Gender and the 2018 U.S. Congressional Election ()
1. Introduction
Gender is an ongoing topic of great interest in political science and political sociology. Politicians, political analysts, campaign advisors, and news media have all devoted considerable attention to the contemporary gender gap, the tendency for a larger percentage of women than men to significantly prefer the Democratic party and Democratic candidates over the Republican Party and Republican candidates (Box-Steffensmeier et al., 2004 ; Center for American Women and Politics, 2022; Diekman & Schneider, 2010; Igielnik, 2020; Kaufmann, 2006; Kaufmann & Petrocik, 1999; Kellstedt et al., 2010; Koeppl-Turyna, 2021; Kittilson. 2016; Lizotte, 2017; Norrander, 1999; Pew Research Center, 2018). The gender gap is generally measured by subtracting the percentage of men who vote for a Democratic candidate from the percentage of women who vote for a Democratic candidate and party preference the percentage of men preferring the Democratic Party from the percentage of women preferring the Democratic Party (Kaufmann, 2006). Outside of the United States, the greater leftward tendency of women voters relative to men voters is also characteristic of Western European democracies, though not of post-Communist states (Abendschon & Steinmetz, 2014; Givens, 2004; Kittilson, 2016; Inglehart & Norris, 2003; Emmenegger & Manow, 2014).
The contemporary political gender gap, however, is the reverse of the one that existed in the first half of the twentieth century. In the years immediately following women winning the legal right to vote, the majority of women voters were generally more conservative than men in terms of party preference and voter choice (Kittilson, 2016; Lipset, 1960). Political observers concluded that this was due to multiple factors. One was that in the period immediately after obtaining the franchise, women were more subject to religious influence and more closely tied to religious organizations than men and thus were more likely to prefer religiously oriented parties and politically conservative parties reflecting traditional religious beliefs in their policy goals. Another factor suggested by political scientists and sociologists was that at that time, only a relatively small percentage of women were active in the labor force outside the home. Consequently, most women were not subject to certain influences affecting male workers such as labor unions, progressive and socialist political and economic ideas, and the leftist political parties that trade unions typically supported. As populations increasingly urbanized, however, the political sway of religion declined. Furthermore, the enormous growth of female labor force participation meant that more women encountered working place factors and labor union influences that shifted many of them away from right-wing political ideologies and towards ideas favored by leftist parties, narrowing the original political gender gap and eventually reversing it. Further proposed theoretical explanations for why women became more favorable towards left-wing political parties and candidates include the potential impacts of additional changes in the experiences, work, and living conditions of women and men, changes in economic and political environments, shifts in marital and family characteristics, and the overall influence in these contexts of women’s traditional sex role trait of being generally more concerned about the well-being of others than men (Box-Steffensmeier et al., 2004; Diekman & Schneider, 2010; Edlund & Rand, 2002).
The trend of women shifting politically to the left also characterized the United States. As the Democratic Party continued its center-left orientation and, in particular, embraced in the 1960s the racial equality of opportunity goals of the Civil Rights movement, women’s political party preference started to shift towards favoring the Democratic Party over the Republican Party. By the 1980 U.S. election women began to display a significant and consistently greater preference than men for voting for Democratic presidential candidates, although the size of the gender voting gap varied among elections. But this change in the political gender gap was also due a shift of white men, in particular southern white men, away from the Democratic Party and Democratic candidates to the Republican Party and Republican candidates (Kaufmann, 2006; Kaufmann & Petrocik, 1999; Kittilson, 2016; Lizotte, 2017).
Research suggests that the specific factors contributing to the political gender gap in the U.S. have tended to change in response to shifts in the sociopolitical context over time (Kaufmann & Petrocik, 1999; Kittilson, 2016). For example, in times of confrontations with other nations, women have often been less supportive of resorting to war and therefore more supportive of political candidates opposing war than men. Interestingly, however, regarding issues such as equal rights for women and legal access to abortion, no or only small differences have been observed between men and women in recent years (Lizotte 2017). In comparison, one of the most consistently observed politically relevant gender differences over time has been that women have been more supportive of laws and government policies to enhance social welfare, provide wide access to health care, and improve the living condition, financial situation, and income opportunities of the poor. This appears to have been a major reason why in the last several decades women have preferred the Democratic Party whose stated aims included “economic justice” goals such as raising the minimum wage and providing universal healthcare coverage.
