contributors in January 1996. Table one summarizes some important characteristics of these major candidates, along with support for them in two time periods, and their final numbers in the September primary balloting.
Early polls confirmed Suarez role as the front-runner. A February 17, 1996 telephone survey of 454 likely Dade voters demonstrated that they were four viable candidates for Dade's Mayor: Maurice Ferre, Alex Penelas, Xavier Suarez, and Arthur Teele. The poll, while showing that the four top candidates were virtually tied for the lead, also indicated that Suarez was in the best position to win the election. Suarez had the highest name recognition (58.15 percent) and the highest favorable ratings of any of the candidates (41.85 percent). While Ferre's name recognition was only slightly less than Suarez's (57.71 percent), he had a much higher unfavorable rating (19.82 percent compared to 16.3 percent for Suarez). Table Two presents these data.
Alex Penelas scored significantly below both Ferre and Suarez in name recognition (43.17 percent), but he had a very high favorable rating (33.04 percent). Most important for Penelas, his ratio of favorable to unfavorable was the highest among the top four candidates with only 10.13 percent of the respondents giving him an unfavorable. Penelas' biggest advantage was that he had a solid base of support among Latino voters. Whereas Penelas' name recognition was only 43.17 percent countywide, among Hispanics he was known by 62.2 percent of the respondents. Penelas' support for Mayor was almost exclusively in the Hispanic community, with Latinos making up 74.19 percent of his supporters, Anglo consisting of 22.58 percent, and the survey failing to register any black support for his candidacy. At the beginning of the campaign Penelas lacked the broad appeal of either Suarez or Ferre, but he had a much stronger Latino base than either of them.
Teele had the lowest name recognition of the four major candidates, only 36.12 percent and his favorable rating was only 24.3 percent. Considering that he had been Chairman of the County Commission since 1993 this was an extremely poor showing. However, Teele did receive strong support from his co-ethnics. Teele captured 33.33 percent of the Black vote, 8.79 percent of the Anglo, and 6.1 percent of the Latino in the survey. Teele's biggest weakness was his poor name identification, and his almost total lack support among his fellow Republicans.
Only 5.8 percent of the GOP identifiers supported Teele in the survey. The poll also showed very little support for the only Anglo tested in the survey, former County Commissioner and Florida Commerce Secretary Charles Dusseau. Dusseau scored very poorly among voters with only 22.47 percent recognizing his name, and he was the only candidate to get more unfavorable (12.78 percent) than favorable (9.69 percent) ratings. Dusseau only received support from 4.41 percent of the voters questioned. His poor showing was surprising given that non-Latin whites were still the largest voting bloc in the county . Table Three presents ethnic and party registration breakdowns for Dade County registered voters in 1996. Given his poor showing in the polls Dusseau did not enter the mayoral race, leaving no Anglo candidates in the field (Branch, 1996a). Between March and July, despite his poor name recognition, Penelas was overwhelming his competition in fund-raising. By early July, Penelas raised $779,000 mainly from developers and insurance interests, more than twice as much of his nearest rival Teele (Tanfani, 1996b). Penelas justified his vigorous fund-raising by claiming "I haven't got the name recognition some of these other people have. Obviously if people are going to get to know Alex Penelas a little better and get to hear his message, I need money to do it" (Tanfani, 1996b). Trying to match Penelas dollar for dollar was Teele, who amassed $302,000 just two weeks after announcing his candidacy for Mayor. Teele, who had served in the Reagan administration, was counting on his nationwide network of both Republican and African-American friends to fund his campaign. Meanwhile, putative frontrunner Suarez was counting on organization and a network of volunteers to offset the monetary advantage enjoyed by his rivals. Suarez, who had raised $216,000 and had only $27,000 left by July, stubbornly refused to raise more than another $100,000 before the election. Instead, he placed all his hope on a network of neighborhood organizations he had knitted together.
