The Partisan Gender Gap in the United States: A Generational Replacement?

The Partisan Gender Gap in the United States: A Generational Replacement? Abstract To what degree have generational differences contributed to partisan changes in the American electorate, and what role did they play in the emergence of the gender gap in party identification? This paper sheds light on parallel and contradicting partisan trends among subgroups of the American electorate that affected the partisan gender gap over the past decades. By unpacking the gap by region, race, and generations, the analysis reveals that the effect of generational replacement on the gender gap varied in terms of size and direction between different subgroups. While in the South newer generations of white women diverged from the New Deal generation, consequently having a greater effect on the gender gap, in the rest of the country shifts among white men affected the gap to a greater extent than shifts among white women. Among African Americans, a decline in Democratic support was shown among newer generations of men, but less so among women. The findings highlight the importance of such political and historical contexts, and raise questions about the future of the partisan gender gap as the New Deal generation is replaced. Gender differences in voting and partisanship—a gender gap—have characterized US elections since at least 1952. Although women today identify as Democrats more so than men (the “modern gender gap”; see Inglehart and Norris [2000]), the pattern was precisely the opposite in the 1950s (the “traditional gender gap”). Scholars have shown that the modern gender gap resulted from structural, economic, and cultural changes that affected the lives of women, shifting their political preferences toward more liberal and prosocial policies generally identified with the Democratic Party (e.g., Manza and Brooks 1998; Edlund and Pande 2002). But these female-centered explanations for the gap tell only part of the story. A closer look shows that it was men, especially white men, who changed their preferences and moved to the Republican Party in higher numbers (Kaufmann and Petrocik 1999; Norrander 1999). The emergence of the gender gap, it has been shown, belongs to a set of broader partisan changes and political realignments that took place in the United States in the twentieth century. One powerful mechanism for political changes of this kind is generational replacement, in which newer cohorts with distinctive political preferences enter the electorate and replace older cohorts. Indeed, most explanations for the current modern gender gap suggest that women of younger generations should have been more affected by those external conditions. Although young generations played a significant role in shaping political trends in the United States (e.g., Carlson 2008), the generational perspective of the gender gap in the United States has largely been overlooked, leaving lacunae in our understanding of how the gap originated and is manifested. This paper seeks to address these lacunae by exploring how generational differences contributed to partisan-related trends in the United States. The paper unpacks the gap by region and over time. The analysis shows how generational trends varied between gender, region, and race. The contribution of this study is therefore twofold. First, it connects the gender gap literature to not only the literature on party realignment and political socialization in the United States, but also the impact of generational difference on political outcomes. Second, the study moves from analyzing the gender gap to analyzing the intergenerational dynamics of partisanship within each gender. Utilizing the ANES dataset between 1952 and 2016, the study emphasizes the role played by the New Deal generation and, more importantly, by the newer generations in influencing the partisan gender gap in the United States. The Partisan Gender Gap in the United States The gender gap in voting and in party identification has been a stable phenomenon in American politics for a few decades now, with some fluctuations over the years (Mattei and Mattei 1998; Box-Steffensmeier, De Boef, and Lin 2004). The gap, which started out as a traditional gender gap reflecting those in other Western countries, changed during the 1960s with a shift among women toward the Democratic Party, resulting in a small modern gap. Similar to patterns seen in Europe (Inglehart and Norris 2000; Giger 2009), men’s support for the Democratic Party declined during the 1970s and 1980s, widening an already growing gender gap (Kaufmann and Petrocik 1999; Norrander 2008). Most studies on the United States and elsewhere focus on the period of the modern gap. Explanations are mostly women centered (Gidengil 2007), attributing the emergence of the modern gender gap to changing structural, economic, and cultural factors that affected women’s policy and party preferences. For example, the decline in marriage and higher divorce rates (Edlund and Pande 2002), changes in women’s participation in the labor market (Manza and Brooks 1998), and inequality in gender division of household work (Iversen and Rosenbluth 2006) all served to make women more independent but also more economically vulnerable (Abendschön and Steinmetz 2014). Other studies have emphasized cultural change such as the decline in religiosity (Emmenegger and Manow 2014) as well as the rise of feminism (Giger 2009). The literature on the US gender gap focuses on how it is shaped by issue preferences and cultural factors, for example, social welfare issues (Erie and Rein 1988) and their salience for each gender (Kaufmann and Petrocik 1999), evaluation of the national economy (Welch and Hibbing 1992; Chaney, Alvarez, and Nagler 1998), and values such as reproductive rights and female equality (Conover 1988; Kaufmann 2002). Studies also show that the modern gender gap emerged as a result