ó Would You Vote For Colin Powell? White Support for a Not Quite Colorless Black Candidate* April Gresham Villanova, P A Robert Maranto Villanova University Sigelman et al (1995), Williams (1990), and Sonenshein (1990) find that whites stereotype African-American candidates as less competent and more lib­ eral than white candidates. Yet increasing numbers of African-Americans win election in white constituencies. We used a natural experiment, the media boom • for a Colin Powell presidential bid in 1994, to test the degree to which various race-related attitudes affected Powell support among white voters. We con­ dueled a phone 5urvey of 350 while,Perm::;yluaniu resitlentt; from June to No­ vember 1994. Powell had as much support as any Republican. Powell support was unrelated to conservatism, party, and affirmative action. For Democrats, Powell support was related to attitudes toward NAFTA and immigration. For Republicans, Powell support was related to authoritarianism and populism. We conclude with speculation about future directions for research. Introduction and Hypotheses Schuman, Steeh, and Bobo (1988 73-81) and Thernstrom and Thernstrom (1997) report that the percentage of whites willing to vote for an African-American presidential candidate rose sharply from the late 1950s to the early 1970s, and slowly thereafter. By the early 1980s, more than 80% of whites expressed willingness to vote for a black presi­ dential candidate. Further, some African-Americans have won election to overwhelmingly white constituencies. These include former Senator Carol Moseley Braun (D, IL), former Virginia Governor L. Douglas Wilder, former U.S. Representative Alan Wheat (D, Ma, who lost a 1994 Senate race), former U.S. House member Gary Franks (R, CT) and current U.S. Representative J.e. Watts (R, OK)(Barone and Ujifusa 1993). Yet black candidates in white constituencies still face obstacles. Vot­ ers use personal information about candidates, including ethnicity, to make presumptions about political views (Popkin 1992, 63-64, 77-79). As Williams (1990) finds, national surveys show that whites assume black candidates to have certain generic racial attributes, among them liberal- ... We wish to thank Lafayette College and Villanova University for material sup­ port of this pwject, Paul Snidennan fur hdping U~::;igll mca:surement :strategies, and Tali Mendelberg and several anonymous reviewers for their comments. The usual caveats apply. 53 ism and relative incompetence in office. Using experimental manipula­ tions of the race and ideology of hypothetical candidates, Sigelman et al (1995) show that whites assume even conservative African-American candidates to be less competent and more sympathetic to disadvantaged groups than comparable white candidates. Successful African-Ameri­ can politicians in white constituencies have overcome stereotypes in part by empha~izing conservative stands or social characteristics. For example, former Los Angeles Mayor Tom Bradley stressed his background as a policeman in early races and his business support in later ones (Sonenshein 1990). Similarly, former Massachusetts Senator Edward Brooke and Virginia Governor L. Douglas Wilder emphasized their tough stands against crime. Wilder also stressed his southern style and status as a military veteran (Sonenshein 1990,1993). A natural replication of such research was made possible by the 1994- 95 media boom about a hypothetical presidential bid by former Joint Chiefs of Staff Chairman Colin Powell, the first ~frican-American (tech­ nically Caribbean American) to hold the post. Though he did not de­ clare his Republican affiliation until October 1995 and had contributed campaign money to Democrats and Republicans alike, Powell's military career flourished under Republican preSidents, and he was widely as­ sumed to be Republican. Powell's military career appealed more to Re­ publicans than to Democrats, as did his frequent talk on religious issues and the need for strong families. Powell was popular among white reli­ gious conservatives, as shown by the Powell biographies frequently fea­ tured at religiOUS bookstores (Interview 1; Woodward 1991). Powell was a particularly interesting candidate to examine since pre­ vious black presidential candidates Gesse Jackson, Shirley Chisholm) were liberals, while Powell was thought to be moderate or conservative. In 1994-1995 draft Powell organizations formed, one led by historian and Eisenhower biographer Stephen Ambrose. As Ambrose said, "when people look at him [Powell], it makes you proud to be an American ... because, like Ike, he makes you believe in the American dream" (Elliason 1995). A Times-Mirror survey conducted in August 1994 showed a Republican Powell beating Clinton by 51 to 41 %. The survey also found a 74% Powell approval rating and 8% disapproval. Powell himself encouraged presidential speculation, stating, liThe Chairman of the JCS is the best job in the world, except for maybe one or two others,"l Notably, Powell's perceived success in Operation Desert Storm and the status of the military might well inoculate him against racial stereotypes of incompetence (Elliason, II All Things Considered," March 31, 1995; Interview 1; Raum 1994). Similarly, African-American Republicans in general, and former generals in