The Marriage Gap: Fact or Fiction?
Jim Tinnick
Saint Francis University/University of Pittsburgh
A 2001 Democratic Leadership Council article explores the “marriage gap,” the theory that married voters favor Republicans, while unmarried voters lean Democratic. The DLC’s rationale for this trend is “family values” – married voters stress issues like television sex and violence and school safety, while unmarried voters do not.
The DLC article raises many questions. Do political differences exist between married and unmarried individuals? Do divorced or separated individuals hold different opinions than the married or the never married? Using National Election Studies data, this paper will examine differences between married and unmarried individuals, and determine whether marital status effects electoral decisions.
Introduction
“Gap” studies have become quite common in social science over the years. Political scientists, for example, have made much of the “gender gap,” in which women tend to favor Democratic candidates and men tend to favor Republicans. More recent studies have focused on the existence of a “marriage gap,” where it is hypothesized that married voters tend to vote Republican in national elections, while unmarried voters tend to vote Democratic. Empirical studies of this trend have yielded mixed results (Weisberg, 1987; Plutzer and McBurnett 1991). Nonetheless, political consultants and pollsters continue to focus on the marriage gap. Greenberg (2001), in a piece for the Democratic Leadership Council, argues that the Democratic Party needs to make gains among married voters by focusing on “family values” issues. Rauch (2001) follows a similar tack, opining that “marriage is displacing both income and race as the great class divide of the new century.” Clearly, political practitioners hold that marriage matters.
Greenberg’s article inspired this piece, as it raised several questions. First, Greenberg notes the gaps between married and unmarried voters in the 1998 and 2000 elections: In 1998, Democrats received support from 44 percent of married voters, compared to 60 percent of unmarried voters; In 2000, Vice President Gore received support from 44 percent of married voters, compared to 57 percent of unmarried voters. While on the surface this appears to be evidence of a marriage gap, it is fair to ask whether this gap is real or caused by other factors – race or income, for instance (Weisberg 1987). Therefore, this paper seeks to revisit the question of the marriage gap with recent data.
Second, most studies of the marriage gap have focused on vote results as their dependent variable, with martial status as the key independent variable. But this focus on just voting differences between marital groups might be masking other trends that some studies see as important. Greenberg, for instance, offers “family values” attitudes as a partial explanation for the marriage gap. This stance, however, assumes that there is indeed some difference between marital groups on political issues. This paper will also examine whether issue-specific differences exist between martial groups.
The Political Importance of the Marriage Gap
Marriage has taken on considerable political importance of late, both as a political issue and as a political cleavage. In 2001, marriage itself was an issue in the debate over the reauthorization of the 1996 welfare reform law (Rauch 2001). Rising divorce rates have led some states to institute “covenant” marriages. A new single’s-rights group – The American Association for Single People (AASP)– has even formed, citing “single’s discrimination” in employment, housing, insurance and credit (State Laws Against Martial Discrimination).
With marriage taking on salience as an issue, scholars and pollsters alike are examining marriage as a potential political cleavage. The AASP reports that there are currently 82 million unmarried American adults – about 40% of the adult population (Facts About Single People), and Greenberg notes that 65 percent of the electorate is married.
Weisberg (1987) offered the first scholarly treatment of the marriage gap. His study showed an apparent “marriage gap” in presidential elections dating back to 1972. This is particularly important to Weisberg as the “proportion of unmarried voters doubled from about 18% of the public in 1964 to 36% in 1984.” Nonetheless, Weisberg expresses reservations about the conclusion that marital status causes individuals to vote differently:
We should be suspicious of whether the marital voting differences are totally due to changes in marriage situations. This raises the issue of whether the marriage gap is simply a reflection of other voting differences.
In short, Weisberg argues that the marriage gap may be nothing more than a spurious relationship between marriage, vote choice, and other factors such as race, gender, age and income. Analyzing National Election Studies data, this is indeed what he finds: “The marriage gap… disappears completely when [race and family income] are taken into account.”