of men deserting the Democratic Party, while women’s partisanship remained more stable over time (Kaufmann and Petrocik 1999; Norrander 2008). The emergence of the partisan gender gap therefore can be examined in light of overall partisan trends and political realignments, especially in southern states, where much of the gap was attributed to differences in partisan realignment between genders (Norrander 1999). What previous studies on the voting gender gap in the United States tended to overlook was the generational component of political trends, which over time can change overall preferences in society. The effect of generational replacement on the electoral gender gap was suggested by Inglehart and Norris (2000, 2003), who argued that younger women were more affected by modernization-related processes (see also Shorrocks [2018]). However, in the United States, a modern gender gap is pronounced within both young and old birth cohorts who were born many years before modernization processes took place.1 Yet, as this study claims, generational differences have played a role in fostering partisan changes within each gender, consequently affecting gender gap trends. Generations in the United States as a source of political change Any aggregate change in attitudes can be the product of two mechanisms: period effects and generational replacement. In the former, people of all cohorts change their preferences over time. In this case the Zeitgeist, the spirit of the times, affects all individuals alike, regardless of age. Where generational replacement occurs, older generations continue to hold on to old attitudes, while newer generations with distinctive preferences enter the electorate and cause an overall change. Here, the circumstances of each period have a powerful influence on young people, and these effects persist over the lifecycle of a generation’s members (Mannheim 1952; Neundorf and Niemi 2014). In political research specifically, scholars have acknowledged the impact of major political actors during the period when voters come of age. This impact on political preferences lasts throughout the voters’ lives (Tilley 2002). As voting for the same party over time strengthens attachment to that party (Dinas 2014), we would expect to identify generational effects on political preferences according to the prevailing major political actors during the critical years of each birth cohort (Ghitza and Gelman 2014). In times of change, political events are more likely to affect voters who have not yet had the chance to participate in political acts such as voting or are still not committed to a particular party (Miller 1992) than older people whose political preferences have been firmly established (Green, Palmquist, and Schickler 2002). This mechanism suggests that new generations could be agents of political change. While there are numerous ways to define generations depending on the context and goals of the research, this paper specifically follows the literature on partisanship in the United States, which distinguishes generations according to the period in which they politically came of age. Here, in brief, I discuss the sociopolitical circumstances underpinning the paper’s classification of generations. The literature on partisanship further develops these distinct circumstances. A vast body of literature emphasizes the 1930s and 1960s as periods of substantial turnover in mass partisanship. The partisan realignment of the 1930s, which occurred in light of Franklin D. Roosevelt’s presidency and the New Deal, started a period of Democratic dominance that lasted until the 1960s. The South, of course, outnumbered other regions in its Democratic support, and was in practice a one-party system (Black and Black 1987). As debates on segregation and other issues began to polarize the American electorate, and as the Democratic Party moved toward supporting antisegregation policies, white southerners started deserting the party. The partisan transition of the white South was a gradual process, in which Johnson’s Civil Rights Act (1964) and Voting Rights Act (1965) constituted a turning point. Outside the South, Democratic identification declined as well, especially during the 1980s, which raised questions about a new realignment toward a Republican-dominated era (Norpoth 1987). The next two decades showed that this was not the case, as the Democratic Party regained power during the 1990s and 2000s (Knuckey 2015). Based on the critical periods of partisan realignment in the United States, scholars treated people who started voting before 1932 as the pre–New Deal generation, and those who started voting between 1932 and 1963 as the New Deal generation. A third generation, usually comprising people who started voting in 1964 and afterward, is referred to as the post–New Deal generation (e.g., Miller 1992; Nadeau and Harold 1997). I split the members of the post–New Deal generation into two, calling those who started voting in 1980 and afterward the post-Sixties generation.2 GENDERED GENERATIONAL EFFECTS: RESEARCH STRATEGY AND GOALS This paper suggests that generational differences may differ according to gender. As Mannheim (1952) claimed, historical circumstances may affect subunits of the same birth cohort differently. Thus, women and men may have reacted differently to political circumstances such that generational differences among them were not identical (Shorrocks 2016). Another obvious distinction that needs to be made in the US context is between white and African American voters, as they derive from different political contexts and experiences and therefore may exhibit different generational effects (Luks and Elms 2005). The same can be said about regions—the political context of the South in particular requires that this region be analyzed separately. Previous studies deal mostly with the electoral gender gap as a dependent variable. Focusing on women, these studies