particular, might be immune from ste­ reotypes that they are liberal. We thus propose to test: 54 HI: A Powell presidential candiitacy could gaill considerable support among whites. This is not to say that a Powell candidacy would be race neutral. \Ve will explore the possible impact of three different types of racism on voting for Powell: modern or symbolic racism, also referred to as encoded rac­ ism; blatant authoritari~n racism; and populist racism based in realistic group conflict. Modern racism researchers contend that many whites pretend color­ blindness while embracing policies that harm African-Americans; thus much of what passes for conservatism is in fact disguised, or "modern" racism (McConahay 1986; Sears 1988). Hagen (1995) and Bloom (1995) investigate "encoded racism" or "explainable racism/' racism expressed by code phrases such as opposition to welfare and concern about crime. Their preliminary findings indicate that educated voters in particular express such subtle views and vote prejudicially. One way to explore this type of racism is by examining the relation­ ship between views of Powell's candidacy and affirmative action. Affir­ mative action could be seen as a racial issue--one of coded or explain­ able racism (Bloom 1995). If so, support for Powell might be inversely related to affirmative action opposition. On the other hand, Powell had not as of 19941inked himself with affirmative action and seemed conser­ vativ@ on nthpr iSSllPs. Thw;, WI" propnse: H2: Views of affirmative action will not be related to white support for a Powell candidacy. A second type of racism is noted by Sniderman and Piazza (1993), who find that many whites still admit blatantly racist views. They also find that such racism is not strongly related to ideological conservatism but rather to a long-standing personality correlate of prejudice­ authoritarianism (Allport 1954). Notably, this minority of white Ameri­ cans holds authoritarian views on such issues as maintaining respect for America's power, strengthening law and order, making sure all people acquire similar values and ideals, and teaching children respect for au­ thority. Authoritarians are not conservatives. They endorse conventional conservative political views only on law and order issues, and unlike other conservatives they negatively stereotype Jews (a group that stereotypically embodies the Protestant Work Ethic) as well as African­ Americans. In short, authoritarian opinion denigrates all other races (Sniderman and Piazza 1993, 55, ch. 2, 3; Uslaner 2002). Thus, we pre­ dict: 55 H3: Powell support will be inversely related to authoritarianism. A third type of prejudice is realistic group conflict. According to realistic group conflict theory, a stagnant or unstable economy is associated with height­ ened intergroup tensions as groups compete for scarce resources (Bobo 1988; Hofstadter 1964; Fritzsche 1990; Uslaner 2002). Immigrants are prime targets of prejudice since they represent competitor groups. According to this theory, vot­ ers see wealth as finite. Thus, Americans compete with non-citizens (immi­ grants) for jobs in America, and American workers compete with workers in other countries (such as Mexico) in the world market. Such competition is ass0- ciated with prejudice against members of the competing group(s). Two political issues at the time of the survey lent themselves to inves­ tigating realistic group conflict based prejudice: NAFTA and immigra­ tion. These issues highlight a fault line among Republicans. "Optimist" conservatives such as Jack Kemp argue that free trade and immigration help all Americans; they thus support both NAFTA and immigration. In contrast, "Nationalist" conservatives such as, Pat Buchanan see economic growth as a zero-sum enterprise: immigration and NAPTA take away II American" jobs. More important, immigration could destroy America's European based culture, and increase crime and disorder. As Buchanan, who other conservatives accuse of racism and anti-Semitism, asks, I/[w]hose country is it, anyway?" (Frum 1994; Lind 1993; Buckley 1992). (On the relationship between optimism and tolerance, see Uslaner 2002.) The contrast between Nationalists and Optimists was especially pro­ nounced in comparisons of the Kemp and Buchanan 1996 presidential campaigns. Kemp's standard stump speech urged the GOP to be the party of Lincoln, a Biblical good shepherd for every sheep. He noted that unemployment in American ghettoes was so high as to justify not only riots, but also "the French Revolution, the Russian Revolution" (C­ SPAN 1992). He saw capitalism as so good that the poor needed to be part of it. He typically urged his (usually all white) audiences to remem­ ber that blacks had the same aspirations for their children. Once, when a hand-shaker urged Kemp to consider Bob Dole for Vice-President, he answered "How about Colin Powell?" (C-SPAN 1992). Republican ac­ tivists in Virginia in the months after the 1992 election distributed Kemp­ Powell buttons (C-SPAN 1992; Interview 1). Kemp campaigned against the anti-illegal immigrant Proposition 187 in California and dropped out of the GOP presidential race after it passed. In sharp contrast, Pat Buchanan's 1995 announcement for president denounced moral decay, the crimes of illegal aliens, and the greed