John W. Williams, Political Science,

Principia College, Elsah, IL 62028

Portions of this paper were presented ath the 1995 annual meeting of the Illinois Political Science Association and the 1995 annual meeting of the Midwest Association for Public Opinion Research, and published in the 1996 issue of the Illinois Political Science Review.

"Since, 1980, female voters have been more likely than male voters to support Democratic candidates," at least in national elections. "Overall, women are more Democratic than are men, with the gap largest among the young." The growing gender gap in American politics, with more men voting Republican and more women voting Democratic, may have played itself out in the 1994 Illinois state elections. Incumbent Republican Jim Edgar handily defeated challenger Democrat Dawn Clark Netsch for the governorship of Illinois (1,973,239, 64.1% to 1,051,068, 34.2%). Edgarís win was part of a Republican sweep of all the "constitutional" or statewide officers, as well as control over both houses of the General Assembly. The unprecedented defeat for the state Democratic Party overshadowed a frequently discussed but quickly forgotten curiosity of the Illinois gubernatorial race. The incumbent Republican candidates for governor and lieutenant governor, Jim Edgar and Bob Kustra, were both men, while the challenging Democrats, Dawn Clark Netsch and Penny Severns, were both women.

As of 1992, women slightly outnumber men in Illinois in the voting age population (52.2%). Politically, Illinois is a competitive, or more precisely a split state. Until this election, the state constitutional officers were from both parties. Throughout the recent historic tenure of Republican "Big Jim" Thompson in the governorís mansion (the longest serving governor in the stateís 170-year history), Democrats still controlled the General Assembly. Even with the November debacle, Democrats control the congressional delegation and both U.S. Senate seats.

The Democrats have fought hard for the governorís mansion since Thompsonís first victory in 1976. U.S. Senator Adlai Stevenson, III, came within 5,000 votes of defeating Thompson in 1982, only to be trounced when LaRouche candidates gained position on the 1986 Democratic slate, turning sure victory into ignoble defeat. Republican Secretary of State Jim Edgar narrowly defeated Democrat Attorney General Neil Hartigan in 1990. Republican control of decennial statewide redistricting process as well as a perceived conservative shift in the state assisted Republicans in winning majority control of the state senate in 1992. On the other hand, newcomer Carol Moseley-Braun defeated an unknown Republican for the open U.S. Senate seat in what was billed for America as "The Year of the Woman."

Incumbent Edgar might have been vulnerable had Illinois Democrats united behind a strong challenger. Instead, the party held a bruising three-way primary, among Cook County board president Richard Phelan, Attorney General Roland Burris, and State Comptroller Dawn Clark Netsch. Phelan was probably the most conservative of the Democrats and could have potentially attracted suburban Chicago "collar county" voters. Burris was the only African-American, until Moseley-Braun, to win statewide office and could draw minority votes, should they chose to vote. Netsch, the first woman to win statewide constitutional office and the first woman to seek the governorship, staked out a liberal position. She won on a plurality in an election filled with negative campaigning. Her ticketmate, state senator Penny Severns, had campaigned as Phelanís running mate. By quirks of the Illinois Democrat Partyís process, their 1994 slate was led by women (as were several of the other constitutional positions). Netsch and Severns hoped to turn out nearly two decades of Republican governors, perhaps by capitalizing on "The Year of the Woman."

Early in the primary process, Bob Kustra, the incumbent Lieutenant Governor, announced his intention to step down. Apparently, the position was so undemanding that he preferred the possibility of becoming a Chicago radio talk show personality. Although the details are not known, after an apparently fruitless search for a replacement, Kustra was talked into staying on the ticket. The strategy of this decision became apparent when Democrat Paul Simon announced his intention not to seek reelection to the U.S. Senate. Likewise, Moseley-Braun is considered vulnerable. Thus, Edgar could run for and presumably win one of the U.S. Senate seats and Kustra would move up to governorship. Both U.S. Senators in neighboring Missouri stepped up from the governorship. If Edgar won Simonís seat, Kustra would be positioned for Moseley-Braunís seat. Whatever the strategy, the Illinois Republican Party stayed with its incumbents, both men.

Given the stakes of the 1994 election, particularly for Illinois Democrats in general and Edgar and Kustra in particular, and the growing gender gap in politics, the 1994 contest would be an interesting matchup. Republican men and Democrat women would have been clear choices. Republican women and, even more, Democratic men could be in a quandary. How these two groups voted could spell victory or defeat. Would these voters vote party or gender?