treat men as the benchmark to which women’s behavior was compared. But if we acknowledge that women and men are different social groups in terms of political behavior, it would be just as reasonable to compare newer generations of women to their female predecessors as to males of their own generation. Women today are different from men of their age (hence the gap between them), but they are also different—in a different way—from women of older generations in terms of education, divorce rates, values, and so on. Of course, the same can be said about men, although only a few studies on the gender gap elaborate on the changes in the male side of the equation (e.g., Kaufmann 2002; Iversen and Rosenbluth 2006). Therefore, the analysis of this paper estimates generational trends among each gender separately in order to learn how each generation’s preferences changed over the years and compared to other generations within the same gender. Methods I utilize the ANES cumulative dataset for the period 1952–2012, to which I added the 2016 preelection survey. The dependent variable, party identification, was originally measured on a seven-point scale, in which respondents were asked their partisanship identification, ranging between Democrat, Independent, and Republican. The answers on this scale included: strong Democrat; weak Democrat; Independent–Democrat; Independent; Independent–Republican; weak Republican; and strong Republican. This variable was recoded into a dummy variable (1 = strong, weak, or leaning Democrat; 0 = strong, weak, or leaning Republican). Pure Independents were excluded from the sample. Some studies excluded leaners and focused on weak and strong partisans only. However, Norrander (1997) showed that there is an independence gap between men and women, according to which men display a greater probability of identifying as leaners while women identify as weak partisans. It is thus important to include both categories in the study. The sample is divided into the southern and nonsouthern United States. The South includes 11 exfederation states: Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Texas, Tennessee, and Virginia. The sample is also divided by race, with the analysis first focus on white respondents, and a complementary analysis of African Americans. METHOD OF ANALYSIS Given the absence of panel data for an extended period of time, this study, like most age-period-cohort (APC) studies, makes use of cross-sectional setups. Such setups make it possible to track people who belong to the same birth cohort over time and to treat them as if they represent a sample of this birth cohort at each point in time (“synthetic cohorts”) (Alwin and Krosnick 1991). The current study employs a descriptive approach side by side with APC regression models. The descriptive method, which presents the total outcome for each generation at each point in time, is the most straightforward way of telling the story of each generation over its life cycle. It helps in identifying the periods in which generations behaved similarly or dissimilarly, as well as critical periods of major change within generations (Firebaugh 1997; Glenn 2005). Yet, this method does not control for the two main confounders—period and age3—or for other compositional differences that result in observed variation between cohorts (Yang and Land 2013). Therefore, I use APC regression models that provide cohort effects that are estimated over all periods and ages4 for three purposes: first, to look for empirical evidence for distinct tendencies that characterize each cohort at all ages and in all periods (Yang and Land 2008); second, to examine whether compositional differences in demographic characteristics between cohorts account for the differences found in partisan preferences (Smets and Neundorf 2014); and third, to evaluate generational effects in a more meaningful way using a simulation of generational replacement. Results FROM THE GENDER GAP TO GENDERED PARTISAN TRENDS What did the partisan gender gap look like among white Americans in each region? Figure 1 presents the percent of Democratic identifiers out of all partisans among white men and women between 1952 and 2016. The gender gaps portrayed are quite similar. In both regions, there was a period of traditional gender gap, of more women than men being Republican, which was followed by a transformation of the gap into its modern form. Figure 1. View largeDownload slide Partisan trends among white men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 1. View largeDownload slide Partisan trends among white men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. However, when moving from the gap to describing men’s and women’s partisan trends in each region, two different stories emerge. Prior to 1960, more white men than women in the South were Democratic, but the rapid decline in Democratic identification, which was less steep among women, caused the gender gap to reverse. Thus, after 1960, women identified with the Democratic Party in higher numbers than men. Outside the South, the partisan gap among white voters emerged through the two following transitions. The first revolved around women, who exhibited an increase in their Democratic support from the 1950s to the 1960s. Before 1960 the average Democratic support among white nonsouthern women was 51.8 percent, while in the years 1960–1968 the average was 55.3 percent. Yet a greater impact on the emergence of the modern gender gap was that of men’s defection from the Democratic Party. Their Democratic support, which was 55.4 percent on average during the 1970s, declined to the average level of 46.5 percent during the 1980s. This description of partisan trends, however, treats each group as composed of the same people, who might or might not have changed their preferences between periods. The next section examines an alternative portrayal of the electorate as composed of different