Plutzer and McBurnett (1991) reexamine the marriage gap, altering Weisberg’s approach in several ways. First, Weisberg’s study dichotomized marital status into married and unmarried groups. Plutzer and McBurnett instead examined martial status in four different ways: a married/unmarried dichotomy; a married/never married/previously married trichotomy; a four-category array including traditional married (homemaker wife), nontraditional married (employed wife), never married, and previously married; and a five-category approach including traditional married (homemaker wife), nontraditional married (employed wife), never married, widowed and divorced. Secondly, as their dependent variable (vote choice) is dichotomous, Plutzer and McBurnett utilize logistic regression rather than OLS regression. Using this approach, Plutzer and McBurnett analyzed NES data for 1972, 1976, 1980, 1984, and 1988 with each of their models. They found marital status effects for only the 1972 and 1984 elections, but argued that marital status trends in the United States necessitated a close monitoring of the marriage gap in future elections.
Divorce and Separation
Weisberg ignores divorce altogether in his article, while Plutzer and McBurnett include it. Divorce is an issue of considerable social importance in the United States. Furstenberg (1990) points out that the rate of divorce in the United States rose tenfold from 1890 to 1990, and estimates that perhaps two-thirds of all marriages from the 1980s will end in divorce. Given that such large segments of the population are divorced, it is important to examine the divorcees’ political attitudes, and look for differences between never married, married, and divorced voters. Stoker and Jennings (1995) found that disruptions in marital status altered the political behavior of individuals: “Changes in marital status typically force people to undergo a period of substantial adjustment… participation rates tended to be depressed in the wake of changes in marital status.” Given these findings, an examination of the effects of marital status change on ideology may be warranted as well.
Data and Methods: Vote Choice
The data for this paper come from the National Elections Studies Archives, to approximate a replication of the Plutzer and McBurnett study for the years 1996 and 2000. A total replication, unfortunately, is not possible, due to several NES questionnaire changes: the 1996 NES did not ask about children living at home for panel participants; the 1996 NES did not ask respondents about the size of their hometown; and the 2000 NES did not inquire about spouses’ employment, one of the key independent variables in Plutzer and McBurnett’s five model system. This paper, then, is forced to compare fewer models to the baseline model.
The dependent variable for this study is the two-party vote for president in 1996 and 2000 (Democrats = 1, Republicans = 0). Since the dependent variable is dichotomous, the best approach is to follow Plutzer and McBurnett and utilize logistic regression. Independent variables for all models include dummy variables for race (nonwhite = 1), region (South, Midwest, and West included, East excluded), and homeownership (homeownership = 1, renting or “other” = 0). Each model also includes interval measures of income (24 categories in 1996 and 22 categories in 2000, slightly different than Plutzer and McBurnett’s 17), age in years, education in years, and a seven-point summary of partisan identification. Cases missing data for the dummy variables are excluded from the analysis, and missing income, education, and partisanship data are recoded with the series mean.
Four models are estimated for each year: a baseline model containing the independent variables listed above and making no reference to marriage. Model M1 follows Weisberg’s example and includes a dummy variable for marriage (married = 1, unmarried = 0). Model M2 refines the measure of marriage by adding a dummy for “previously married” (widowed, divorced, or separated = 1). Model M3 refines this even further, incorporating separate dummies for widowhood and for marital disruption (divorced and separated together).
Results and Interpretation: Vote Choice
Insert Table 1 Here
The goodness-of-fit results shown in Table 1 closely parallel the findings of Plutzer and McBurnett. In both 1996 (as well as 1976, 1980, and 1988 in the Plutzer and McBurnett study) the baseline model outperforms the models that include a marital status variable. Lower chi-square values (-2 times the likelihood ratio) show improvements in the model (and are thus preferred), if the chi-square improvement is significant over the difference in the model’s degrees of freedom. In 1996, the largest chi-square difference (between M0 and M1) of .685 falls far short of any acceptable level of statistical significance.
The 2000 models turn out differently. Model M1’s chi-square outperforms the baseline model by a difference of 6.257, which exceeds the critical value of 5.024 for the .025 level of significance over one degree of freedom, and falls just short of the critical value of 6.635 for the .01 level. No other difference approaches significance. This is in line with Plutzer and McBurnett’s findings, as they found that the addition of the marital status variables improved their 1972 and 1984 models.
Since the 2000 model is improved by the addition of the married dummy variable, it is worth further examination. Table 2 reports the logistic regression coefficients for the M1 model.