2. Hypotheses
In the current study, we make use of data from an online survey of voters in five Midwestern states conducted at the time of the 2018 Congressional election. In the months leading up to that election, in addition to economic justice, major politically relevant issues included how to deal with immigration, the desirability or undesirability of a socially diverse United States, whether to tighten gun safety laws, and whether there was a need for criminal justice reform. As for previous elections in recent decades, the authors anticipated that in 2018 there would be a pro-Democratic gender gap among voters in our sample regarding both party affiliation and voter choice.
1) Due to the political context at the time, we hypothesized that there would also be a pro-Democratic gender gap regarding having an unfavorable opinion towards then President Donald Trump and that this factor would have an effect on voter choice similar to that of Democratic Party affiliation.
2) We further hypothesized that there would be gender gaps with regard to certain politically relevant issues of the time such as support for tightening gun control, support for immigration, support for diversity in U.S. society, support for economic justice, and support for criminal justice reform.
3) We additionally hypothesized that the gender effects on voter choice, party affiliation, and having an unfavorable opinion of Trump would be conveyed largely through gender effects on major politically relevant issues contemporaneous with the 2018 Congressional election including support for tightening gun control, support for immigration, support for diversity in U.S. society, support of economic justice, and support for criminal justice reform.
3. Methods
Data come from a random sample of 1746 adults in five Midwestern Great Lake states, Illinois (N = 356), Michigan (N = 354), Minnesota (N = 360), Ohio (N = 338) and Wisconsin (N = 338), organized by the Public Religion Research Institute (PRRI) (Jones & Najle, 2019; PRRI, 2019). Of these states, Michigan, Ohio and Wisconsin were won by the Republican candidate for president in 2016, while Illinois and Minnesota were won by the Democratic candidate for president. Interviews were conducted online in both English and Spanish between December 17 and 23, 2018. Besides obtaining demographic, economic, education and religious data, the study included questions relating to views on major topics of political and social concern in 2018. These included items intended to measure support for stricter gun control, immigration, guaranteeing health care, increasing the minimum wage, the desirability of social diversity in the U.S., and caring out criminal justice reforms. Important for our analyses were questions on political party preference, favorability or unfavorability towards then President Donald Trump, and whether and how one voted in the 2018 Congressional election. For the purposes of our study, we selected only respondents who said they actually voted for either a Democratic or a Republican candidate for Congress in 2018 and who had no missing responses or refusals on any of the items we used. This resulted in a sample of 1228 voters (composed of 653 women and 575 men) for our analyses. All response categories for individual questions were reorganized, when needed, so that a higher score represented higher support for the topic in question. At the appropriate stage in our research, we standardized certain survey items (which had somewhat varying response categories) for the purpose of constructing indices of support for social issues used in regression analyses. Since the ultimate dependent variable, voter choice, is binary, we used logistic regression in the analysis of that variable.
Three items were used to construct the support for the immigration index. These included 1) “Which statement comes closest to your view about how the immigration system should deal with immigrants who are currently living in the U.S. illegally?”; a) “Allow them a way to become citizens provided they meet certain requirements”; b) “Allow them to become permanent legal residents, but not citizens”; c) “Identify and deport them” (PRRI, 2019). 2) “Do you strongly favor, favor, oppose or strongly oppose allowing immigrants brought illegally to the U.S. as children to gain legal resident status?” (PRRI, 2019). 3) “Do you strongly favor, favor, oppose or strongly oppose changing the Constitution to prevent children who are born in the United States to non-U.S. citizens from automatically being granted American citizenship?” (PRRI, 2019).
Two items were selected to comprise the support for religious diversity index. These included 1) “Thinking about what would be your own ideal for the country, please place yourself on the following (1 - 10) scale” with “I would prefer the U.S. to be a nation made up of people belonging to a wide variety of religions” at one end of the scale and “I would prefer the U.S. to be a nation primarily made up of people who follow the Christian faith” at the other end of the scale (PRRI, 2019). 2) “In your opinion, do you think the following is a very important, somewhat important, not too important or not at all important part of being truly American? – Being a Christian” (PRRI, 2019).