The Campaign (July & August 1996)
Alex Penelas dominated the campaign airwaves for Dade county Mayor. From July until Election Day in early September, Penelas was able to control the campaign by building a coalition of Dade's county two principle political machines and by conducting the most extensive advertising blitz in the history of Dade campaigning. Penelas build his electoral base by combining his hometown machine centered in the predominately Cuban enclave of Hialeah with the Jewish retiree political machine that dominated politics in Northeast Dade. Thus, Penelas was able to build a powerful ethnic coalition of Jews and Cubans. Penelas gained significant momentum in the campaign when he secured the endorsement of fellow commissioner Gwen Margolis. Margolis, who herself was considering a possible Mayoral bid, had been discouraged from running by a poll she commissioned from Washington, D.C. based pollster Rob Schroth, which showed her with only 5 percent of the vote (Branch, 1996a). Margolis endorsed Penelas in exchange for his promise to name her Chairwoman of the County Commission if elected. The Margolis endorsement was an important achievement for the Penelas campaign because it assured him the support of the Margolis political machine. The so-called "condo commandos", mostly Jewish retirees living in the condominiums along Biscayne Bay, has been an important part of Dade county Democratic Party politics for over three decades. In the past, Democratic presidential, senatorial, and gubernatorial candidates have made pilgrimages to the condominiums of Northeast Dade to assure themselves the support of these
voters.
Penelas was also assured of the support of the Hialeah machine. Penelas began his political career in Hialeah, Dade's second largest city, as a city commissioner where he became a protege of Herman Echevveria, the powerful chairman of the City Commission. Echevveria and Hialeah Mayor Raul Martinez jointly controlled a machine that dominated politics in that working class Cuban city. In 1993 Penelas was re-elected to the County Commission from a Hialeah-based district. Although Martinez supported Ferre for Mayor, Penelas entered the Dade Mayoral campaign with strong support from his hometown and its political machine. Table Four summarizes the support of the four candidates in these two political "machine" areas of Dade County. The Hialeah Commissioner enjoyed a 14-percentage point lead over Art Teele, his closest August rival, in the non-machine areas of Dade County. Still, he also held a 14% lead over Teele in the North Dade machine area, and a commanding 38% leads over Maurice Ferre in Hialeah.
Although Penelas had put together a strong coalition, as late as June 27, 1996 he had been trailing in the polls. Table Five includes Mayoral vote preference from respondents to three surveys conducted in February, June, and late August 1996.
A Florida Scientific Survey poll conducted June 24-27 showed that Ferre and Suarez were locked in a neck-and-neck battle for Mayor. Indeed, in a sample of 600 registered voters, only four votes separated these two leading candidates. Teele and Penelas made up a "second tier," trailing the two leaders by approximately 4 to 5 percentage points. However, with four candidates separated by only 5.59 percent and with a large portion of the electorate still undecided, Penelas was still very much in the running. Moreover, since February Penelas had improved his name identification to 58 percent from 43.17 percent; this was still far behind Suarez and Ferre who each enjoyed 75 percent name recognition.
The Penelas campaign felt that they could easily make-up this deficit by launching an aggressive campaign advertisement blitz. Beginning in early August, the Penelas campaign began a million-dollar media campaign on television and on Hispanic radio. In August alone, Penelas spent over $750,000 on media (Finefroch, 1996). The strategy of the campaign was to take advantage of Penelas' low name recognition to introduce him to the citizens of Dade county. Penelas' TV ads had the desired affect. By August 18 a Miami Herald poll showed that he was the frontrunner (Finefroch, 1996). Penelas was the choice of 26.2 percent of likely voters polled, with Teele and Ferre receiving 19.5 and 17.4 percent, respectively. Xavier Suarez, the former frontrunner, was the choice of only 13.7 percent of likely voters. Moreover, Penelas' name recognition had ballooned to an impressive 70.4 percent in these seven weeks.
Penelas' new status as frontrunner was confirmed in a late August tracking poll conducted by Florida Scientific Survey one week before the September 3 primary. This poll of 642 likely voters, which forms the backbone of our multivariate analysis below, showed Penelas at 33.33 percent of the vote, with Teele running a distant second at 18.48 percent. The early frontrunners Ferre and Suarez were now witnessing their support collapse; they were receiving only 15.58 percent and 11.06 percent of the vote respectively. The dramatic increase in Penelas' support was due in large part to his overwhelming support in Dade's large Hispanic community. At the beginning of the campaign Penelas was receiving only 36 percent of the Hispanic vote in surveys, with the two other Hispanic candidates (Ferre and Suarez) each receiving 26.91 percent and Teele, the only non-Latin, trailing badly with about 10.18 percent of those polled. Table Six shows candidate support in August broken down by ethnicity.