generations that enter and leave the electorate over the years. Figure 2 unpacks partisan trends of southern white men and women according to the four generations described above: Pre–New Deal generation, New Deal generation, post–New Deal generation, and post-Sixties generation. The figure shows that at the beginning and among both genders, the New Deal generation exhibited trends similar to those of its predecessor, the pre–New Deal generation. During the 1950s and 1960s, these two older generations experienced a decline in Democratic support, while the next generation exhibited even lower levels of Democratic support. The members of this post–New Deal generation started voting when southern white voters massively left the Democratic Party. The young voters of the 1960s and afterward might have acquired their partisanship in Democratic families, but by the time they reached adulthood their parents’ generation had already started abandoning this political home. The graphs also show that the members of the newest generation, who came of age during the Republican years of the 1980s, were more Republican than all other cohorts in the beginning, although they did not become more Republican with time. Figure 2. View largeDownload slide Partisanship among white southern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long dash light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 2. View largeDownload slide Partisanship among white southern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long dash light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. While the graphs show similar generational stories for both genders, men in all cohorts were less likely to identify as Democrats compared to women. One explanation suggested by Schreckhise and Shields (2003) was that men were more affected by racial issues compared to women, who were more influenced by socioeconomic issues that made them more pro-redistribution and therefore more Democratic. Regarding white women outside the South (figure 3), the analysis clearly shows a Republican tendency of the oldest pre–New Deal generation. The following New Deal generation was much more Democratic than its predecessor at that time, and remained Democratic over most of its life cycle. Interestingly, though, this generation showed a decline in Democratic support in the years 2012 and 2016. Figure 3. View largeDownload slide Partisanship among white nonsouthern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long-dashed light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 3. View largeDownload slide Partisanship among white nonsouthern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long-dashed light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. The post–New Deal generation, which came of age in the second half of the 1960s and 1970s, was the main suspect for creating the modern gender gap. This new generation of women matured into adulthood during the second wave of feminism, when both women’s level of education and their share in the labor force rose. Indeed, at first, this generation was more likely to identify as Democrats compared to women of previous generations. However, the percentage of women identifying as Democrats among the post–New Deal generation continually declined and reached average levels among women during the 1990s and 2000s. The newest generation, which started voting in a rather right-wing period, was more Republican at the beginning, but then exhibited an increase in Democratic support over the 2000s. Regarding the overall decline in Democratic support among men, figure 3 provides some evidence for generational differences. The defection from the Democratic Party is pronounced by members of the New Deal generation, but this change was not great compared to the following generations. The post–New Deal generation, which started as more Democratic compared to other generations at that time, experienced a decline in Democratic support through the 1970s and 1980s. The next generation, who started voting in Reagan’s era, was even more Republican than previous generations. Interestingly, in the 2000s this generation showed a considerable increase in Democratic identification. BEYOND TIME: ESTIMATING NET GENERATIONAL EFFECTS In order to more precisely measure generational differences while controlling for period and age, I ran four logistic regression models separately for each of the subgroups described in the previous section: white southern men, white southern women, white nonsouthern men, and white nonsouthern women. The dependent variable in all models is Democratic identification.5 Independent variables include age and its second and third polynomials, three dummy variables for generations with the New Deal generation as the reference category, and survey-year dummies. The regression equation for each model is as follows: Logit(Democrat)=β0+β1Age+β2Age2+β3Age3+β4PreNewDeal+                                      β5PostNewDeal+β6PostSixties+∑j=127γjYearj As shown in table 1, on average, the New Deal generation was more Democratic than both its predecessors and followers, as indicated by the negative coefficients for most generational effects. However, only some of these generational effects are statistically significant. Table 1. Generational effects on democratic partisanship White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 Note.—Logistic regression models in which the dependent variable is Democratic identification. Reference categories are as follows. Generations: New Deal generation. Survey year: 1952. Full results including year coefficients are presented in table A1 in the Online Appendix. Standard errors in parentheses. Source.—ANES data 1952–2016. * p < 0.05. View Large Table 1. Generational effects on democratic partisanship White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 Note.