Support for national social diversity was measured by one item: “Thinking about what would be your own ideal for the country, please place yourself on the following (1 - 10) scale” with “I would prefer the U.S. be a nation made up of people from all over the world” at one end of the scale and “I would prefer the U.S. be a nation primarily made up of people from Western European heritage” at the other end of the scale (PRRI, 2019).
The criminal justice reform index was made up of two items. 1) “Do you strongly favor, favor, oppose, or strongly oppose a law in your state requiring drug treatment instead of jail time for people convicted of illegal drug use for the first or second offense?” (PRRI, 2019). 2) Do you “completely agree”, “mostly agree”, “mostly disagree” or “completely disagree” with the statement that “a person who has been convicted of a felony should be allowed to vote after they have served their sentence?” (PRRI, 2019). The economic justice index was composed of two items. 1) “Do you strongly favor, favor, oppose, or strongly oppose increasing the minimum wage from $7.25 an hour to $15 an hour?” (PRRI, 2019). 2) “Do you strongly favor, favor, oppose, or strongly oppose protecting health care coverage for people with pre-existing conditions?” (PRRI, 2019). Attitude towards gun control was measured by a single item, “Do you strongly favor, favor, oppose, or strongly oppose enacting stricter gun control laws?” (PRRI, 2019).
Party affiliation was measured by a question about what political party a respondent identified with. If a subject responded by giving an answer other than Republican Party, Independent, or Democratic Party, they were omitted from the analyses. If a subject responded that she or he was an Independent, they were asked if they leaned to the Republican Party or leaned to the Democratic Party. Then respondents were placed into one of three categories: Republicans and those independents leaning Republican scored 1, Independents scored 2, or Democrats and those independents leaning Democrat scored 3. Favorability or unfavorability towards President Trump was gauged by a question which asked respondents “Would you say your overall opinion of Donald Trump is very favorable (scored 1), mostly favorable (scored 2), mostly unfavorable (scored 3), or very unfavorable (scored 4)?” (PRRI, 2019).
Besides gender, which was coded male as 1 and female as 2, there were six additional exogenous variables used in the analyses: race/ethnicity, household income, educational level, age, religious affiliation, and frequency of attendance at religious services. Race/ethnicity was coded White non-Hispanic as 1 and all others as 0. Household income was registered on a 21 point scale with less than $5000 coded as 1 and $250,000 or more as 21. Educational level was measured on a 14 point scale with no formal education coded as 1 and professional or doctorate degree as 14. Ages of respondents ranged from 19 to 92. Religious affiliation was coded as White Evangelical Protestant 1 and all others 0. Frequency of attendance at religious services was measured by response to a question asking how often, other than weddings and funerals, did a respondent attend religious services with never coded as 1, seldom 2, a few times a year 3, once or twice a month 4, once a week 5, and more than once a week 6.
4. Results
4.1. Gender Gaps
In the sample, 59.1 percent of women and 52.9 percent of men reported voting for a Democratic candidate in the 2018 Congressional election representing a pro-Democratic Congressional voting gender gap of 6.2 percent. Regarding party affiliation, 57.9 percent of women voters reported either being members of the Democratic Party or being independents who leaned towards the Democratic Party compared to 49.0 percent of men voters. Thus, the pro-Democratic gender gap for party affiliation in the sample was 8.9 percent. Among women voters, 64.2 percent responded that they had an unfavorable opinion of then President Donald Trump compared to 57.5 percent of men voters resulting in a gender gap of 6.7 percent regarding having an unfavorable opinion of Trump.