of such (Jewish named) compa­ nies as Goldman-Sachs that gave executives huge bonuses while laying off 2,000 clerks (C-SPAN 1995). Accordingly, we predict that: 56 H4: Support for NAFTA and immigration will be directly related to support jar Powell. Optimists and Nationalists can be likened to the libertarian and popu­ list currents in mass ideology described by Maddox and Lilie (1984). Like Optimists, libertarians favor less government regulation in both the eco­ nomic and social-cultural spheres of life. Libertarians also tend to be younger, better educated, and relatively optimistic. In contrast popu­ lists are pessimistic about human nature and want a strong government, both to control business and social morality and to protect the working class. Populists are older, poorer, less educated, more skeptical about programs aiding African-Americans, and less trusting generally (Maddox and Lilie 1984; Uslaner 2002). . Maddox and Lilie suggest that as of 1982, populists and libertarians could be found in substantial numbers in each party, unlike liberals and conselvatives. Suggesthl.g thal populist-nationalist Dem.ocrats are Inore likely to believe in social control,. to stereotype others, and to hold rela­ tively authoritarian views may explain the surprising findings of Sniderman et al (1993) that stereotypes about African-Americans were strongly re­ lated to welfare policy views for economic liberals, but not for economic conservatives. While conservatives oppose assistance for anyone, high ste­ reotyping liberals oppose assistance targeted to blacks. Perhaps these sup­ posed "liberals" are in fact the populist-nationalists discussed above: eco­ nomically liberal but socially conservative or even authoritarian. Thus: H5: Support for Powell will be inversely related to populism. Procedure Sniderman and Piazza's findings are based on California and Ken­ tucky surveys focusing on racial issues. We have partly replicated Sniderman and Piazza's measures of racial stereotyping and authoritar­ ian values using two surveys. Our Racism survey focuses on racial ste­ reotyping and affirmative action, though it also includes a few general questions about political attitudes. Our Politics survey uses an embed­ ded design in which a few race-oriented questions are embedded within a broader survey of political attitudes. The following analyses use ques­ tions from the Politics survey, except tests of HI (Tables 2-3), which pool the data sets.2 We conducted a phone survey of Pennsylvania's 15th U.S. House dis­ trict (the Lehigh Valley and surrounding area) from June 15 to Novem­ ber 30, 1994. Respondents were normally called three times before being counted as refusals or four times before being categorized as unreach­ able. The household and household member to be interviewed were selected randomly (Lavrakos 1992). Of the 753 households contacted, 55 57 were unable to participate due to inability to speak English or poor hear­ ing. Three hundred and seventy of the remaining 698 potential subjects completed the survey, for a 53% completion rate. Of the 370 individuals surveyed, 350 (94%) are included in the analyses that follow.3 Of the 350, 187 completed the Politics survey and 163 completed the Racism survey. As Table 1 shows, sample demographics resembled those for the 15th District as a whole as reported from Barone and Ujifusa (1993,1113). Table 1 Comparing lhe Sample with the 15th U.S. House District Attributes Sample 15th District Median Household Income $30-45,000 $33,049 Percent White 94 92 Percent Over Age 64 16 15 Percent W / College Degree 36 37 Percent, Who, in 1992, Voted for Clinton-Bush-Perot 41-43-16 41-37-22 Forty percent of survey respondents report being Democrats and 38% call themselves Republican. Official registration figures for the district show Democrats with 56% of the two party registration total, though registered independents tend to think of themselves as Republicans. This is particularly true of the many recent immigrants from New Jersey, an open primary state. Many of the newcomers came to escape high taxes, high housing costs, and crime. In addition, as explained below, Lehigh Valley Democrats increasingly vote Republican (Interview 2). The Lehigh Valley is an interesting area in which to study the inter­ section of race, class, and politics. The area is overwhelmingly white but ethnically diverse. It was originally settled by British and German im­ migrants whose descendants still make up more than half the popula­ tion. In the late 19th and early 20th centuries came waves of Irish, Ital­ ian, . and Eastern European immigrants. Survey respondents are 36% Catholic, 55% Protestant, and 1% Jewish. Immigrants worked in such large manufacturing industries as Bethlehem Steel, Dixie Cup, and Mack Trucks. These heavy industries hit hard times in the study period, mak­ ing NAFTA a significant political issue. Despite economic change, job­ lessness was near the national average, with new industries moving in from New York and New Jersey. In addition, many factory workers (or their children) moved down Interstates 78 and 81 to the Sun Belt. Still, 58 the Valley remained heavily blue collar and 25% rural, with politically significant unions, gun and sportsmen's groups, und ethnic clubs. The district had 78,000 veterans, and