Classic American voter studies have found that partisanship or identification with party, is one of the best explanations for voting decisions. Although partisanship or party loyalty has decreased substantially in recent decades, party identification still remains the single best predictor of voting. Although not identified as a cause of the decline of party identification, the phenomenon of the gender gap has begun to reshape our understanding of partisanship. Partisanship and gender are coming into conflict and the 1994 Illinois gubernatorial contest provided a testing ground for this battle.

Consistent with the commonalties of partisanship voting and gender voting, this study hypothesized:

I. Republican men would strongly support the male Republican incumbents; and

II. Democratic women would strongly support the female Democratic challengers.

Because the study of the conflict between partisanship and gender is just beginning, there is little basis for hypothesizing how Democratic men or Republican women might vote, given just those two variables -- party and gender. The gender gap may be increasing as partisanship declines, so that gender replaces, at least in part, party identification. In effect, former partisans become avowed independents and then cast votes based on gender. In addition, voters may resolve party/gender conflicts by not voting. In the 1992 Illinois U.S. Senate race, Republican women either crossed over or sat out the election. Democratic men (especially whites), however, probably sat out the race. The Moseley-Braun election was complicated by the race factor, with Moseley-Braun an African-American coming from the racially charged Cook County, where she was Recorder of Deeds. Therefore, this study is preliminary in examining the relationship between party and gender. Rather than hypotheses, the study poses two research questions:

III. Will Republican women more strongly support the Republican males or the Democratic females?, and

IV. Will Democratic men more strongly support the Democratic females or the Republican males?

However, this study does hypothesize that the strength of partisanship, considered a strong measure of voting behavior, influences the struggle between party and gender. Thus, the study hypothesized:

V. Strong partisans are more likely to vote party rather than gender; and

VI. Weak partisans are more likely to vote gender than party.

Election outcomes are influenced by many factors. In many races, including the governorís race, the outcomes appear influenced most heavily by money. Edgar outspent Netsch. In Illinois politics, race is an important factor, and traditionally Democratic African-Americans, particularly in the Chicago area, apparently sat out the election. So, instead of looking to the results, this study examined potential voter opinion in the week prior to the election.

This study is based on a random telephone poll of 346 registered voters conducted in the week prior to the November 1994 general election. Two Illinois political jurisdictions -- the Village of Godfrey and the City of Alton -- were identified. Both incorporated municipalities are located in Madison County in southwestern Illinois, within the St. Louis, Missouri, standard metropolitan statistical area. Both municipalities are also Illinois townships, the most basic Illinois political subdivision. Alton is an old Mississippi River town, pre-dating the Civil War, of approximately 34,000 people. Godfrey, one of Illinoisí newest municipalities, is a mix of suburban bedroom community and rural farmland, with a population of just over 16,000. The former is a Democratic stronghold, while the latter is a Republican break-away from heavily Democratic Madison County.

The telephone prefixes for the two municipalities were identified and the four-digit suffixes were randomly generated. Respondents were filtered for registered voters. Nearly a thousand working numbers were dialed in order to obtain 346 registered voters. Non-usable numbers included businesses, answering machines, constant busy singles and unanswered calls (systematic dial-back instructions were used), as well as households with no registered voters. No effort was made to obtain a balance by party or by gender.

Response sheets included the telephone prefix, in order to reconfirm the working assumption held by most that Alton is heavily Democratic and Godfrey is heavily Republican. This is not a foolproof method, since the prefix zones do not match the jurisdictional borders, although they are close.

Respondents were asked their gender, their party identification, and the strength of their party support or leaning. Those reporting themselves as independents were asked toward which party they leaned. Using a scale of 1 to 6 (low to high), respondents were asked to rank Jim Edgar and Dawn Clark Netsch on four qualities: competency, leadership, job effectiveness, and integrity. Respondents were asked two questions on four-element Likert scales: "Should the gender of a candidate be an issue for voters?" and "Is gender of a candidate ever an issue with you?" Finally, respondents were asked "If the candidates were absolutely equal in every regard, but gender, would you vote based on gender?" Respondents were also asked their age (in ranges) and years of education.