In order to calculate estimates of gender differences regarding politically relevant social issues, the response options of the social issue variables were divided into just two categories with one indicating lack of support for the topic in question and the other indicating support. For example, for the immigration question “Which statement comes closest to your view about how the immigration system should deal with immigrants who are currently living in the U.S. illegally?”; a) “Allow them a way to become citizens provided they meet certain requirements”; b) “Allow them to become permanent legal residents, but not citizens”; c) “Identify and deport, them” the a and b options were categorized as support for immigration while the c option was categorized as lack of support for immigration. For the questions with ten options along a scale of 1 to 10, choices 1 through 5 comprised one combined response category and 6 through 10 the other combined response category. For items with four options, the first two were joined into one response category, while the other two were joined into the other response category. For example, “completely agree” and “mostly agree” were combined into a single agree category while “mostly disagree” and “completely disagree” were combined into a single disagree category. Then the binarized social issue items were combined into initial indices for the purpose of calculating percentage gender gaps. Figure 1 below presents the results.
The greatest social issue difference between women and men voters was on the topic of stricter gun control. Women’s support for this issue exceeded men’s support by 11.7 percent. On the topic of immigration, 7.2 percent more women than men expressed support for immigration. For economic justice, 6.7 percent more women than men responded with support. Support for national social diversity was expressed by 5.9 percent more women than men. Regarding criminal justice reform, 3.5 percent more women than men expressed support. And 2.5 percent more women than men responded with support for religious diversity in the U.S.
4.2. Factors Affecting Support for Social Issues
A second set of social issues indices was constructed suitable for use in regression analyses through standardizing the original response options of index items. Table 1 shows the effects of the exogenous variables on the social issue variables. All variance inflation factors for the regressions in Table 1 are under 1.38. Controlling for the other exogenous variables, being female had statistically significant positive effects on support for stricter gun control, support for immigration, support for economic justice, support for national social diversity, support for criminal justice reform, and support for religious diversity. Educational level also had statistically significant positive effects on support for all these issues. In comparison, frequency of attendance at religious services had statistically significant negative effects on support for all six issues.
4.3. Factors Affecting Party Affiliation and Opinion of Trump
The next questions addressed were how and to what extent did respondents’ views on the measured social issues influence political party preference and having an unfavorable opinion of Trump. Table 2 presents the results of ordinary least
![]()
Table 1. Regressions of support for stricter gun control, immigration, economic justice, national social diversity, criminal justice reform, and religious diversity on seven exogenous variables.
Standardized Regression Coefficients and Adjusted R2. N = 1228. *p < .05, **p < .01, ***p < .001.
![]()
Table 2. Regressions of democratic party affiliation and unfavorable opinion of Trump on support for stricter gun control, immigration, economic justice, national social diversity, criminal justice reform, and religious diversity and gender and six other exogenous variables.
Standardized Regression Coefficients and Adjusted R2. N = 1228. *p < .05, **p < .01, ***p < .001.
squares regressions of Democratic Party affiliation and unfavorable opinion of Trump on support for stricter gun control, support for immigration, support for economic justice, support for national social diversity, support for criminal justice reform, and support for religious diversity, and gender and the six other exogenous variables. All variance inflation factors for the regressions in Table 2 are under 2.4. The results showed that among the social issue variables linked to gender, support for stricter gun control had the strongest positive statistically significant effect on affiliation with the Democratic Party followed respectively by support for immigration, support for economic justice, support for religious diversity, and support for criminal justice reform. Controlling for other variables in the regression, support for national social diversity had no independent effect on affiliation with the Democratic Party. Even after controlling for the effects of the social issue variables, being White non-Hispanic, income level, age, being a member of a predominantly White Evangelical Protestant religious organization and frequency of attendance at religious services had statistically significant negative effects on affiliation with the Democratic Party, though these effects were generally weaker than those of the social issue variables. Regarding having an unfavorable opinion of Donald Trump, among the social issue variables support for stricter gun control again had the strongest positive statistically significant effect followed by support immigration. Support for economic justice and support for religious diversity had statistically significant positive effects of similar strength on having an unfavorable opinion of Trump. Support for criminal justice reform also had a significant positive effect on having an unfavorable opinion of Trump while support for national social diversity had no effect. Notably, once the effects of the social variables on having an unfavorable opinion of Trump were controlled for, being female registered an unexpected significant negative effect. Being White non-Hispanic, age, and being a member of a predominantly White Evangelical Protestant religious organization also had statistically significant negative effects on having an unfavorable opinion of Trump, but these were on average weaker than the social issue effects.