military issues often figured in local politics. Then US. Representative Paul McHale, a Democrat, upset Re­ publican incumbent Don Ritter in 1992 in part by emphasizing his mili­ tary record. In its heavy industry days the district elected Democratic US. House members from 1932 to 1978, when Ritter upset a longtime incumbent. Later congressional races were closely contested. The dis­ trict was Democratic in state and local elections through the 1970s, but since then it has trended Republican because of Cold War issues, eco­ nomic changes, and migration from New Jersey. While Hubert Humphrey beat Richard Nixon by a 50% to 46% margin in 1968, George McGovern could manage only 40% in 1972. Jimmy Carter won the district with 52% in 1976, but Ronald Reagan beat Carter by 50% to 39% in 1980. George Bush won 55% in 1988, but in a bad local economy in 1992 he lost to Clinton 41 % to 37%, with Perot winning 22%.4 At the same time, from 1982 to 1994 the GOP narrowed 'the two party registration edge from 63% to 37% to 56% to 44%. Many of the 15th's Democrats are considered "God and country" or "Reagan Democrats" who vote Republican nationally on cultural and defense issues but have ethnic and economic links to the Democratic party and tend Democratic for local offices. Al­ though Democrats are a plurality in our survey, 51 % of respondents con­ sider themselves conservative-35% liberal. Indeed, the presence of so many Reagan Democrats led the Clinton campaign to conduct focus groups in Allentown before the 1992 Democratic convention.5 Despite the local losses for Bush and U.S. Representative Ritter in 1992, the GOP continued its gains in the area. In 1986 Democrats held 14 of the Lehigh Valley's 15 state house seats, and all three major city mayor's offices (Allentown, Bethlehem, Easton). As of 1994, GOP mayors ran the three cities and Republicans held 12 of 15 house seats (Barone and Ujifusa, 1975, 1981, 1983, 1989, 1993; Interview 2). While economic and national issues seemingly dominated area voting behavior, race also may have had an impact. Overwhelmingly white, the Lehigh Valley has grown more diverse. Hispanics, African-Americans, and A~ian Americans made up 8% of the population in the study period, and they were even more in the three major cities. Downtown Allentown and Easton were widely perceived as unsafe because of ethnic minorities. '!he dominant local newspaper gave substantial coverage (often includ­ ing photos) to crimes committed by ethnic minorities and to moves to make Allentown an "English only" city (Lehigh Valley Morning Call 1995). In addition, most Valley residents watch New York and Philadelphia tele­ vision. Further, as Kinder and Mendelberg (1995) find, racial isolation actl1ally f'nhancf's the impact of prejudice on white opinion. Similarly; 59 DiolU1e (1991, 88) recalls Nixon campaign speeches denouncing urban crime and welfare dependency in such places as New Hampshire. An area need not have "urban" problems to see minorities as a threat. In­ deed such appeals work best in white rural areas long fearful of all man­ ner of urban evils (Hofstadter 1964). The Lehigh Valley had an active klavern and neo-Nazi groups, one of which came into prominence when two brothers in the group killed their religious, racially tolerant parents and younger brother. A local Republican political consultant maintained that it was difficult for African-Americans to win election in the Lehigh Valley, though Colin Powell might be the one exception (Interview 2).6 In short, while overwhelmingly white, the Lehigh Valley is an inter­ esting place to study whites' willingness to vote for African-Americans. Its economic and social transitions and social conservatism make it a good place to look for authoritarian and populist tendencies in Ameri­ can politics. Testing the Hypotheses As Table 2 shows, Hypothesis 1 is consistent with the data. Colin Powell would have been a serious presidential candidate, essentially ty­ ing then-Senate Majority Leader Robert Dole as the first GOP choice among Lehigh Valley voters, with 23% to Dole's 25%. Powell led all other GOP contenders as a second choice and in combined (first and second place) support. Of course, one might suspect that as a nonparti­ san figure Powell will have more support among Democrats than other "Table 2 Support for Powell and other GOP Candidates in a Hypothetical Presidential Primary Candidales Percenl Percent Percents Chosen Listing Listing Combined* Dole 25 13 20 Kemp 18 14 16 Quayle 8 8 8 Buchanan 2 4 3 Powell 23 22 23 Gramm 3 3 3 Other / Refuse 22 35 100% 100% N=100 N=100 ~Du~:; nut ~yudllIl~dll of flrsllwo l:uluuUls u~l:ause of differenl num­ bers of missing cases. 