By late August, Penelas' support among Hispanic voters had shot-up to a remarkable 57.47 percent, while Ferre's support was stagnant at 25.57 percent, and Suarez's support had dramatically dropped to 14.16 percent; Art Teele, the African-American candidate, was at a merger 2.68 percent of the Hispanic vote in polls.
As Penelas was clearly becoming the candidate of choice for Dade's Hispanic voters, Teele's second place finish in the August survey, and ultimately in the September election, was due largely to his overwhelming support in the Black community. At the beginning of the campaign Teele had the support of only 62.50 percent of African-American voters, with Ferre and Suarez receiving 22.12 percent and 13.46 percent among Black voters, respectively. But by August, Teele's support among his co-ethnics had grown to 82.09 percent, while both Ferre and Suarez had dropped below 10 percent among Black voters. While Teele was receiving overwhelming support among his fellow African-Americans, he had almost no support among fellow Republicans. Teele, the only Republican in the race, was polling only 7.05 percent among the GOP, while Penelas and Ferre, both lifelong Democrats, were receiving 58.15 percent and 20.70 of the GOP vote, respectively. Table Seven shows this seemingly anomalous trend.
Conversely, Teele was the choice of 41.54 percent of the Democrats, while Penelas, himself a Democrat, received the support of only 25.13 percent of the Democrats polled in August.
Penelas was able to ride his support in the Hispanic community to a first place finish in the September primary. In the subsequent October runoff with Art Teele, Penelas trounced his fellow County Commissioner by a 20-percentage point margin, receiving over 95% of the Hispanic vote, with Teele polling over 95% of the black vote.
The Power of Ethnicity over Partisanship in a Tri-Ethnic Polity
The proceeding separate analyses of ethnic and partisan influences on voting present a powerful argument: ethnic bloc voting was rampant in the Dade County Mayoral election. A full 97.32% of Hispanic voters reported support for one of the three Hispanic candidates in our August poll, while 82.9% of African-American respondents supported Black Republican Art Teele. Further, although partisanship was a significant predictor of voting, it was extraordinarily counterintuitive. Table Seven reported that the plurality of Democratic voters in Dade County supported the only Republican in the race, while an overwhelming majority (78.85%) of Republican voters supported one of the two Democratic candidates, and the plurality of Independents supported Alex Penelas, with only 18.99% supporting co-partisan Xavier Suarez. Of course one would be tempted to dismiss the partisan findings as simply functions of ethnicity. After all, one might argue, almost all Dade County Latinos are Republican, and nearly all Blacks in the county are Democrats. This would be a true statement, but it ignores the Anglo voters, and directly dodges the question of the power of ethnicity versus partisanship in predicting vote choice. This section falls back on our original three hypotheses, using a multivariate analysis to directly test the power of ethnicity versus partisanship in predicting vote choice in racially and politically polarized South Florida.
Applying our hypotheses to the Dade County Mayoral election's four major candidates, predicted vote choices and countervailing factors are:
Hispanic Democrats Penelas or Ferre No Dissonance
Hispanic Independent Suarez No Dissonance
Hispanic Republicans Penelas, Ferre, or Suarez Partisan Dissonance
Anglo Democrats Penelas or Ferre No Co-Ethnic
Anglo Independents Suarez No Co-Ethnic
Anglo Republicans Teele No Co-Ethnic
Black Democrats Teele Partisan Dissonance
Black Independents Teele Partisan Dissonance
Black Republicans Teele Partisan Dissonance
Table Eight presents the actual vote breakdowns for these partisan-ethnic combinations of voters. Unfortunately, given the very low number of African-American Independent and Republican voters in our August survey, this table lumps all Black voters into one category. Hispanic Democrats voted largely as hypothesized, with 86.28% of them voting for either Alex Penelas or Maurice Ferre, the two Hispanic Democrats in the race. However, only 17.86% of Hispanic Independents voted for co-ethnic and co-partisan Xavier Suarez, with Alex Penelas receiving majority support here. Hispanic Republicans, lacking a co-partisan, were hypothesized to vote for one of the three Latino candidates, and 96.7% did.