Insert Table 2 Here
The marriage coefficient, reported in logged odds for logistic regression, is -.490. Converting the logged odds into percentage change ([%Δ = eb – 1] * 100; see Pampel, 2000, 23) yields –39%. This means that the odds of married respondents voting for Gore in the 2000 were 39% lower than for unmarried respondents.
A cursory look at model M1 for 1996 (output not shown) yielded no significant effects for marriage (p = .408). Given the small improvement made in the model’s chi-square when the marriage variables were included, this is not unexpected.
Why marriage proves significant in 2000 and insignificant in 1996 is difficult to ascertain. Several possible explanations exist. One lies in the three-way nature of the 1996 presidential election. If Perot drew disproportionately from the Republican base, he would have blunted the Dole’s advantage among married voters. Another possible explanation is Clinton’s electoral strategy in 1996. Clinton’s “triangulation” approach (see Morris, 2000) led him to talk about family values: school uniforms, the V-chip, et cetera. If Greenberg’s estimation that the marriage gap is driven by family values is correct, then Clinton’s ability to undermine Dole’s family values base and garner support from the married would diminish the differences between married and unmarried voters.
Data and Methods: Marital Status and Political Issues
Outside of voting, does marital status make a difference? Greenberg argues that:
In a period of economic prosperity, issues that normally dominate electoral politics receded, and the election [of 2000] was fought on the cultural terrain of morality and values – from guns to gays, to character and leadership, and more broadly, the moral state of the nation.
This section of the paper focuses on political issues and marital status. Of interest are various economic and social issues, as well as different types of marital status: married, unmarried, divorced/separated, and widowed. The data are from the 2000 NES survey. The primary analytical tool used here will be OLS regression, as the variables are not dichotomous.
The independent variables for these models are as follows: the respondent’s age in years; the respondent’s education in years; the respondent’s household income (22 categories, with missing data recoded to the series mean); a seven point measure of partisan identification; a dummy variable for race (1 = nonwhite); dummies for region (Midwest, South, and West; East excluded); and dummies for marital status (married, widowed, and divorced/separated; never married the excluded category).
The first dependent variable is the respondent’s preferences on a five-point government spending and services scale[1][1], with lower values indicating a preference for less government spending on fewer government services. The regression results are reported in Table 3.
Insert Table 3 Here
The results from the regression are somewhat surprising. The coefficients for marriage, divorce/separation, and widowhood are all positively signed, with marriage and divorce/separation significant in excess of the .05 level, and widowhood significant at .06. The positive signs of the coefficients show that those in all three marital status categories are more likely to prefer government spending. These results, however, are offset by the negative coefficients for age, education, and household income, all of which are highly significant and distributed over a larger range.
The next dependent variable is a five-point measure of whether the respondent believes that the government should provide a job for all individuals. Lower scores indicate that the respondent feels that the government should provide jobs, higher scores indicate that the respondent feels that “the government should let each person get ahead on their own.” The regression results are reported in Table 4.
Insert Table 4 Here
Here, the coefficients for marital status all prove insignificant. This is perhaps unsurprising, as the means for each marital category attained or exceeded the midpoint of three on this question (see the Appendix for a table of the mean scores on the examined scales).
The first two models focused on economic matters. However, if Greenberg is correct, the proper avenue for analysis of the marriage gap is the social domain. Table 5 shows the regression results for the gays in the military question (on a four-point scale), with lower results indicating a more favorable response to allowing gays to serve.
Insert Table 5 Here
According to the results, neither widowhood nor divorce or separation has an effect on attitudes toward gays in the military. Being married is also insignificant at the .05 level, although is passes muster at the .10 level (p = .096). The coefficient is positive, indicating that the married are less likely than the single to favor allowing homosexuals to serve in the armed forces.
The next social issue is gun control. Table 6 displays the OLS results for responses to a five-point measure of favoring gun restrictions, with lower responses favoring more difficult restrictions.
Insert Table 6 Here
Here, no marital status variable approaches statistical significance. An examination of the mean scores for this scale, however, reveals a tight clustering around the mean of 1.99 for all marital groups.