—Logistic regression models in which the dependent variable is Democratic identification. Reference categories are as follows. Generations: New Deal generation. Survey year: 1952. Full results including year coefficients are presented in table A1 in the Online Appendix. Standard errors in parentheses. Source.—ANES data 1952–2016. * p < 0.05. View Large In the South (models 1 and 2), generational replacement has been more pronounced among women compared to men. While for white women the coefficients for the two youngest generations are statistically significant, among men, only the post–New Deal is significantly less Democratic than the New Deal generation (p < .05). This difference between women and men probably has to do with the greater loyalty of the New Deal generation to the Democratic Party on the female side. On average, southern white women of the New Deal generation did not defect from the Democratic Party as much as men from the same generation, at least up to the beginning of the 2000s (see their moderate slope in this period in figure 2). Next, the idea of a generational replacement accounting for men’s decline in Democratic support outside the South is only partially supported (models 3 and 4). That is, the post-Sixties generation among white men has significantly lower Democratic support compared to the New Deal generation, but the post–New Deal does not. Nevertheless, the picture of the average effect hides the trend of decreasing Democratic support among the post–New Deal generation during the 1980s, which can be detected in figure 3. Among women, the New Deal generation emerges as significantly more Democratic than its predecessor. The younger generations, however, show mixed results, depending on gender. Among white women, the post–New Deal generation shows higher levels of Democratic support compared to the New Deal generation (their coefficient is positive), but this result is not significant. The post-Sixties generation was on average less Democratic than the New Deal generation, but again this result is not significant. This finding confirms the idea of generational convergence among nonsouthern white women. Do these generational differences stem from intergenerational differences in demographic composition, or do they hold when controlling for demographic factors? To answer this question, I ran the same four models of table 1 with additional controls for college degree, income, work status, and marital status, four demographic variables that are available in the ANES throughout the period of this research. The results, which can be found in table A3 in the Online Appendix, show that the pattern of generational effects that emerged from table 1 persists. AFRICAN AMERICANS, GENERATIONS, AND THE PARTISAN GENDER GAP While some studies do not find gender gap in party identification among African Americans (e.g., Conway 2008), Junn (2017) emphasizes the role of women of color in fostering and stimulating the gender gap in voting in the United States. This does not mean that men of color are not Democratic overall, but that increasing turnout levels and demographic changes made women of color an important subgroup of voters affecting voting gap between women and men in the United States. Although sample sizes of African Americans in the ANES do not allow us to separate this subgroup into regions, the analysis of overall partisan trends among black women and men, and in particular its generational aspect, carry important insights about the gender gap in the United States. The partisan gender gap among African Americans since 1980 was 3.7 percent on average according to the ANES data, but it is not statistically significant, probably resulting from the small number of respondents in each election year. Figure 4 shows that an increase in Democratic support occurred among both genders during the 1960s and 1970s. Similarly to nonsouthern whites, a decline in Democratic support among black men widened the gender gap. Figure 4. View largeDownload slide Partisan trends among African American men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 4. View largeDownload slide Partisan trends among African American men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. To estimate generational differences among each gender, I ran an APC analysis (models 5 and 6 in table 1), which shows that all coefficients are negative, thus all generations are less Democratic than the New Deal generation, but only the pre–New Deal generation among men is significantly significant (p < 0.05). As the generational categorization suggested above is not necessarily relevant for African Americans, I conducted a robustness check of generational trends using Generalized Additive Models (see table B1 and figure B3 in the Online Appendix). The results show that newer cohorts of black women were a bit less likely to be Democratic, but the confidence intervals around these effects are quite large. The picture of generational effects among black men is clearer, with members of older cohorts more likely to be Democrats, and members of younger cohorts less likely. BACK TO THE GENDER GAP: A GENERATIONAL REPLACEMENT? The last empirical section estimates to what degree generational replacement alone affected the total gender gap among white Americans over the past decades.6 This analysis examines the impact of generational replacement on party identification, using a simulation in which newer generations hold the same political preferences as older generations.7 In particular, the simulation here investigates what partisan trends would look like if the post–New Deal and post-Sixties generations had party identification levels identical to the New Deal generation. Figure 5 presents the results of the simulation for southern whites on the left and for nonsouthern whites on the right. The bars represent the difference between the predicted percent of Democratic identifiers based on the original model and the predicted percent of Democratic identifiers based on