Exploring the Gender Gap
Goal
The goal of this exercise is to explore the differences between men and women with respect to public opinion in areas important to political decision-making. Crosstabulation, Chi-square statistics, comparisons of means, and t-statistics will be used.
Concept
In political science literature, the gender gap refers to gender-based differences in public opinion, party preference, and vote choice. Research suggests that men and women differ when making political choices; specifically, women tend to identify with the Democratic Party more often and tend to be more ideologically liberal. As a result, women tend to have more liberal policy preferences than men.
Some research questions concerning the gender gap are:
- What factors lead women to identify more with the Democratic Party?
- How do men and women differ with respect to issue attitudes and policy preferences?
- Can the ideological and political differences between men and women be explained?
This exercise will use the 2004 American National Election Study. The American National Election Studies (ANES) grew out of the Survey Research Center and the Center for Political Studies of the Institute for Social Research at the University of Michigan. These organizations, together, have been covering elections since 1948. ANES produces high quality data on voting, public opinion, and political participation to serve the research needs of social scientists, teachers, students, policy makers and journalists who want to better understand the theoretical and empirical foundations of national election outcomes.
Respondents in the 2004 ANES were interviewed before and after the November election. Questions asked of respondents during the survey cover a broad range of topics including demographics, attitudes toward candidates and parties, attitudes on different segments of the American public, attitudes on foreign policy matters, and political behavior. The ANES uses random sampling in order to produce the most representative data possible on the American electorate.
Variables used in this exercise include:
- Gender (V041109A)
- Liberal-Conservative (V045117)
- Party identification (V043116)
- Interventionism by diplomacy/military (V043107)
- Government spending and services (V043136)
- Women's role (V043196)
- Importance of women's equal role (V043197)
Ideology
First examine the difference between men and women with regard to political ideology. Respondents were asked to rate their ideological position on a scale of 1 to 7 where 1 = "extremely liberal" and 7 = "extremely conservative." Run a comparison of means of ideology by gender. What is the mean ideology score for men and for women? What is the t-statistic for men's ideology score and for women's ideology score? What does the t-statistic tell you about the difference in men's and women's ideology scores?
Party Identification
Ideological identification and partisan identification should be closely related. Party identification (v043116) is measured with a 7-point scale ranging from Strong Democrat to Strong Republican, but also includes those who refuse to state their party preference. We recoded the variable, preserving the 7-point scale but eliminating the five respondents who did not fit into one of the seven response categories. The new variable is called "PARTYID."
Conduct a comparison of means test on PARTYID by gender. Is there support for the relationship between gender and party identification?
Attitudes toward Foreign Policy
Examining gender differences with regard to issue attitudes may give more insight into the gender gap. First consider attitudes toward intervention in foreign affairs. Respondents were asked to place themselves on a scale between 1 and 7 where 1 represents the position "U.S. should solve with diplomacy and international pressure" and 7 represents the position "U.S. must be ready to use military force."
Run a comparison of means of attitudes toward foreign policy and interventionism (V043107) by gender. When you look at the table of means, how do you interpret the difference between men and women?
Attitudes toward Government Services
Next consider attitudes toward government spending and services. In variable V043136 respondents placed themselves on a scale between 1 and 7 where 1 = "Govt should provide many fewer services" and 7 = "Govt should provide many more services." Look at the comparison of means test of attitudes about government spending by gender. Do women favor more government services than men?
Attitudes toward Gender Roles
Now look at gender differences in attitudes toward gender roles. In variable V043196, respondents placed themselves on a 7-point scale where 1 = "Women and men should have equal roles" and 7 = "A woman's place is in the home." Look at the comparison of means of V043196 by gender. From the comparison of means, what can you say about men's and women's attitudes about gender roles?
Keeping in mind these results, run a crosstabulation of the same two variables. Do you gain any further knowledge about gender and gender role attitudes? From the crosstab analysis and chi-square statistic (found below the main table), are you able to conclude that women are more likely than men are to say that women and men should have equal roles?
Finally, consider whether there are gender differences in the importance placed on the issue of gender equality. Respondents were asked, "how important is the issue of women's equal roles?" with response options ranging from "extremely important" (1) to "not at all important"(5).
Restricting your analysis to only those men and women who said that "women and men should have equal roles," run both a comparison of means and a crosstab of the importance of women's equality (V043197) by gender. How strongly, on average, do men feel about the importance of women's equal roles? How does this compare to the importance for women? Remember that everyone in this analysis believes that women and men should have equal roles. From the crosstab, which response category (row) shows the largest difference between men and women?
Think about your answers to the application questions before you click through the interpretation guide for help in answering them.
What is the mean ideology score for men and for women? What is the t-statistic for men's ideology score and for women's ideology score? What does the t-statistic tell you about the difference in men's and women's ideology scores?
Is there support for the relationship between gender and party identification?
How do you interpret the difference between men and women in attitudes toward using diplomacy versus military force?
Do men and women vary significantly on attitudes about government spending?
What can you say about men's and women's attitudes about gender roles?
Are you able to conclude that women are more likely than men are to say that women and men should have equal roles?
Considering only those who believe that men and women should have equal roles, is there a difference between men and women regarding the importance of the issue of women's equality?