Table 7 reports the OLS results for attitudes toward capital punishment, gauged by a four-point scale. Lower scores indicate greater support for the death penalty. None of the martial status variables approach significance, although the mean scores indicate a fairly close clustering of the marital groups.
Insert Table 7 Here
Table 8 displays the OLS results for attitudes towards women’s roles on a five-point scale. None of the marital status variables are significant at the .05 level, although the widowhood variable approaches significance at the .10 level (p= .126). An examination of the means shows that widows and widowers are more apt to opine that “a woman’s place is in the home” than are other groups. Also worth noting in the results is the significance of the age variable, which is probably collinear with the widowhood variable.
Insert Table 8 Here
Table 9 shows the regression results for responses to a four-point abortion attitudes scale, with lower scores advocating more restrictions on abortion. The table shows that the married and the widowed support greater restrictions on abortion, with the results significant in excess of the .05 level. Divorce or separation, however, proves to be an insignificant predictor of abortion attitudes (p = .812).
Insert Table 9 Here
Table 10, next page, displays the regression results for respondents’ reaction to the statement “the newer lifestyles are contributing to the breakdown of our society.” On a five-point scale, lower scores indicate stronger agreement. The table shows that marriage and widowhood far exceed the .05 level of significance, and divorce or separation is significant at .06. In all cases, the sign is negative, showing that those who are or have been married are more likely to agree that “new lifestyles” are contributing to societal breakdown. A look at the mean scores for this question reinforces these findings as the never married and especially the “partnered, not married” categories have very high means.
Insert Table 10 Here
Finally, Table 11, next page, displays the OLS results for responses to the question “This country would have many fewer problems if there were more emphasis on traditional family ties.” This variable is set on a five-point scale, and lower scores indicate stronger agreement. As Table 11 shows, the results for all three marital status categories are highly significant with strong negative coefficients. Being or having been married leads individuals to favor, at least in principle, “family values.”
Insert Table 11 Here
Interpretation
This component of the paper examined the effects of marital status on political attitudes. The findings are mixed. On the two economic issues examined, marital status had an effect on government spending preferences (albeit the effects might easily be mitigated by age, education and income variable effects), but not on government-guaranteed jobs attitudes. Among specific social issues, weak effects among married voters were found for gays in the military; no effects were found for gun control; no effects were found for capital punishment; very weak effects were found among widowed voters on women’s roles (although this might be a function of age); and effects were found for the married and widowed, but not the divorced and separated, on abortion.
However, very strong effects were found for all marital categories on the “new lifestyles” and “family values” variables. Why would these value questions show strong effects, while specific issues fail to show any effects? One possibility might arise from the very vagueness of these two variables. Specific issues, like gun control or women’s roles, require the respondent to come to a reasoned decision. The mere raising of the “new lifestyles” and “family values” issues, however, might bring about an emotional reaction, “activating core values… from a normative perspective” (Barker, 2002, 31-32). One does not need to know specifics in order to feel strongly about these issues – in fact, the normative reaction brought about by priming these core values tends to neutralize issue sophistication (Barker, 2002, 31).
The divorce/separation variable proved significant in only three models: spending preferences, “new lifestyles” (at the .06 level), and “family values.” In all three of these models, marriage and widowhood had similar effects. A second look at Tables 9, 10, and 11, however, reveal an interesting trend. First, divorce or separation fail to predict attitudes toward abortion, while the marriage and widowhood strongly predict opinions on abortion. Secondly, on “new lifestyles” and “family values”, the coefficients though signed properly are a good deal weaker for divorce than for marriage, indicating less agreement with these statements about family values among the divorced. Could these findings be an indication of greater social liberalism among divorcees? A look at the mean scores for “new lifestyles” and “family values” do indicate a pattern of greater liberalism among the divorced and separated, however, this examination is cursory and clearly requires further research.
Summary
This paper has examined the marriage gap in two ways. First, this study replicated (as nearly as possible) and updated Plutzer and McBurnett’s examination of the marriage gap. The results were similar to Plutzer and McBurnett’s as well – clear marital status effects in some years (1972 and 1984 for Plutzer and McBurnett; 2000 for this study) but not in others (1976, 1980, 1988; 1996). And like Plutzer and McBurnett’s study, I am unable to identify a root cause for the marriage gap. It is clear that a real and not illusory gap exists, but it is far from clear why the gap emerges in some years and not in others.