the simulation, for each decade. The dark gray bars stand for women, and light gray bars for men. Whiskers mark 95 percent confidence intervals.8 Figure 5. View largeDownload slide The difference in partisan trends in the absence of generational replacement. The bars represent the difference between the predicted percent of Democratic identifiers based on the original APC model and the predicted percent of Democratic identifiers based on the simulation (i.e., no generational replacement), for each decade (table 1, models 1–2 for the left panel and models 3–4 for the right panel). Positive gaps indicate by how much the level of Democratic support would have been higher in the case of no generational replacement. Dark gray bars stand for women, and light gray bars for men. Whiskers mark 95 percent confidence intervals. Figure 5. View largeDownload slide The difference in partisan trends in the absence of generational replacement. The bars represent the difference between the predicted percent of Democratic identifiers based on the original APC model and the predicted percent of Democratic identifiers based on the simulation (i.e., no generational replacement), for each decade (table 1, models 1–2 for the left panel and models 3–4 for the right panel). Positive gaps indicate by how much the level of Democratic support would have been higher in the case of no generational replacement. Dark gray bars stand for women, and light gray bars for men. Whiskers mark 95 percent confidence intervals. As can be seen, almost all bars show positive differences, which means that if the two newer generations had the same preferences as the New Deal generation, white women and men in both regions would have supported the Democratic Party in higher numbers. The graph clearly shows the differences between regions. In the South, generational differences were larger among white women, while outside the South white men exhibited larger intergenerational differences compared to women. In particular, white women in the South would have been more Democratic by six percentage points in the 1990s had no generational replacement taken place. This difference was even larger in the following years and reached the level of an additional eight percentage points in the 2010s. Generational replacement affected white southern men as well, with six percentage points higher Democratic support in the 2000s had no generational replacement taken place. However, not all these differences are statistically significant. Outside the South, the picture is the opposite. White men’s partisanship was more affected by generational replacement compared to women’s. If newer generations had political preferences identical to the New Deal generation, white men’s support for the Democratic Party would have been higher by three percentage points in the 2000s and by four percentage points in the 2010s. White nonsouthern women would also have been a bit more Democratic if no replacement had taken place, but these differences are smaller and not statistically significant. The last part of the analysis asks what the partisan gender gap among whites would look like if no generational replacement had taken place. To answer this question, I went back to the raw data and calculated the partisan gender gap found in the ANES for each decade since 1980—with and without generational replacement.9 Findings show that if generational replacement had not taken place, outside the South the gap between white women and men would have been consistently smaller. The entrance of newer generations of men who did not identify with the Democratic Party as their predecessors had done enlarged the gap between white women (who remained loyal to the Democratic Party) and men. In the 1990s, for example, the gender gap could have been smaller by 5.5 percentage points if generational replacement had not taken place. The impact of generational differences diminished in the 2000s due to the increase in Democratic support among white men of the youngest generation (see their trend line in figure 3). Interestingly, a sharp decline in Democratic support among white women of the New Deal generation in the 2012 and 2016 elections, which was shown in figure 3, caused a reverse gender gap among this generation in that period. This finding demonstrates how cohorts of voters can change their political preferences at older ages, even if only temporarily. The results concerning southern whites are mixed. Up to 2000, the gap among the New Deal generation was smaller than the total gap. In the 2000s, the gender gap among the New Deal generation increased as its men departed from the Democratic tendencies prevalent among their female counterparts. At the same time, the gender gap among younger generations declined due to younger women becoming less Democratic and more similar to their male counterparts. The result of these trends was a smaller gender gap in the 2000s compared to what could have happened to the gap had newer generations held the preferences of the New Deal generation.10 Conclusions This paper analyzes several aspects of the origins of the partisan gender gap in the United States, asking to what degree generational replacement contributed to the emergence of the gap. The analysis, which unpacked the gap by gender, region, race, and generation, reveals differences between these subgroups that affected the gender gap within each region/race. In particular, the analysis highlights the role of the New Deal generation, which came of age during a period of Democratic dominance, and the extent to which generations that came afterward diverged (or not) from this generation’s Democrat tendency. The analysis shows that among white women, younger generations in the South were significantly less Democratic than women of the older New Deal generation. Outside the South, white women of the New Deal generation were clearly more Democratic than their predecessors, but