Anglo voters are, of course, a potential wildcard in this analysis, since there were no major Non-Latin White candidates in the race. Here, therefore, our model predicts Anglo voters will fall back on their partisan identities to choose a Mayoral candidate. Table Eight, however, indicates that Anglo voters did not follow the decision rules as presented in Figure One. Only 46.99% of Anglo Democrats supported one of the two Democratic candidates; a plurality actually supported Art Teele, the only Republican in the race. Likewise, a scant 16.67% of Anglo Independents reported supporting Xavier Suarez, with the great majority equally splitting their support between Democrat Alex Penelas and Republican Art Teele. Finally, Anglo Republicans absolutely abandoned partisan cues, with only 18.6% supporting Republican Teele, and a full 67.44% supporting one of the two Democrat (and Hispanic) candidates. Black voters overwhelmingly supported co-ethnic candidate Arthur Teele, giving him 82.09% support.
These rather curious findings, especially for Anglo voters, demand further analysis. Table Nine presents the results of a
Multinomial Logistic Regression (MNL) analysis of the August 1996 pre-election poll. Here, our dependent variable is vote
choice, and is coded 1=Penelas, 2=Suarez, 3=Ferre, and 4=Teele. Undecided voters were excluded from the analysis, leaving
500 usable cases. Since the dependent variable is measured at the nominal level, neither OLS regression, dichotomous logistic
regression, nor ordered logit are acceptable statistical modeling methods.
MNL has the advantage of producing coefficients that have an interpretation similar to any logit model. That is, for a one unit change in an independent variable, there is a I change in the logged odds of choosing one alternative over another in the dependent variable. Multinomial logistic regression presents k-1 equations, where k is the number of categories of the dependent variable. One of the categories of the dependent variable is fixed, and thus the equations are interpreted as the logged odds of a respondent choosing each k alternatives over the fixed one. In this case, voting for Alex Penelas is the fixed category, so the three equations presented in Table Nine are the logged odds of choosing Suarez over Penelas, Ferre over Penelas, and Teele over Penelas. So, a negative coefficient for an independent variable indicates increasing likelihood of voting for Penelas over the alternatives, and a positive coefficient indicates an increasing likelihood of voting for the alternative over Penelas. Statistically, which category of the dependent variable to fix is arbitrary, but we chose Penelas because he received the most support in our August poll and ultimately won the election.
All the independent variables included in the model are dummy variables, so the interpretation of the MNL coefficients is straightforward: a voter having the characteristics of the independent variable is ln(I ) more likely to vote for the alternative candidate over Alex Penelas. Included are dummy variables for voters who are: Hispanic Democrats, Hispanic Republicans, residents of the Hialeah or North Dade "machine" areas, Spanish speakers(2), Anglo Democrats, Anglo Republicans, and Blacks. The omitted categories were non-Black Independents, and the constant represents this omitted category.
Column One of Table Nine is the equation for the likelihood of voting for Xavier Suarez (a Hispanic Independent) over Alex Penelas (a Hispanic Democrat). The only significant predictor of vote choice in this case is for Blacks, who are significantly more likely to vote for Suarez than Penelas. Our hypotheses predicted that there should also be significant differences for Hispanic Democrats (predicted to vote for Penelas), Anglo Democrats (predicted to vote for Penelas), and Independents (predicted to vote for Suarez), but these variables show no statistical significance.
Column Two, the choice between Maurice Ferre (a Hispanic Democrat) and Penelas, further indicates that Blacks are more likely to vote for Ferre than Penelas. Also, the constant is statistically significant, in this case representing the fact that Independent Anglos and Hispanics are more likely to vote for Penelas than Ferre. This relationship should be insignificant according to Hypothesis 1. Lastly, Column Three shows that both Blacks and Anglo Democrats were significantly more likely to support Black Republican Art Teele over Alex Penelas, while Hispanic Democrats and Republicans significantly supported Penelas over his rival. All of these relationships were predicted by Hypothesis 1, with the exception of the Anglo Democrats, who partisan cues hypothetically should have led them to support Alex Penelas. Further, Anglo Republicans were hypothesized to support their co-partisan Teele, but the coefficient here is not statistically significant.