The second part of this paper looked for differences between the different marital status groupings on various political questions. Again, the results were mixed. On some questions, marital status failed to have any impact, while on other matters all three examined categories (married, widowed and divorced/separated) reported strong effects. Only the abortion question yielded a clear-cut difference between marital status categories, although weak effects were found for the married on the gays in the military question (p = .096). It is possible, though, that these findings show some differences between the divorced and the married. On the “new lifestyles” and “family values” questions, the divorced produced weaker coefficients, and an examination of their mean scores on these questions showed a tendency toward the more liberal responses. Clearly, more research is needed to determine if these findings are real or residual.
Further Research
Many avenues exist for further empirical research within the marriage gap, apart from the already mentioned “new lifestyles” and “family values” differences between the divorced or separated and the married.
1992 The election of 1992 was excluded from this study. In order to replicate Plutzer and McBurnett, logistic regression had to be utilized. Logistic regression requires a dichotomous dependent variable, and 1992 was a three-way race. My solution to this in 1996 was simply to exclude Perot from the analysis and only work with the two-party vote. With a 49-41-8 split in the electorate, this was possible. The 43-35-19 split in 1992 was more problematic. With Perot gaining almost a fifth of the votes, it was more difficult to justify operating only from the two-party base. This raises the question: how are we to study marital effects in 1992 in a way comparable to the rest of the elections? Perhaps more fundamentally, given the unique nature of 1992, would including it in overall trends be of any use to scholars?
Family Values vs. Specific Issues
Another possible avenue to explore is the strong results found in the “family values” questions versus the weak results from the specific issue questions. Why does marital status seem to make such a difference with amorphous concepts but make none with specific issues? And what effects does this have on the electorate?
Sources of the Marriage Gap
The big question, however, remains the most elusive. Why is the marriage gap present in some elections but not in others? What is it that leads the married to favor Republicans, how can they take advantage of it, and what can Democrats do to neutralize these efforts? If Greenberg is right that this is primarily a family values effect, and if specific issue stances do not move marital status blocs, then it appears that controlling campaign discourse about family values would be enough (see Barker, 2002, 10). If this is indeed the case, Greenberg’s advice to Democratic candidates to take family values from Republicans seems quite sound.
Appendix
Table 1
Goodness-of-Fit for Logistic Regression Presidential Vote Models
-2*log (Likelihood Ratio)
Model df 1996 2000
M0 (baseline) 9 629.925 680.345
M1 (M0+married) 10 629.240 674.088
M2 (M0+married+
Prev. married) 11 629.117 673.585
M3 (M0+married+
Divorced/separated+
Widowed) 12 629.102 673.428
N 1028 1094
Table 2
Logistic Regression Equation for Model M1 for Election 2000
| | B | S.E. | Sig.
| Step | RACE | .776 | .311 | .013
| 1(a) | -------- | ------ | ---- | ----
| | MIDWEST | .065 | .300 | .828
| | -------- | ------ | ---- | ----
| | SOUTH | -.447 | .291 | .125
| | -------- | ------ | ---- | ----
| | WEST | .647 | .329 | .049
| | -------- | ------ | ---- | ----
| | Own/Rent | -.065 | .262 | .803
| | -------- | ------ | ---- | ----
| | Educ/Yrs | .071 | .046 | .124
| | -------- | ------ | ---- | ----
| | PARTYID | -1.180 | .066 | .000
| | -------- | ------ | ---- | ----
| | HH Income| -.011 | .033 | .748
| | -------- | ------ | ---- | ----
| | AGE | .001 | .007 | .913
| | -------- | ------ | ---- | ----
| | MARRIED | -.490 | .224 | .029
| | -------- | ------ | ---- | ----
| | Constant | 2.742 | .798 | .001
| | -------- | ------ | ---- | ----
N = 1094 L-square = 674.088
Table 3
OLS Regression Results for Government Spending and Services
| | Unstandardized Coefficients | t | Sig. |
| -------------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| 1 | (Constant) | 4.659 | .202 | 23.114 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | Respondent age | -6.761E-03 | .002 | -3.454 | .001 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | Highest grade completed | -3.977E-02 | .012 | -3.368 | .001 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | INCOME | -1.902E-02 | .008 | -2.254 | .024 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | Party ID | -.167 | .013 | -12.665 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | Race | .317 | .073 | 4.361 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | Midwest | -.137 | .082 | -1.680 | .093 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | South | -.167 | .077 | -2.165 | .031 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | West | -.111 | .084 | -1.329 | .184 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | married | .155 | .075 | 2.071 | .039 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | Divorced or Separated | .256 | .091 | 2.801 | .005 |