4.4. Factors Affecting Voter Choice
Given that the ultimate dependent variable in this study, whether the respondent voted for a Republican candidate or a Democratic candidate in the 2018 Congressional election, is binary, we examined which factors had direct independent statistically significant effects on voter choice through two logistic regressions. In the first, we regressed whether the respondent voted for a Republican candidate or a Democratic candidate in the 2018 Congressional election on the six social issue variables and the seven exogenous variables.
The results are presented in Table 3 below. Support for stricter gun control, support for immigration, support for economic justice, and support for religious diversity had the strongest statistically significant effects. All four of these social issue variables had positive effects on the tendency to vote for a Democratic candidate in the 2018 Congressional election. Support for criminal justice reform had a weaker statistically significant positive effect on voting for a Democratic
![]()
Table 3. Logistic regression of 2018 congressional democratic candidate voter choice on social issues and exogenous variables.
Unstandardized logistic regression coefficients with standard errors in parentheses. N = 1228. *p < .05, **p < .01, ***p < .001.
candidate. Among the exogenous variables, being female had a weak negative relationship to voting Democratic with the effects of the social issue variables removed. As with the results for the regressions of Democratic Party affiliation and having an unfavorable opinion of Donald Trump, this indicates that the pro-Democratic gender gap was conveyed entirely through the effects of being female on the social issue variables. Being White non-Hispanic, household income, being a member of a predominantly White Evangelical Protestant religious organization, and frequency of attendance at religious services all had relatively weak negative effects on the tendency to vote for a Democratic candidate in the 2018 Congressional election.
In the additional logistic regression analysis, 2018 Congressional Democratic Candidate Voter Choice was regressed on the six social issue variables, the seven exogenous variables, Democratic Party affiliation, and having an unfavorable opinion of Donald Trump. The results are presented below in Table 4.
The results show that the effects of the social issue variables and the exogenous variables on 2018 Congressional Democratic Candidate Voter Choice were conveyed almost totally through their effects on Democratic Party affiliation and having an unfavorable opinion of Trump. Only support for religious diversity continued to have a statistically significant positive effect independent of Democratic Party affiliation and having an unfavorable opinion of Trump.
![]()
Table 4. Logistic regression of 2018 congressional democratic candidate voter choice on social issues, exogenous variables, democratic party affiliation, and unfavorable opinion of Donald trump.
Unstandardized logistic regression coefficients with standard errors in parentheses. N = 1228. *p < .05, **p < .01, ***p < .001.
5. Conclusion
In the months leading up to the 2018 Congressional election, politically relevant issues included whether to enact legislation to tighten gun control, how to deal with undocumented immigrants, the desirability or undesirability of social diversity in the United States, whether the minimum wage should be increased, and whether there was a need to carry out criminal justice reform. In that election, it was anticipated that there would be a continuation of the political gender gap observed in recent decades in which a larger percentage of women than men favor the Democratic Party and vote for Democratic candidates. Due to the political context at the time of the election, we predicted that there would also be a gender gap regarding attitude toward then President Donald Trump with more women having an unfavorable opinion of Trump than men and that this factor would have a positive effect on voting for Democratic candidates similar to that of Democratic party affiliation. We hypothesized further that there would be gender gaps with regard to social issues with larger percentages of women than men expressing support for legislation to tighten gun control, support for immigration, support for economic justice, support for diversity in U.S. society, and support for criminal justice reform. We also hypothesized that the gender effects on voter choice, party affiliation and having an unfavorable opinion of Donald Trump would be conveyed primarily through gender effects on the social issues.
Consistent with the trend of recent decades, women voting in 2018 tended to prefer Democratic candidates and the Democratic Party more than men. The findings also confirmed that a larger percentage of women than men voters in 2018 had an unfavorable opinion of then President Trump. Gender gaps were also found with regard to all six social issue variables used in the study. Women’s support for enacting legislation to ensure stricter gun control exceeded men’s support by 11.7 percent, on support for immigration women’s lead was 7.2 percent, on support for economic justice 6.7 percent, on support for national social diversity 5.9 percent, on support for criminal justice reform 3.5 percent, and on support for religious diversity 2.5 percent. Furthermore, in multivariate regression of the social issues on seven exogenous variables including gender, race/ethnicity, income, educational level, age, religious affiliation, and frequency of attendance at religious services, being female had an independent positive effect on support for all six social issues. Beyond being female, only educational level had independent positive effects on support for all six social issues.