The responses were subjected to cross-breaks and chi-square tests. For this study, three variables were of primary interest: ratings of qualities of candidate (five qualities for each of Edgar and Netsch), party affiliation of respondents, gender of respondents. Cross-breaks and chi-squares were calculated two difference ways. First, party affiliation was crossed with rating of each candidateís qualities, controlling for gender. Second, gender was crossed with rating of each candidateís qualities, controlling for party identification.

Since so few respondents, only 29 or 8.4% of all respondents, indicated that their party loyalty was weak, the decision was made to combine independents who expressed a party preference with "weak" party identifiers. There is some controversy over how to classify independent leaners. Some argue that they are "hidden" partisans who should be considered identifiers. Cross-breaks and chi-squares were run with independents separated out, hence with three categories of political preference (Democrat, independent, and Republican), and with independent party-identifiers combined with the respective parties, hence two categories of partisanship or party leaning (Democrats and Democrat-leaning, and Republicans and Republican-leaning). If weak identifiers, as hypothesized in hypotheses V. and VI., have weaker commitment to party and greater commitment to gender, their exclusion from the two parties (as in the three-way combination) should result in strong or significant differences between Democrats and Republicans. Likewise, the inclusion of weak identifiers within the respective parties (as in the two-way combination) should result in weak or no significant differences between Democrats and Republicans. This should occur, as Abramson has argued, because independent "hidden" identifiers are considered even weaker partisans than are professed weak partisan identifiers.

The respondents reported themselves as: 124 Democrats, 88 Republicans, 112 independents, and 22 unknown. A little over half were men, 52.6%, and 47.4% were women. The 282 respondents reporting themselves as partisans (even as independents) were relatively distributed.

Chart I. Partisanship and Gender (n, %)

The 322 respondents reporting a political preference of Republican, independent or Democrat were also relatively distributed.

Chart II. Political Preference and Gender (n,%)

One of the anomalies that surfaced was the willingness of the respondents to comment on Governor Edgar yet their reluctance to comment on Comptroller Netsch. Less than 4% of the 346 respondents refused to answer the questions about Edgarís qualities, while approximately 26% refused to answer the questions about Netschís qualities. This may reflect general unfamiliarity with Netsch, who was a state senator from a Chicago legislative district, before rising to relative obscurity as state comptroller. Edgar, on the other hand, served two terms as Secretary of State, before becoming Governor. One of the duties of the Secretary of State is issuance of driverís licenses and motor vehicle registration. During his first term, Edgar had the state license plates redesigned. His re-election posters, billboards and bumper stickers duplicated the colors, typeface and style of the license plates.

PARTISANSHIP VS. CANDIDATES

The first set of results was based on political preference and party identification or leaning. There were significant differences between how Republicans and Democrats rated Netsch on all four qualities. The Democrats rated Netsch higher than did the Republicans.

Chart III. Partisanship v. Dawn Clark Netsch (reporting chi-square and actual significance)

There were significant differences between how Republicans and Democrats rated Edgar on all four qualities. The Republicans rated Edgar higher than did the Democrats.

Chart IV. Partisanship v. Jim Edgar (reporting chi-square and actual significance)

GENDER VS. CANDIDATES

The second set of results was based on gender. There was only one significant difference between how men and women rated either candidate on their qualities. Men rated Jim Edgar higher on job effectiveness than did women (chi-square=10.876, df=5, p=.05384). There were not significant differences between how men and women rated Edgar on his other three qualities. There were not significant differences between how men and women rated Netsch.

PARTISANSHIP VS. CANDIDATES, CONTROLLING FOR GENDER

The third set of results was based on political preference and party identification or leaning, controlling for gender. There were significant differences between how male Republicans and male Democrats rated Netsch on all four qualities. The Democratic men rated Netsch higher than did the Republican men.

Chart V. Male Partisanship v. Dawn Clark Netsch (reporting chi-square and actual significance)

There were no significant differences between how male Republicans and male Democrats rated Edgar on competency, integrity, job effectiveness, or leadership on a three-way split between Republicans, independents and Democrats. However, when just partisan identifiers (including partisan independents) are compared, there was a significant difference in Edgarís job effectiveness rating (chi-square=11.40, df=5, p=.04395). Otherwise, there was no statistical difference in how Republican men and Democratic men rated the incumbent governor.

There were significant differences between how female Republicans and female Democrats rated Netsch on all four qualities. The Democratic women rated Netsch higher than did the Republican women.