| | --------------------------- | ------------------------ | ---------- | ------- | ---- |
| | widowed | .242 | .126 | 1.923 | .055 |
|-------------------------------------------------------------------------------------------|
N = 1441; R-squared = .171
Table 4
OLS Regression Results for Government Provision of Jobs
| -------------------------------- | ------------------------------------- | ------ | ---- |
| | Unstandardized Coefficients | t | Sig. |
| -------------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| 1 | (Constant) | 2.088 | .241 | 8.681 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Respondent age | 6.102E-03 | .002 | 2.580 | .010 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Highest grade completed | 2.410E-02 | .014 | 1.715 | .087 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | INCOME | 1.926E-02 | .010 | 1.854 | .064 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Party ID | .166 | .016 | 10.232 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Race | -.409 | .086 | -4.749 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Midwest | .246 | .101 | 2.430 | .015 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | South | .298 | .094 | 3.158 | .002 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | West | .258 | .103 | 2.502 | .012 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | married | 7.367E-02 | .091 | .808 | .419 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Divorced or Separated | -.113 | .111 | -1.014 | .311 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | widowed | -.168 | .153 | -1.097 | .273 |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
N = 1524; R-square = .l33
Table 5
OLS Regression Results for Gays in the Military
| -------------------------------- | ------------------------------------- | ------ | ---- |
| | Unstandardized Coefficients | t | Sig. |
| -------------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| 1 | (Constant) | 1.928 | .196 | 9.845 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Y1x. Respondent age | 8.496E-03 | .002 | 4.343 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Y3. Highest grade completed | -6.828E-02 | .012 | -5.895 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | SMEAN(V000994) | -1.556E-02 | .009 | -1.765 | .078 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | SMEAN(PARTYID) | .141 | .014 | 10.412 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Race | .412 | .071 | 5.788 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Midwest | 9.071E-02 | .084 | 1.080 | .280 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | South | .243 | .079 | 3.083 | .002 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | West | -1.614E-02 | .086 | -.187 | .852 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | married | .124 | .075 | 1.664 | .096 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Divorced or Separated | -2.848E-04 | .092 | -.003 | .998 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | widowed | -2.861E-03 | .128 | -.022 | .982 |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
N = 1643; R-square = .125
Table 6
OLS Regression Results for Gun Control
| -------------------------------- | ------------------------------------- | ------ | ---- |
| | Unstandardized Coefficients | t | Sig. |
| -------------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| 1 | (Constant) | 1.982 | .181 | 10.973 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Respondent age | 5.440E-04 | .002 | .307 | .759 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Highest grade completed | -3.775E-02 | .011 | -3.582 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | INCOME | 2.405E-03 | .008 | .302 | .763 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Party ID | .135 | .012 | 10.912 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Race | -.122 | .064 | -1.894 | .058 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Midwest | .161 | .076 | 2.107 | .035 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | South | .276 | .072 | 3.830 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | West | .167 | .079 | 2.114 | .035 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | married | -2.981E-02 | .069 | -.433 | .665 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Divorced or Separated | -.111 | .084 | -1.325 | .185 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | widowed | -3.400E-02 | .115 | -.296 | .767 |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
N = 1715; R-square .090
Table 7
OLS Regression Results for the Death Penalty
| -------------------------------- | ------------------------------------- | ------ | ---- |
| | Unstandardized Coefficients | t | Sig. |