60 Table 3 Percent Listing Powell as a Preferred 1996 GOP Presidential Candidate by Party. Respondent Self-Rating Democrat Independent Republican Views of Powell Strong Weak Weak Strong* Does Not List Powell 63 69 66 52 66 Powell Listed as Second GOP Choice 14 13 20 17 19 Powell Listed as First GOP Choice 24 19 14 31 15 Total percentage 100a 100a 100 100 100 X2(8) =7.75,)2=.46 N=335 * Total candidatp pf'r('pnta8f'~ in this table are greater than in Table 2 since this table combines two variables (FIRST and SECOND choice) for individual cases: it is additive rather than a mean. a Actual percentages add to 100; there is slight inflation in the table due to rounding error. GOP contenders can muster. In fact, this is not the case. As Table 3 shows, relationships with party are not statistically significant. Powell seemingly appealed most to weak Republicans, 48% of whom list him as either their first or second choice. Only 34% of strong Repub­ licans and 36% of the sample as a whole do suo Sl.rong Rt:publil.:aIU::i pre­ ferred more partisan candidates. Yet we found no clear tendency for Powell supporters to back him less against Clinton. On the Politics sur­ vey respondents were asked whether, if the election were held today, they would vote for President Clinton or for their favorite Republican. Those who choose Powell as their first choice break 63% to 37% for Powell; those chOOSing other GOP candidates support them by 75% to 25% mar­ gins, but the differences do not approach statistical significance in the small sample, X2(5)=4.01, ns. Naturally those unwilling to choose a Re­ publican backed President Clinton. In short, HI is supported. In investigating the other hypotheses, we first examine zero-order cor­ relations bel ween Powell suppurt ami hyputhe:::;i~ed preuidun;; affir­ mative action (H2), authoritarianism (H3), NAFTA and immigration sup­ port (H4), and populism (H5). In addition, we examine the correlations 61 Table 4 Zero-order Correlations between Powell Support and Hypothesized Predictors by Education Level Education Level Hypothesized No College College or More Predictor r N r N A uthoritarian Scale .02 75 - .26 *** 97 Support for Affirmative Action -.02 77 .08 96 Conservative Ideology -.08 83 - .05 101 Anti-Immigration -.17 85 - .17 100 , Pro-NAFTA -.03 83 .25 ** 99 Pro-Defense Spending .10 84 -.03 101 Populism Scale .01 85 - .23 ** 101 N=85 N=101 **}2<.05, ***}2<.Ol between Powell support and two other variables less relevant to our hypotheses, ideology and defense spending. We present correlations separately for those with no college and those with some college or above (See Table 4) since attitude constraint is strongly related to education (Devine 1970). Sniderman and Piazza (1993) and Uslaner (2002) find education inversely related to authoritarianism. In addition, as Popkin (1991) notes, issues important to the more atten­ tive public in the year before a campaign may become important to the less attentive in the months of a campaign; thus issue relationships among more educated respondents now may predict those among the less edu­ cated in the 1996 campaign season. For both groups, there are no relationships between support for Powell and general ideology, defense spending, and affirmative action. (Items are available on request at robert.maranto@villanova.edu.) Low educa­ ti'on respondents show no relationships between support for Powell and measures of authoritarianism, populism, and NAFTA. For both educa­ tion groups, those opposing (legal) immigration are somewhat less likely to support Powell, though the relationship attains statistical significance only when the two groups are combjned for a higher n, £(186)=-.18,12=.02. High education respondents show statistically significant relationships 62 Table 5 Zero-order Correlations between Powell Support and Hypothesized Predictors by Respondent Party Respondent Party Hypothesized Democrat Republican Predictor R N r N Authoritarian Scale -.10 59 -.23 *72 Support for Affirmative Action .01 59 .03 76 Conservative Ideology -.07 62 .02 81 Anti-Immigration -.27 ** 63 -.04 81 Pro-NAFTA .28 ** 61 .11 78 Pro-Defense Spending -.04 61 .04 81 Populism Scale -.03 63 -.20 * 81 N=63 N=81 *~ <.08, **~<.05 between support for Powell and authoritarianism, popUlism, and NAFTA. More authoritariah, populist respondents who oppose free trade and immigration are less likely to support Powell. In short, correlations tend to support H2, H3, and H4, and H5, at least for the educated. Affirma­ tive action support does not predict support for Powell, but support for NAFTA and (po~~ihly) immigration do. Gt>nf>ral idt>ology dof'~ not prf'­ diet support for Powell relative to other GOP candidates, nor does de­ fense spending, but authoritarianism and populism do. As Table 5 shows, presenting Powell support correlations by party tells a somewhat different story. Of the hypothesized variables, for Demo­ crats, only NAFTA and Immigration have statistically significant rela­ tionships with Powell support. For Republicans, authoritarianism and populism are nearly significant at .05, with I2.-values of .054 and .078, respectively. In short, realistic group conflict variables influence white Democratic support; the ideological variables of authoritarianism and populism seem to affect white Republican support. Thus, the zero-order correlations strongly support H2, work for H3 and perhaps H5 with Re­ publicans, and support H4 for Democrats. Zero-order correlations only examine the relationship between the pre­ c1irtor variable ;:mcl Powpl1 ~lJPport without "'lm1l1t;.mf'o1J~ly ("ontmlHn8 63 Table 6a Final Hierarchical Multiple Regression Summary Table for Less Educated Respondents. Predictor Variables B SEB Beta T-test Sig. T Affirmative action -.005 .en -.05 -.33 .742 Authoritarian scale .006 .01 .06 .46 .645 Pro-NAFTA .009 .03 .04 .27 .785 Anti-immigration -.05 .04 -.18 -1.37 .177 Populism Scale -.05 .10 -.07 -.47 .640 