Chart VI. Female Partisanship v. Dawn Clark Netsch (reporting chi-square and actual significance)

There were significant differences between how female Republicans and female Democrats rated Edgar on integrity (chi=square=24.676, df=10, p=.00599) and leadership (chi-square=20.758, df=10, p=.02285). Republican women rated Edgar higher than did Democratic women. There were no significant differences between how female Republicans and female Democrats rated Edgar of competency or job effectiveness.

GENDER VS. CANDIDATES, CONTROLLING FOR PARTISANSHIP

The fourth set of results was based on gender, controlling for party identification. There were significant differences between how female Democrats and male Democrats rated Netsch on all four qualities.

Chart VII. Democrats v. Dawn Clark Netsch (reporting chi-square and actual significance)

There were no significant differences between how female Republicans and male Republicans rated Netsch.

With one exception, there were no significant differences between how female Democrats and male Democrats rated Edgar. When weak partisans are calculated with party identifiers, there is a significant difference in ratings of Edgarís job effectiveness (chi-square=11.48, df=5, p=.04259). Male Democrats tended to rate Edgar higher than female Democrats.

There were no significant differences between how female Republicans and male Republicans rated Edgar.

GENDER AS AN ISSUE FOR VOTERS

The fifth set of results was based on the question: "Should gender of a candidate ever be an issue for the voters?" Almost every respondent answered this question and few believed that gender should be an issue with voters.

Chart VIII. Should gender of a candidate ever be an issue for the voters? (n,%)

The ratings of each candidateís qualities were compared to responses to the question on a four-element Likert scale from "yes, very relevant" to "no, never relevant." There were significant differences in respondentsí answers for two of Edgarís qualities: competency (chi-square= 30.746, df=15, p=.0095) and leadership (chi-square=26.48, df=15, p=.03327); and one of Netschís qualities: leadership (chi-square=26.919, df=15, p=.02941). Those who said gender should be a relevant issue tended to rate Edgar lower on competency and leadership and Netsch higher on leadership than did those who said gender should not be an issue. Democrats were more likely to believe that gender should be a relevant issue that Republicans (chi-square=14.64, df=6, p=.02325). However, there was no significant difference between women and men.

GENDER AS AN ISSUE FOR THE RESPONDENT

The sixth set of results was based on the question: "Is the gender of a candidate ever an issue with you?" Like the previous question, almost all the respondents answered the question, and few believed that gender was ever an issue with them.

Chart IX. Is the gender of a candidate ever an issue for you? (n,%)

The ratings of each candidateís qualities were compared to responses to the question on a four-element Likert scale from "always" to "never." There were not significant differences in respondentsí answers to any of Edgarís qualities. However, there were significant differences in responses to two of Netschís qualities: competency (chi-square=28.806, df=15, p=.01704) and integrity (chi-square=37.81, df=15, p=.00096). Those who claim to always vote on gender rated Netsch higher in both categorizes than did other respondents. There was not a significance difference between Democrats and Republicans or women and men on this question.

EVERYTHING BEING EQUAL

The seventh set of results was based on the question: "If the candidates were absolutely equal in every regard, but gender, would you vote on gender?" And, like the previous two questions, almost every respondent answered the question, and few answered an unequivocal "yes."

Chart X. If the candidates were absolutely equal in every regard, but gender, would you vote on gender? (n,%)

The ratings of each candidateís qualities were compared with the answers to the question. There were not significant differences in respondentsí answers to any of Netschís qualities. However, there was significant difference in response to one of Edgarís qualities: leadership (chi-square=19.833, df=10, p=.03087). Respondents who said they would vote on gender, rated Edgar higher on leadership, than those who said they would not vote on gender. There was not a significant difference between Democrats and Republicans or women and men on this question.

CORRELATIONS

Although not part of the original design, the various variables in this study were analyzed for correlations. As would be expected, Netschís four qualities were highly correlated (correlations ranged from .7640 to .8659, all had p<.000) and Edgarís four qualities were highly correlated to each other (correlations ranged from .7123 to .8319, all had p<.000). Though not strong, Edgarís qualities were correlated to party identification (correlations ranged from .2043 to .2189, all had p<.000) and Netschís qualities were negatively correlated to party identification (correlations ranged from .-.2895 to -.4489, all had p<.000). Finally, though not strong, the correlations among the last three questions, involving gender as a voting issue, were significant (correlations ranging from .4330 to .6459, all had p<.000).