Interpretation
Things to think about in interpreting the results:
- It is always important to consider how missing data might affect the ability to make inferential statements about the population from which the sample was drawn. Large amounts of missing data can limit our ability to generalize sample results. Even small amounts of missing data can affect the results, especially if missing responses are not randomly distributed across categories of the independent (grouping) variable.
- Reading the results: Briefly, the comparison of means calculates the mean of a dependent variable within categories of the row (independent) variable. The t-statistic measures whether the mean in a cell is larger or smaller than the overall mean. You can also think about it as how far the observed value in the cell is from the expected value in the cell (the sample mean) with the distance being measured in number of standard errors. The t-statistic also takes into account the total number of cases (n) in a cell. If there are only a few cases in the cell, then the deviation from the overall mean is not as significant as if there are many cases in that cell. Remember, when the absolute value of the t-statistic is large, the probability that you would obtain that value if the null hypothesis were true is small. That is, the probability that you would get the large t-statistic due to sampling error alone is small.
- Frequency distribution tables (crosstabs) display percentages in each cell of the people who fall into the overlapping categories (if two variables are used), followed by the actual number of people that represents in the sample. The coloring in the tables is based on the Z-statistic and demonstrates how the observed numbers in each cell compares to the expected number if there were no association between the two variables. Paler colors represent lesser significance of the difference between observed values and expected values. The Z-statistic is very similar to the t-statistic except that, unlike the t-distribution, the Z-distribution does not take into account sample size. The accompanying bar charts display the patterns visually as well.
- Weights (mathematical formulas) are often used to adjust the sample proportions, usually by race, sex, or age, to more closely match those of the general population. The analyses used in this guide did not use any weights, which may reduce the generalizability of the findings, but the resulting tables are accurate descriptions of the relationships found between these variables among these respondents.
- The analyses show the following:
- Although the difference between men and women in mean ideological identification appears to be relatively small, the absolute value of the reported t-statistics is greater than 2. Given that we have a relatively large sample, the t-statistic tells us that the difference between men and women is statistically significant. Since the female mean ideology score is lower than the male mean, there is support for the hypothesis that women are more liberal than men. Actually, it may be more appropriate to say that women are less conservative than men since the mean scores for both men and women exceed the midpoint of the scale (4) which suggests a very slight leaning toward the conservative end of the scale.
- The mean PARTYID score for men is 4.08 and it is 3.69 for women which is still a relatively small difference. The t-statistic for men is slightly higher in the PARTYID analysis than in the ideology analysis and the t-statistics for women are about the same in both analyses. Again, both t-statistics have an absolute value greater than 2 which suggests that there is a relationship between party identification and gender. There is support for the hypothesis that women tend to identify as a Democrat more than men, but we certainly cannot say that a majority of women identify as Democrat nor can we say that a majority of men identify as Republican.
- From the table, note that although there is a relatively small difference in the mean interventionism score between males and females, this difference is statistically significant and suggests that females are more likely to support foreign policy that features diplomacy than are males. Even so, males' mean score of 3.98 is right in the middle of the diplomacy-military force scale.
- With regard to government spending and services, the difference between males and females is not very large in terms of the mean response (males = 4.32, females = 4.69), however, the difference is still statistically significant. We know this because of the dark shading in both cells and because the absolute value of the t-statistic is greater than 2. There is a very small chance (less than 1%) that the observed difference between the male mean and the female mean could be this large or larger due only to sampling error. Females do appear to be more supportive of increased government spending on services than males.
- There is very little difference between the male (1.96) and female (1.88) mean response on the first equal roles variable. In fact, given that the corresponding t-statistics are below 1, we are not able to conclude that there is any difference between males and females in their outlook upon equality of roles. This conclusion is borne out by the findings from the crosstab where we can use the chi-square as a summary statistic of the entire table and note that the relatively low chi-square combined with a high p-value tells us that we cannot reject the null hypothesis of no relationship between gender and attitudes about equality of roles. Finally, we can look at the Z-score (in each cell below the percentage) and note that no cell has a Z-score greater than 2 suggesting that there is no response option in which men and women had a significant difference of opinion.
- A different picture emerges when we consider attitudes about the importance of the issue of women's equality. This analysis only included those that hold egalitarian gender role attitudes. We find that there is a difference in mean response between males (2.30) and females (1.84) and that this difference is statistically significant because the absolute value of the t-statistics is greater than 2 (actually greater than 5 in this analysis). When we look at the crosstab, we notice that 62.3% of women in our subsample feel that the issue of women's equality is an extremely important issue compared with 36.7% of men.
Summary
The goal of this exercise was to explore the differences between men and women with respect to public opinion in areas important to political decision-making. Taken as a whole, the above analyses suggest that there are real and significant gender-based differences in party identification and on attitudes toward public and social policy. Further research might consider why a gender gap exists on some issues. This research might take into consideration context as well as measurement and question design. Future research might also consider examining the gender gap over time. It would be interesting to see if and how much historical context affects the size of this gap. One possible approach is to explore any possible differences between the 2004 ANES data and previous ANES studies.
CITATION: Inter-university Consortium for Political and Social Research. Exploring the Gender Gap: A Data-Driven Learning Guide. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2009-04-16. Doi: https://doi.org/10.3886/gendergap
This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.