their followers did not show much different party preferences and were only a bit less Democratic. Nonsouthern white men were the ones who exhibited a decline in support for the Democrats, shared by newer generations as well, which narrowed the gender gap among whites in that region. Among African Americans, and similarly to nonsouthern whites, generational differences affected men to a greater extent. Younger generations of black men were less Democratic than older black voters, but among black women there was no substantial decline in support for the Democrats along cohorts. Three main points arise from these findings. First, they demonstrate how important it is to unpack the quantity called the gender gap into men’s and women’s separate preferences, especially when studying transformations therein over time. Although the gap is important in itself, it might conceal variations within each gender—and other subdivisions—that explain central trends in American politics. Second, although a lot of attention has been given to the role of women, and to changes that occurred on the female side of the equation, the results show that some trends are attributed to men, and those that are attributed to women are in the opposite direction from the one that might be expected, as women of newer generations showed diminished Democratic support compared to women of older generations. Third, some political transitions are attributable to the entrance of newer generations and some are not. Only by separating those trends over time and within subgroups can we identify when generational replacement mattered and when it did not. This kind of perspective is missing for the youngest birth cohorts, some of which show very high levels of Democratic identification. If these are meaningful trends representing generational phenomena or a short-term effect, only time will tell. Appendix This article utilizes the ANES Cumulative Time-Series data set from 1952 to 2012 and the ANES pre-election survey for 2016. The most common ANES study design is a cross-section, equal-probability sample, although in some surveys they oversampled certain subpopulations. Additional information can be found online at http://www.electionstudies.org/studypages/anes_timeseries_cdf/anes_timeseries_cdf.htm. I utilize the weights provided by the ANES [variable: VCF0009z] and the weighting variable V160101 for the 2016 survey. Most of the surveys were carried out face-to-face, except for 2012 and 2016 in which an internet based survey was added. The response rates are as follows (http://www.electionstudies.org/overview/dataqual.htm): 1952: 77.2; 1956:85.5; 1958: 78.1; 1960 and 1962: not ascertained; 1964: 80.6; 1966: 77.1; 1968: 77.4; 1970: 76.6; 1972: 75.0; 1974: 70.0; 1976: 70.4; 1978: 68.9; 1980: 71.8; 1982: 72.3; 1984:72.1; 1986: 67.7; 1988: 70.5; 1990: 70.6; 1992: 74.0; 1994: 72.1; 1996: 59.8; 1998:63.8; 2000: 60.5; 2002: 55.3; 2004: 66.1; 2008: 59.5; 2012: 38.0 for the face-to-face component and 2.0 for the Internet component; 2016: 50.0 for the face-to-face component and 44.0 for the Internet component. Measures 1952-2012 cumulative file Partisanship (VCF0301). Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or what? (IF REPUBLICAN OR DEMOCRAT) Would you call yourself a strong (REP/DEM) or a not very strong (REP/DEM)? (IF INDEPENDENT, OTHER [1966 AND LATER: OR NO PREFERENCE; 2008: OR DK) Do you think of yourself as closer to the Republican or Democratic party? (Strong Democrat, Weak Democrat, Independent–Democrat, Independent–Independent, Independent–Republican, Weak Republican, Strong Republican) Race (VCF0106). What race or ethnic group or groups best describes you? (white; black; other) Education (VCF0110). What is the highest grade of school or year of college you have completed? (grade school or less (0–8 grades); high school (12 grades or fewer, incl. non-college training); some college (13 grades or more but no degree); college or advanced degree) Household income (VCF0114). Information about income is very important to understand how people are doing financially these days. Your answers are confidential. Would you please give your best guess? [responses recoded by ANES in the Cumulative File] (0 to 16 percentile; 17 to 33 percentile; 34 to 67 percentile; 68 to 95 percentile; 96 to 100 percentile) Male (VCF0104). Do you identify as a man or a woman? (male, female) Age (VCF0101). What is your date of birth? (age 17 and below and 97 and up were excluded). South. Using variable (VCF0901), individuals living in states included in the original Confederacy were coded 1 (Virginia, Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Texas, Tennessee); otherwise, 0. Work status (VCF0118). 1968-1970: Are you presently employed, or are you unemployed, or retired, (a housewife), (a student), or what? 1972-1978: (1972: We’d like to know if you are looking for work, working now) (1974- 1978: We’d like to know if you are working now, or are you unemployed,) retired, (a housewife) a (student), or what? (IF HOMEMAKER OR STUDENT) Are you doing any work for pay at the present time? (IF R IS HOMEMAKER OR STUDENT AND R IS WORKING FOR PAY:) About how many hours do you work on your job in the average week? 1980 AND LATER: We’d like to know if you are working now, temporarily laid off, or are you unemployed, retired, permanently disabled, (a homemaker), (a student), or what? Marital Status (VCF0147). 1952: Are you married? 1956-2004: Are you married now and living with your husband/ wife (2002: spouse)-- or are you widowed, divorced, separated, or have you never married? 2008: Are you married now and living with your husband/wife-- or are you widowed, divorced, separated, or have you never married? / Are you married, divorced, separated, widowed, or have you never been married? 