| -------------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| 1 | (Constant) | 1.241 | .197 | 6.292 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Y1x. Respondent age | 3.920E-03 | .002 | 2.016 | .044 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Y3. Highest grade completed | 5.185E-02 | .012 | 4.492 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | SMEAN(V000994) | -5.379E-03 | .009 | -.615 | .538 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | SMEAN(PARTYID) | -.107 | .014 | -7.788 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Race | .401 | .072 | 5.603 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Midwest | .214 | .085 | 2.506 | .012 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | South | -3.330E-02 | .080 | -.415 | .678 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | West | -6.933E-02 | .088 | -.790 | .430 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | married | -4.294E-02 | .075 | -.569 | .570 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Divorced or Separated | 3.131E-02 | .092 | .340 | .734 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | widowed | 1.302E-02 | .126 | .103 | .918 |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
N = 1655; R-square = .082
Table 8
OLS Regression Results for Women’s Roles
| -------------------------------- | ------------------------------------- | ------ | ---- |
| | Unstandardized Coefficients | t | Sig. |
| -------------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| 1 | (Constant) | 1.966 | .179 | 10.964 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Respondent age | 4.560E-03 | .002 | 2.605 | .009 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Highest grade completed | -5.185E-02 | .010 | -4.958 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | INCOME | -2.069E-02 | .008 | -2.627 | .009 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Party ID | 6.271E-02 | .012 | 5.126 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Race | 4.590E-02 | .064 | .716 | .474 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Midwest | 3.869E-02 | .076 | .512 | .608 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | South | 6.416E-02 | .071 | .902 | .367 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | West | 1.598E-02 | .078 | .204 | .838 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Married | 5.509E-02 | .068 | .808 | .419 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Divorced or Separated | -7.260E-02 | .083 | -.874 | .382 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Widowed | .174 | .114 | 1.531 | .126 |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
N = 1675; R-squared = .058
Table 9
OLS Regression Results for Abortion
| -------------------------- | ------------------------------------- | ------ | ---- |
| | Unstandardized Coefficients | t | Sig. |
| -------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------- | ------------------------ | ---------- | ------ | ---- |
| 1 | (Constant) | 2.707 | .174 | 15.579 | .000 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Respondent age | -2.927E-03 | .002 | -1.710 | .087 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Education | 6.120E-02 | .010 | 6.027 | .000 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Income | 2.974E-02 | .008 | 3.820 | .000 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Race | -.345 | .062 | -5.526 | .000 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | PARTYID | -9.254E-02 | .012 | -7.683 | .000 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Midwest | -.304 | .074 | -4.101 | .000 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | South | -.153 | .070 | -2.200 | .028 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | West | 5.416E-02 | .077 | .706 | .480 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Married | -.215 | .066 | -3.238 | .001 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Divorced or Separated | 1.916E-02 | .081 | .237 | .812 |
| | --------------------- | ------------------------ | ---------- | ------ | ---- |
| | Widowed | -.261 | .110 | -2.364 | .018 |
| -- | --------------------- | ------------------------ | ---------- | ------ | ---- |
N = 1708 R-square = .119
Table 10
OLS Regression Results for “New Lifestyles”
| -------------------------------- | ------------------------------------- | ------ | ---- |
| | Unstandardized Coefficients | t | Sig. |
| -------------------------------- | ------------------------ | ---------- | | |
| Model | B | Std. Error | | |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| 1 | (Constant) | 2.613 | .231 | 11.294 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Respondent age | -1.324E-02 | .002 | -5.934 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Highest grade completed | 6.536E-02 | .013 | 4.873 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | INCOME | -1.723E-03 | .010 | -.171 | .864 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Party ID | -.109 | .016 | -6.999 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Race | -.327 | .085 | -3.867 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Midwest | 2.050E-02 | .097 | .211 | .833 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | South | 9.805E-02 | .092 | 1.071 | .284 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | West | .324 | .100 | 3.229 | .001 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | married | -.323 | .088 | -3.672 | .000 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | Divorced or Separated | -.204 | .109 | -1.882 | .060 |
| | --------------------------- | ------------------------ | ---------- | ------ | ---- |
| | widowed | -.358 | .143 | -2.501 | .012 |
| -- | --------------------------- | ------------------------ | ---------- | ------ | ---- |
N = 1476; R-square = .121
Table 11
OLS Regression Results for “Family Values”
| -------------------------------- | ----------------------