Constant .87 .80 1.09 .281 Multiple R=.18354 R2=.034, E(5,63)=.44 , p=.82 for other relationships. We used a hierarchical multiple regression to determine the pattern of Powell support with each predictor variable while taking into account its relationship with other predictors. We en­ tered predictor variables in the following order: affirmative action, authoritarianism, NAFTA, Immigration, and Populism. Affirmative ac­ tion was entered first to test H2 and authoritarianism was entered on the second step to test H3. Next, views on specific issues were entered, to see if they added explanatory power over and above the first two (more general) pred.iclors. Finally, pupulblll ~cure::; were entered last tu ::;ee if they added any explanation, over and above the other predictors, which are better established in the literature. This hierarchical multiple regression was run for four separate subsamples: respondents with high school education or less (Table 6a), those with some college or more (Table 6b), Democrats (Table 6c), and Republicans (Table 6d). Tables present the betas and tests for individual significance after all variables have been entered into the equation. Al­ though the tables show different patterns for these subsamples, indicat­ ing the necessity of splitting up the sample, lack of power is a problem due to small N's. However, multicollinearity is not a problem, as predic­ tors' intercorrelations never reach .3 and are seldom significant. As Table 6a indicates, for less educated respondents Powell support is not significantly related to any predictors by the end of the analysis. None of the E-values for the R2 change approached significance at any step-though immigration comes closest-indicating that Powell sup­ port was not related to predictor variables at any step. 64 Table 6b Final Hierarchical Multiple Regression Summary Table for Respondents with Some College and Above. Predictor Variables B SEB Beta T-test Sig. T Affirmative action -.005 .01 -.04 -.43 .666 Authoritarian scale -.02 .01 -.22 -2.16 .034 Pro-NAFTA .08 .03 .25 2.55 .0l3 Anti-immigration -.03 .03 -.09 -.85 .397 Populism Scale -.19 .08 -.25 -2.44 .017 Constant 1.81 .58 3.14 .002 Multiple R=.43261 R2=.18715 E(5,87)=4.00, 12=.00 Table 6b tells a more interesting story for the better-educated respon­ dents. Affirmative action was not significant at Step I, but the R2 change of .063 was significant for authoritarianism at Step 2, E(2,90)=6.09, 12=.016. NAFTA views, entered on the third step, also accounted for a significant increase in R2 change of .057, E(3,89)=5.8, ]2.=.018. Immigration (Step 4) was non-significant, but populism scores, entered last, accounted a defi­ nite R2 change of .055, £(5,87)=5.95, 12=.017. The £ for the final equation was highly significant, f(5,87)~4.01, 12-~.003. Democrats, portrayed in Table 6c show very little predictability in supporting Powell. The one variable that indicated a significant R2 change was NAFTA, entered on Step 3, E(1,54)=2.06, 12=.05. Immigration had the next highest T, but did quite not reach significance, £(1,54)=-1.52, }2.=.14. The F for the overall equation at that step was not significant; neither was the final equation with all five predictors entered.7 Table 6d shows that for the Republican subsample affirmative action is not significant but authoritarianism is. When authoritarianism was entered at Step 2, it led to a significant R2 change of .05, E (1,66)=3.95, }2.=.051. The other significant predictor is populism, entered last, which results in an R2 change of .058, with E (1,66)=4.19, .12=.045. Discussion: Not Quite Color Blind Survey results must be interpreted with caution. Saying that one will vote for an African-American candidate and actually doing so are differ­ ent things. 8 For our findings in particular, the small n of cases limit the 65 Table 6c Final Hierarchical Multiple Regression Summary Table for Democrat Respondents Predictor Variables B SEB Beta T-test Sig. T Affirmative action -.01 .02 -.09 -.64 .527 Authoritarian scale .002 .01 .02 .17 .865 Pro-NAI'TA .09 .04 .28 2.06 .045 Anti-immigration -.07 .04 -.21 -1.52 .136 Populism Scale -.13 .12 -.15 -1.08 .285 Constant 1.09 .84 1.29 .202 Multiple R=.37049 R2=.136, £:(5,50)=1.59, 12=.18 power of analyses. Further, the sheer prestige of Colin Powell may limit our ability to apply findings for less prominent African-American candi­ dates. Still, our findings accord with a growing literature examining the determinants of growing white support for African-American politicians. Results suggest that had he run, Colin Powell would have been a formi­ dable presidential candidate. At the time of our survey, in a "Reagan Democrat" area, Powell was essentially in a first place tie among white voters with then Senate Majority Leader and eventual 1996 Republican nominee Robert Dole. Further, in contrast to Bloom (1995) but in line with the findings of Sniderman and Piazza (1993), Powell support was not generally linked to support for affirmative action. Perhaps actual candidates, as opposed to experimental simulations, can overcome ra­ cial stereotypes-at least up to a point. Still, race matters. For better-educated respondents and for RepUblicans, authoritarianism and popUlism inversely correlated with support for PowelP This finding seemingly counters