Overall, the MNL model fits the data fairly well, with 54.41% classified correctly, much better than either a naïve model of 25% or a model which would have predicted everyone voted for Penelas (42.54%). In the end, this model is more suited to our hypotheses than the contingency tables presented earlier, in that the effects of ethnicity and partisanship on voting can be tested simultaneously. The reader can see very little support in either these models or in the earlier tables for any rational partisan basis to voting in the Dade County Mayoral election. In fact, some of the significant partisan differences are counterintuitive: Anglo Democrats supporting a Black Republican over an Hispanic Democrat, and Hispanic Independents supporting one Hispanic Democrat over another. Still, just as with raw logit coefficients, MNL coefficients can be a bit difficult to interpret. Fortunately, MNL shares a desirable feature with the other maximum likelihood categorical procedures such as logistic regression: the coefficients can be exponentiated to recover probabilities of voting for one candidate over the other, holding the other independent variables constant. Table Ten presents these probabilities for every possible two-candidate combination. These odds are recovered from the original MNL model.
Starting in Column One, our model predicts that an African-American voter is 9.97 times more likely to vote for Xavier Suarez than Alex Penelas. This gap between black support for Penelas and other candidates continues, with Blacks being an astounding 82.27 times more likely to support co-ethnic Art Teele than Penelas. The next set of odds is for Spanish-speaking Hispanic Democrats and Republicans living in Hialeah. This combination was chosen because it is the group of people most likely to support Alex Penelas, the Hispanic County Commissioner from the Hialeah machine area. Here, one sees that both Democrats and Republicans are 0.27 times more likely to support Suarez than Penelas. In other words, these people are almost four times more likely to support Penelas. This gap is different in the Penelas/Ferre choice combination. Here, Hispanic Hialeah Democrats are 1.73 times more likely to support Democrat Maurice Ferre than Democrat Penelas. With the Penelas/Teele combination, both Democrats and Republicans in this group are over 28 times more likely to support Penelas over Black Republican Art Teele (odds of .03 and .04, respectively). Interestingly, that gap between support for fellow Hispanics and support for Teele continues for the Suarez/Teele and Ferre/Teele combinations, indicating very little partisan difference in vote preference for Spanish-speaking Hispanics living in Hialeah - race is clearly the deciding factor in these people's vote choices. The next set of coefficients is for Hispanic Democrats and Republicans who do not live in Hialeah and who answered out August survey in English. Why break out the Latino
population in this way? First, we want to see some of the combined effects of living in a machine area (Hialeah) and being a predominant speaker of English versus Spanish. There is a good reason to suspect that these are simply two different groups of voters: working class, recent immigrants living in Hialeah, versus more upscale, more assimilated Cubans and other Hispanics living in more affluent and integrated areas of the county. Clearly, according to the coefficients in Table Ten, there are some differences in voting behavior between these groups of people. First, the odds of voting for Suarez versus Penelas are not as dramatic for non-Hialeah English-speaking Hispanics. Still, however, gap between the likelihood of voting for Art Teele versus the three Hispanic candidates remains large for non-Hialeah Latinos; these people are still nearly ten times more likely to support Alex Penelas over Black Republican Art Teele, and the pattern is very similar for the Teele/Suarez and Teele/Ferre combinations. On balance, then, there is very little support for an argument that, when given a choice between a black candidate and various Hispanic candidates, Hispanics will deviate from ethnic bloc voting and use partisanship as a major voting cue. Hispanic Republicans, no matter where they live and what language they speak, are more than willing to vote against a fellow Republican and vote for a fellow Hispanic, even though that candidate may be a Democrat.
The bottom half of Table Ten pays attention to the voting probabilities of various groups of Anglo voters. The first group here is Anglo Democrats living in North Dade - the classic "condo commando" predominantly Jewish retirees. There is very little predictive power here, except in the Suarez/Teele and Ferre/Teele vote choice pairs. Here, Anglo residents of the Jewish machine area of the county are about twice as likely to vote for Art Teele - a Black Republican - as they are to vote for Xavier Suarez (a Hispanic Independent) or Maurice Ferre (a Hispanic Democrat). Further, they are 1.62 times more likely to support Teele over Penelas, even though the North Dade machine endorsed the latter. Clearly, the machine has little effect on vote choice, al else being equal. Further, these White Democrats were more willing to vote for a Black Republican than any of the three Hispanic candidates, two of whom were fellow Democrats. Neither race nor partisanship, then, predicted voting behavior in the hypothesized manner for these people. Anglo Democrats in the rest of the county followed similarly murky patterns in their vote choice probabilities, though they seemed to support Maurice Ferre over his fellow Hispanic Democrat Alex Penelas.