We hypothesized that the positive relationship between being female and preference for the Democratic Party, having an unfavorable opinion of Donald Trump, and voting for Democratic candidates during the 2018 Congressional election would be conveyed through the social issues as mediating variables. The results of the regressions of Democratic Party affiliation, having an unfavorable opinion of Donald Trump, and voting for a Democratic candidate on the six social issue variables and the seven endogenous variables supported this hypothesis. With the effects of support for stricter gun control, support for immigration, support for economic justice, support for religious diversity, and support for criminal justice reform, all of which had statistically significant independent positive effects on Democratic Party affiliation, having an unfavorable opinion of Trump, and voting for a Democratic candidate in the 2018 Congressional election, removed, being female no longer displayed statistically significant positive effects.
It is interesting to note that while there was a gender gap for the issue of support for national social diversity, this variable, in contrast to the other five social issue variables, had no independent effect on Democratic Party affiliation, having an unfavorable opinion of Trump, or voting for a Democratic candidate in the 2018 Congressional election. In comparison, support for religious diversity did have statistically significant positive effects on Democratic Party affiliation, having an unfavorable opinion of Trump, or voting for a Democratic candidate in the 2018 Congressional election. It is possible that the reason for the difference in the effects of these two diversity variables has to do with how they were worded. The support for national social diversity variable was based on responses to the statement: “Thinking about what would be your own ideal for the country, please place yourself on the following (1 - 10) scale” with “I would prefer the U.S. be a nation made up of people from all over the world” at one end of the scale and “I would prefer the U.S. be a nation primarily made up of people from Western European heritage” at the other end of the scale. But selecting the first choice does not imply that the respondent favors that U.S. have a non-White majority population or a majority non-Western culture population. It only states that the U.S. be a nation with some level of population from diverse geographical regions of the world. In comparison, the items that make up the religious diversity variable ask the respondent to, in effect, choose whether she or he would be willing to support the U.S. having a non-Christian majority population. It is likely that the concept of the U.S. being comprised of a non-Christian majority is more unacceptable to Republican inclined voters than the concept of the U.S. including an unspecified percentage of people from geographically different global locations.
It is also important to recognize that the support for immigration variable is also in part a measure of support for social diversity because the items making up this index, whether immigrants in the U.S. illegally should be allowed pathways to citizenship or permanent legal residency and whether a new law should prevent children born in the U.S. to non-citizens from automatically becoming citizens, also have implications for the level of social diversity in the U.S.
While this study analyzed a relatively large sample of voters, it had several limitations. Although the data came from five large Midwestern states, it is not certain that the findings obtained here would also have characterized the rest of the United States. Second, it is possible that there were gender gaps regarding one or more politically relevant social issues that were not in the covered in the survey from which the data were drawn. Future research should extend a similar analysis to a large national sample of voters and incorporate measures of social issues about which there is large scale public concern in the period immediately preceding the relevant election.
The Trump administration’s controversial policies on gun control and immigration likely contributed to the mediating roles that these issues played between gender and Democratic Party affiliation, having an unfavorable opinion of Donald Trump, and voting for a Democratic candidate in the 2018 Congressional election. It is also noteworthy that while being female and having a relatively high level of education had positive relationships to support for stricter gun control, support for immigration, support for economic justice, support for criminal justice reform, and support for religious diversity, frequency of attendance at religious services had negative effects on all these variables and being a member of a White Evangelical Protestant religious organization also had a negative effect on all of these except support for criminal justice reform. These findings suggest that the 2018 Congressional election in great part involved a pitting of women and the relatively well educated against the highly religious, in particular, white Evangelical Protestants. The effect of gender on voter choice in the 2018 Congressional election, like effects of other exogenous variables, appeared to be conveyed almost entirely through support or opposition to certain specific social issues.