Sniderman et al's (1993) findings that race related views have more impact on Democratic than on Republi­ can opinion when considering "liberal" issues like welfare. Yet our finding .complements theirs in that Powell-a black Republican candidate-may provide more inconsistent cues for high stereotyping Republicans than for high stereotyping Democrats. In other words, the minority candidate's ideology, like the nexus of issue and party noted by Sniderman et al (1993), could moderate the impact of racial beliefs on political behavior. For Democrats, support for NAFJA and immigration was associated with in­ creased Powell support, suggesting {he utility of realistic group coriflicl models in 66 Table 6d Final Hierarchical Multiple Regression Summary Table for Republican Respondents Predictor Variables B SEB Beta T-test Sig. T Affirmative action -.04 .Ul .UU2 .U2 .987 Authoritarian scale -.03 .01 -.26 -2.10 .039 Pro-NAFTA .03 .04 .09 .73 .468 Anti-immigration -.008 .05 -.02 -.18 .861 Populism Scale -.19 .09 -.27 -2.05 .045 Constant 2.14 .77 2.77 .007 Multiple R=.36768 R2=.135, E(5,62)=1.94, J2=.10 explaining Democratic support. Our finding is somewhat consistent with McLauren (1995) who found that holding democratic norms increases benevolence towards non-f,uropean immigrants in several European countries. These trends may re­ flect the self-perceived economic vulnerability of white working class Democratic constituencies. In short, ideological variables seem best to explain Republican sup­ port for Powell; for Democrats, group conflict variables hold more promise. Our finding that populism predicts Powell support for Republicans and higher educated respondents shows the potential of a populist-lib­ ertarian dimension for explaining racial attitudes and voting behavior. While widely used by political consultants, this dimension is rarely ad­ dressed in the voting behavior literature and needs more research. For instance, determining whether the basis for populism lies more in an authoritarian Democrat or a working class liberal faced with realistic group conflict would provide much needed construct validation. This should be a topic of future research. The failure of party and general ideology to explain Powell support compared to the success of authoritarianism, populism, and NAFTA sug­ gests that, as Sniderman and Piazza show, political scientists should not confuse conservatism with racism. Similarly, researchers should he mmE' attentive to issue dimensions cutting across the conventionalliberal-con­ servative continuum. Finally, researchers must pay heed to the context of issues for different populations. In particular, for losers in the market­ place, realistic group conflict models may have more impact. In short, race still matters, though in a more complex manner than in the past. 67 Notes 1. At the 1995 Gridiron dinner in Washington (at which Congress members and others hold skits), both the Democratic and Republican presidential race skits featured a Powell look-alike adorned with medals and a halo, introduced with "Colin Powell, Superstar" to the tune of "Jesus Christ, Superstar" (C-SPAN, "Washington Journal," March 27, 1995). 2. We expected the embedded design to yield ditierent responses on contro­ versial racial issues. In fact it did not, so questions used in both surveys Can be pooled for analysis. 3. Minority respondents included 4 Asians, 4 blacks, 6 Hispanics, 1 Native American, 3 multi-racial, and 13 don't know I refuse. The 6 Hispanics and 1 Na­ tive American were included with the 343 whites in all subsequent analyses. 4. Perot had a strong local organization. In the 1994 House election, Patriot Party (Perot) candidate Vic Mazziotti took 5%. Since first term incumbent Paul McHale won by only 400 vote~ uut of 150,000 casL and the adiculale Mazziotti ran as a fiscal conservative, many local GOP activists felt that Mazziotti kept the party from picking up a House seat. Clinton was perceived as unpopular in the di~trkt, partil:ularly ,Hllong velelans, gun owners, SeniOl" citizens still resenting the 1960s, and Catholics concerned with social issues. Local GOP candidates sought to tie local Democrats to Clinton in 1994, and did well (Interview 2). 5. At one of these focus groups an Allentown woman complained of Clinton's credibility, saying, "If you asked him his favorite color, he'd say plaid" (Inter­ view 2). In response, the Clinton campaign carefully crafted the candidate's per­ sonal image in the pre-convention days to emphasize his humble background and Arkansas roots. 6. Similarly; in oux interactions with local political activists of both parties, one often hears that it is hard for women to win in the Valley, sometimes fol­ lowed by remarks such as "I think people here might be ready for that now." Female politicians of both parties see sexism as an obstacle. The local state legis­ lative delegation of 19 has not included more than three women. A notable ex­ ception was longtime State Senator Jeanette Reibman, a liberal Democrat who stressed her strong family and National Rif1p A~~oriat1on mf'mbpr~hlp. 7. The same hierarchical regression run for Independents had an n of only 34; no predictors were significant at any time. 8. Indeed, as the 1989 Douglas Wilder-Marshall Coleman race for Virginia Gov­ ernor showed, whites commonly lie in exit polls about their willingness to vote for black candidates so as not to appear racist -even though some may in fact be racist. (We encountered this phenomenon while doing election night comn'1.entary on a Virginia radio station-,-pollsters predicted a Wilder landslide, not a 1 % triumph.) This is as good an argument as any for the utility of modem racism research. 