Finally, what of Anglo Republicans? According to Hypothesis Two, they should be very likely to support Art Teele over any of the other candidates, since he was a fellow Republican, though not a co-ethnic. No such pattern arises. In fact, Anglo Republicans are marginally less likely to support Teele that Democrat Alex Penelas, and there is simply no difference in their predicted probabilities of supporting Teele over the other two candidates. Apparently, the Anglo Republican vote followed no discernible racial or partisan pattern in the Mayoral Election.
Conclusion: The Overwhelming Dominance of Ethnic over Partisan Voting
This paper hypothesized that when ethnicity and partisanship are considered as independent variables in the prediction of vote choice, ethnicity will be the more powerful predictor. However, we also specifically proposed that partisanship should have a significant effect on vote choice, but only in the absence of ethnic voting cues. We proposed that Hispanics would vote for Hispanic candidates, regardless of partisanship, and that African-Americans would vote for Black candidates, even if they were Republicans. Our analysis finds very little support for this dual influence of ethnicity and partisanship on voting, with even a minor secondary role for partisanship in predicting vote choice. According to our analyses, Hispanic Republicans were perfectly content to vote for Hispanic Democrats over Black Republicans, while Black Democrats were equally comfortable with supporting Black Republicans. Partisanship, even when it introduced dissonance into one's vote choice (e.g. Blacks voting overwhelmingly for a Republican candidate for Mayor), had absolutely no independent effect on voting.
Further, Non-Latin White voters, who did not have the opportunity to vote for a co-ethnic candidate, still had the ability to rely on partisan cues, since the race contained two Democrats, an Independent, and a Republican. However, according to both our contingency tables and our MNL model, partisanship had few effects on Anglo vote choice, and where there were significant effects, they were reversed from hypothetical expectations, with Anglo Democrats significantly more supportive of a Republican (Teele) than a Democrat (Penelas). On top of this, over 80% of all Anglo Republicans reported an intention to vote against the only Republican in the race!
What is going on in Dade County? Is there anyway that one can avoid the conclusion that the 1996 County Mayor election was about anything but ethnicity. In a word, no. Absolutely not. This election was about nothing but ethnicity. The one place where one could actually find significant partisan differences that could not be attributed to ethnicity - among Anglo voters - showed counter hypothetical findings, with Anglo partisans supporting candidates from the "wrong" party. Even here, one must speculate that there is an element of unmeasured ethnic voting by Dade County Anglos. A plurality of Anglo Democrats supported the Black Republican Art Teele. Many Anglo Democrats in the county are North Dade Jewish retirees. One could speculate that these voters simply could not stomach the idea of a Hispanic Mayor, which they may have seen as yet another element of Hispanic domination in Dade politics, which until recently had been dominated by Democratic Jewish politicians.(3) On the other hand, over 80 percent of Anglo Republicans voted for a candidate besides Art Teele, the only Republican in the race. Again speculating, perhaps these Republicans saw a Hispanic Mayor as preferable to a Black one. These questions must remain unanswered here, since our current data offer no way to directly test them empirically.
What can the Dade County Mayor's race tell us about politics in general, and urban voting in particular? Can we learn anything of a generalizable nature, or is Dade County so hopelessly unique that the findings here must remain esoteric? To be sure, ethnic bloc voting is a feature of American politics at almost every level. The demographics of this country are changing, and nowhere is this more evident than in our largest cities. The old black versus white polarization still exists. But increasing Hispanic populations in almost all large cities, and a larger and larger Asian presence in a few others, must surely hint to us that the Dade County Mayor's election of 1996 will probably be repeated in similar form elsewhere. There may not be many elections anytime soon that mirror the unique partisan and ethnic "head standing" of this South Florida election. After all, one cannot find many examples of Black Republicans or Cuban Democrats running against each other. Still, as the political and ethnic diversity of the United States increases, races that pair off all possible partisan-ethnic candidate combinations will become more and more common.
References
Branch, Karen. 1996a. "Dade Politics." Miami Herald, June 24, 1996.
Branch, Karen. 1996b. "Pivotal Election May Pull Puny 36%." Miami Herald, September 3, 1996.