As one might expect, Republicans, both self-professed and hidden identifiers, rated the Republican candidate for governor, incumbent Jim Edgar, significantly higher on all qualities than did Democrats, both self-professed and hidden identifiers. Likewise, those Democrats rated the Democratic candidate for governor, challenger Dawn Clark Netsch, significantly higher on all qualities than did the Republicans.

On the other hand, there was no significant difference between the genders and how they rated either of the candidates, with the one exception being over Jim Edgarís job effectiveness. This appears consistent with the strong reluctance of voters to make the gender of a candidate an issue. This seems at odds with the stereotypes held of southern Illinois voters. Lynn Martin, the Secretary of Labor under Ronald Reagan, when she challenged incumbent Democrat Paul Simon for his U.S. Senate seat, undiplomatically announced that those in southern Illinois were "rednecks." She was defeated by a nearly a million vote margin. The votersí answers were, however, consistent.

As hypothesized, Democratic women were stronger supporters of Netsch, the female Democratic candidate, than were Democratic men. However, there was no difference in the support of male and female Republicans for Edgar, the male Republican candidate. Likewise, there was no difference by gender in Republican opposition (or support) for Netsch. Yet, there was, on one quality, a difference between male and female Democrats in their support for Edgar, with men being stronger. Thus, hypothesis I, that Republican men would strongly support the male Republican incumbents, is rejected, while hypothesis II, that Democratic women would strongly support the female Democratic challengers, is retained.

As one might expect, there were differences in how female Democrats and female Republicans rated Netsch, with Democrats rating her higher on all qualities. However, with one exception, male Republicans and male Democrats did not rate Edgar differently.

There were statistical differences between how male Democrats and male Republicans rated Netsch, with Democrats rating her higher on all qualities. Female Republicans rated Edgar higher than did female Democrats on only two qualities. These results do not answer the two research questions clearly: will Republican women more strongly support the Republican men or the Democratic women? and will Democratic men more strongly support Democratic women or Republican men? The evidence suggests, however, that the Democratic men, because of their higher ratings of Netsch, see themselves differently than the Republican men, and, hence, might vote party over gender. The evidence suggests, although less clear, that Republican women, because of their higher ratings of Edgar, might vote party of over gender. This would be consistent with the repeated evidence that the voters are reluctant to make gender an issue.

There was no pattern between the two sets of comparisons -- political preference among Democrats, independents and Republicans; and partisanship between Democrats and hidden Democrats and Republicans and hidden Republicans. The expectations of the last two hypotheses were that strong party loyalists and identifiers would vote party over gender (hypothesis V.) and weak party identifiers would vote gender over party (hypothesis VI.). In order to be so, the first set of comparisons (with hidden identifiers separated out) should have consistently been stronger, while the second set of comparisons (moderated by inclusion of hidden identifiers) should have been consistently weaker. However, there was not pattern of strengths or weaknesses. Therefore, both hypotheses V. and VI. are rejected. If there is any area that should be pursued in future study, it is further exploration of the potential relationship between strength of partisanship and gender of candidates. Future research should explore an alternative research design as well as a larger sample size.

It appears that party loyalty and partisanship (including hidden identifiers among the independents) are still strong in the American polity. This study suggests that partisanship remains a greater influence than gender on the voting decision. This should be good news for those who argue for the American party system and for those who argue for a gender-blind meritocracy. This should also be good news for opponents of negative advertising that makes gender rather than substantial matters the central issue of a campaign. On the other hand, this is not good news for the growing number of women entering politics who need the support of women in order to break the glass-ceiling of politics.

On the other hand, gender was not an issue in the 1994 Illinois gubernatorial campaign. Edgar did not campaign "against a woman" and Netsch did not campaign "as a woman." This may explain, according to Ellen Cannon of Northeastern Illinois University, why Jim Edgar received the "womenís vote." Cannon contends that Edgarís campaign understood how women respond to issues and took it advantage of "family values" as a core issue. Netsch, with her "Iím all business" approach, never conveyed herself as "a nurturing woman who understood raising children, family and mothering." Unlike the 1992 election with "Anita Hill versus Clarence Thomas," this election had no single issue to mobilize a gender vote. Ultimately, the situation Netsch faced was on how to portray herself in order to attract both men and women, or at least winning the support of women without losing the support of men. As Cannon summarized, "women candidates face the daunting challenge of finding ways to communicate nurturing and the ability to run a government at the same time."