9. This correlation may be described best by saying that low authoritarians are more likely to vote for Powell, given the high support of Powell among the typical category for libertarians - weak Republicans (see Table 3). 68 References Allport, Gordon W. 1954. The Nature of Prejudice. Garden City: Barone, Michael and Grant Ujifusa. (1976; 1981; 1983; 1989; 1993). Almanac of American Politics. Washington: National Journal Bloom, Joel David. 1995. II A Wolf in Wolf's Clothing: The Return of Overt Rac­ ism in the 1992 National Election Studies." Paper presented at the annual Midwest Political Science Association Convention in Chicago, April 6-8. Bobo, Lawrence. 1988. "Group Conflict, Prejudice, and the Paradox of Contem­ porary Racial Attitudes." In Eliminating R.acism, ed. Phyllis A. Katz and Uatmas A. Taylor. New York: Plenum Press. Buckley, William F., Jr. 1992. In Search of Anti-Semitism. New York: Continuum. C-SPAN. 1992. "Road to the White House." December 13. __ . 1995. "Road to the White House." March 26. Devine, Donald J. 1970. The Attentive Public: Polyarchical Democracy. Chicago: Rand McNally. Dionne, E.J., Jr. 1991. Why Americans Hate Politics. New York: Simon and Schuster. Elliason, Maura. "All Things Considered." March 31, 1995. Fritzsche, Peter. 1990. Rehearsals For Fascism. New York: Oxford University Press. Frum, David. 1994. Dead Right. New York: Basic Books. Gresham, April W., Jinhui Yun, undEilccn Lcucy. 1995. "Origins of Modern Racism: Protestant Work Ethic or Blatant Prejudice." Poster presented at the annual convention of the Eastern Psychological Association, Boston, March. Interview 1. November 9, 1994 with a Virginia political consultant. Interview 2. March 30,1994 with a Lehigh Valley political consultant. Hagen, Michael G. 1995. "Encoded References to Racial Issues." Paper pre­ ~f>ntpd at the annual Midwest Political Science Association Convention in Chicago, April 6-8. Hofstadter, Richard. 1964. Anti-Intellectualism in American Life. New York: Vin­ tage. Kinder, Donald R., and Tali Mendelberg. 1995. "Cracks in American Apartheid: The Political Impact of Prejudice among Desegregated Whites," Joumal of Politics 57 (2): 402-24. Lavrakos, Paul, Jr. 1992. Telephone Survey Methods. Applied Social Research Methods Series, Vol. 7. Newbury Park: Sage. Lehigh Valley Morning Call. 1995. May 30, May 31, and June 2. Lind, Michael. 1993. "Aliens Among Us: the new right's new gambit on immi­ gration," The New Republic, August 23, 22-23. Maddox, William S., and Stuart A. Lilie. 1984. Beyond Liberal and Conservative. Washington: CATO Institute. McConahay, J.B. 1986. "Modem Racism, Ambivalence, and the Modern Racism Scale." In Prejudice, Discrimination, and Racism, ed. I.E Dovidio & S.L. Gaertner. San Diego, CA: Academic Press. McLauren, Lauren. 1995. "Individual, Contextual, and Cultural Explanations of Support for the Rights of Immigrants in Western Europe." Paper presented at the annual Midwest Political Science Association Convention in Chicago, April 6-8. 69 Popkin, Samuel L. 1992. The Reasoning Voter. Chicago: University of Chicago Press. Raum, Tom. 1994. "Polls find people would pick Powell as next president," Lehigh Valley Morning Call, September 12, All. Schuman, Howard, Charlotte Steeh, and Lawrence Bobo. 1985. Racial Attitudes in America. Cambridge: Harvard University Press. Sears, David O. 1988. "Symbolic Racism." In Eliminating Racism, ed. Phyllis A. Katz and Dalmas A. Taylor. New York: Plenum Press. Sigelman, Carol K, Lee Sigelman, Barbara J. Walkos, and Michael Nitz. 1995. "Black Candidates, White Voters: Understanding Racial Bias in Political Per­ ceptions," American Journal of Political Science 39 (February): 243-65. Sniderman, Paul M., Edward G. Carmines, Philip E. Tetlock, and Anthony M. Tyler. 1993. "The Asymmetry of Race as a Political Issue: Prejudice, Political Ideology, and the Structure of American Politics." Presented at the American Political Science AssocIation Convention 1ll Washington, D.C. and Thomas Piazza. 1993. The Scar of Race. Cambridge: Belknap. Sonenshein, Raphael J. 1990. "Can Black Candidates Win Statewide Election?" Pulitical Science Quarterly. 105 (2): 219-241. __ . 1993. Politics in Black and White. Princeton: Princeton University Press. Thernstrom, Stephen, and Abigail Themstrom. 1997. America in Black and White. New York: Touchstone. Uslaner, Eric M. (2002, forthCOming). The Moral Foundations of Trust. Cambridge: Cambridge University Press. Vidanage, Sharmaine, and David Smrt:;. 1995. "The Fuundations of Public Opinion Toward Immigration: Self-Interest, Group Conflict or Symbolic Politics?" Paper presented at the annual Midwest Political Science Association Convention in Chi­ cago, April 6-8. Williams, Linda. 1990. "White/Black Perceptions of the Electability of Black Political Candidates," NationaZPolitical Science Review 2: 145-64. Woodward, Bob. 1991. The Commanders. New York: Simon & Schuster. 70