Cain, Bruce E., and D. Roderick Kiewiet. 1984. "Ethnicty and Electoral Choice: Mexican American Voting Behavior in the California 30th Congressional District." Social Science Quarterly. 65:315-27.
Campbell, Angus, Phillip Converse, Warren Miller, & Donald Stokes. 1960. The American Voter. Ann Arbor: University of Michigan Press.
Dahl, Robert A. 1961. Who Governs? Democracy and Power in an American City. New Haven, CT: Yale Univerisity Press.
De la Garza, Rudolph O. 1977. "Mexican-American Voters: A Responsible Electorate." Mexican-Americans: Political Power, Influence or Resource. Lubbock: Texas Tech Press. 63-76.
Downs, Anthony. 1957. An Economic Theory of Democracy.
Fiedler, Tom. 1996. "To Win, Teele Must Climb an Even Higher Mountain." Miami Herald, September 4, 1996, 1A.
Filkins, Dexter. 1992a. "Dade Strong-Mayor Plan Wins Easily on Third Try." Miami Herald, October 2, 1992, 1A.
Filkins, Dexter. 1992b. "Districting Dispute May Kill County Mayor Post." Miami Herald, December 22, 1992, 1B.
Finefrock, Don. 1996. "Penelas Emerges as Front-runner, but Runoff is Likely in Race for Metro-Dade Mayor." Miami Herald, August 20, 1996, 1A.
Hero, Rodney. 1992. "The Elections of Federico Pena." Latinos and the U.S. Political System: Two-Tiered Pluralism. Philidelphia, PA: Temple University Press. 116-130.
Jongho Lee & Scott Graves. 1997. "Ethnicity and Voting: The Case of the 1996 U.S. Senate Election in Texas." Paper presented at the annual meeting of the Midwest Political Science Association, Chicago, IL April 10-12, 1997.
Moreno, Dario and Christopher Warren. 1996 "Conservative Enclave Revisited: Cuban-Americans in Florida." in Rodolfo O. De la Garza and Louis DeSipio, Ethnic Ironies: Latino Politics in the 1992 Elections. Boulder, Co: Westview Press.
Moreno, Dario and Christopher Warren. 1992. "The Conservative Enclave: Cubans in Florida." in Rodolfo O. De la Garza and Louis DeSipio From Rhetoric to Reality: Latino Politics in the 1988 Election. Boulder, Co: Westview Press.
Moreno, Dario and Nicol Rae. 1992. "Ethnicity and Partnership: The Eighteenth Congressional District in Miami." in Guillermo Grenier and Alex Stepick, Miami Now. Gainesville: University Press of Florida.
Strouse, Charles. 1992 "Does Dade Mayor Need More Power? Dade Voters to Decide Thursday." Miami Herald, September 22, 1992, 1B.
Tanfani, Joseph. 1996a. "The Battle Ahead: Penelas vs. Teele: Old Rivals Face Off in Round 2." Miami Herald, September 4, 1996, 1A.
Tanfani, Joseph, Karen Branch, and Don Finefrock. 1996a. "At the Peak of Power: Penelas." Miami Herald, October 2, 1996, 1A.
Tanfani, Joseph. 1996b. "Penelas Has Huge Lead in Race for Cash." Miami Herald, July 11, 1996, 1B.
Wolfinger, Raymond. 1965. "The Development and Persistence of Ethnic Voting." American Political Science Review. 59:896-908.
Return to Table of Contents
1. 0 This term, while applicable in the case of immigrants, seems a bit more problematic when dealing with African-Americans. This fact may be a function of the differences in their respective experiences. As such, it reminds us of the methodological difficulties which arise when race and ethnicity are lumped together in the same category.
2. 0 Because of the ethnic breakdown of Dade County, we utilized interviewers who were fully bilingual. If a household was called and the person answering the phone offered a salutation in Spanish, the interview was conducted in Spanish. Approximately 38% of the interviews with Hispanic voters were conducted in Spanish. 0 Interestingly, the one major area of the county still controlled by Jewish Democrats is the City of Miami Beach. The Miami Beach City Commission, which is the last major legislature in the county elected in at-large districts, is 100% Jewish Democrat; all of the seats are held by Jewish Democratic Commissioners, even though the City is actually majority Hispanic.