2012: Are you now married, widowed, divorced, separated or never married? Are you currently living with a partner, or not? 2016 pre-election survey List of variables: Age (V161267), gender (V161342), race (V161310x), region (V161010e), partisanship (V161158x), income (V161361x), work status (V161276x), education (V161270), marital status (V161268). Supplementary Data Supplementary data are freely available at Public Opinion Quarterly online. The research for this article was completed while the author was a doctoral candidate in the Department of Political Science at Hebrew University of Jerusalem. For helpful comments and suggestions, the author thanks Sylvia Bashevkin, André Blais, Elisabeth Gidengil, Guy Grossman, Orit Kedar, Einat Lavy, Odelia Oshri, Michael Shalev, and Rosalind Shorrocks, as well as the editors and three anonymous reviewers. Footnotes 1. A modern gender gap is found among white nonsouthern voters within most cohorts who were born after 1906, while among white southerners this gap is observed within cohorts born after 1926. Among African Americans, the modern gender gap is found only among cohorts born after 1946 (with the exception of the 1956–1965 cohorts). 2. The relevant birth years for each generation according to their chronological order are: 1880–1907, 1908–1939, 1940–1958, and 1959–1990. I use different generational categorizations for robustness checks, which generally support this categorization. These are Generalized Additive Models (table B1) and 10-year interval cohorts per Firebaugh and Chen’s (1995) categorization (table B2). For more details see the Online Appendix, Section B. 3. Since age (life cycle) effects generally create political continuity at the societal level, they cannot account for an aggregate political change. Methodologically, age effects are confounded with cohort at each point in time and therefore need to be controlled for. 4. Every APC analysis encounters an inherent problem of identification (Mason and Fienberg 1985) due to the perfect collinearity between the three components. Various methods for dealing with this problem have been proposed over the years. Nonetheless, as Tilley and Evens (2014) noted, all APC methods carry assumptions that may or may not fit the data in a given case. 5. For robustness checks, I ran the analyses with the original seven-point scale variable. The main results were replicated, except for African American men, whose generational effects become significant using OLS model. See table A2 in the Online Appendix. 6. Sample sizes of African Americans are too small for this procedure. 7. I conducted this analysis by calculating the predicted probability to identify as a Democrat of each respondent based on the coefficients from the relevant regression model (table 1) and the observed values for that respondent (Hanmer and Kalkan 2013). Then I calculated another set of predicted probabilities; this time, newer generations are awarded the coefficient of an older cohort instead of the coefficient of their original one. 8. The standard errors of the difference were calculated via bootstrap over 1,000 draws. 9. The first was calculated as the total gender gap in each decade, and the second as the gender gap among the New Deal generation only. 10. Similarly to nonsouthern whites, a tremendous decline in Democratic support took place in the 2012 and 2016 elections among white southern women of the New Deal generation. This trend led to an inverse partisan gender gap among that generation. References Abendschön , Simone , and Stephanie Steinmetz . 2014 . “ The Gender Gap in Voting Revisited: Women’s Party Preferences in a European Context .” Social Politics 21 : 315 – 44 . Google Scholar CrossRef Search ADS Alwin , Duane F. , and Jon A. Krosnick . 1991 . “ Aging, Cohorts, and the Stability of Sociopolitical Orientations over the Life Span .” American Journal of Sociology 97 : 169 – 95 . Google Scholar CrossRef Search ADS Black , Earl , and Merle Black . 1987 . Politics and Society in the South . Cambridge, MA : Harvard University Press . Box-Steffensmeier , Janet M. , Suzanna De Boef , and Tse-Min Lin . 2004 . “ The Dynamics of the Partisan Gender Gap .” American Political Science Review 98 : 515 – 28 . Google Scholar CrossRef Search ADS Carlson , Elwood . 2008 . The Lucky Few: Between the Greatest Generation and the Baby Boom . London : Springer . Chaney , Carole K. , R. Michael Alvarez , and Jonathan Nagler . 1998 . “ Explaining the Gender Gap in U.S. Presidential Elections, 1980–1992 .” Political Research Quarterly 51 : 311 – 39 . Conover , Pamela Johnston . 1988 . “ Feminists and the Gender Gap .” Journal of Politics 50 : 985 – 1010 . Google Scholar CrossRef Search ADS Conway , M. Margaret . 2008 . “ The Gender Gap: A Comparison across Racial and Ethnic Groups .” In Voting the Gender Gap , edited by Lois Duke Whitaker , pp. 170 – 83 . Champaign : University of Illinois Press . Dinas , Elias . 2014 . “ Does Choice Bring Loyalty? Electoral Participation and the Development of Party Identification .” American Journal of Political Science 58 : 449 – 65 . Google Scholar CrossRef Search ADS Edlund , Lena , and Rohini Pande . 2002 . “ Why Have Women Become Left-Wing? The Political Gender Gap and the Decline in Marriage .” Quarterly Journal of Economics 117 : 917 – 61 . Google Scholar CrossRef Search ADS Emmenegger , Patrick , and Philip Manow . 2014 . “ Religion and the Gender Vote Gap: Women’s Changed Political Preferences from the 1970s to 2010 .” Politics & Society 42 : 166 – 93 . Google Scholar CrossRef Search ADS Erie , Steven P. , and Martin Rein . 1988 . “ Women and the Welfare State .” In The Politics of the Gender Gap: The Social Construction of Political Influence , edited by Carol M. Mueller , pp. 173 – 91 . Newbury Park, CA : Sage Publications . Firebaugh , Glenn . 1997 . Analyzing Repeated Surveys . Thousand Oaks, CA : Sage Publications . Google Scholar CrossRef Search ADS Firebaugh , Glenn , and Kevin Chen . 1995 . “ Vote Turnout of Nineteenth Amendment Women: The Enduring Effect of Disenfranchisement .” American Journal of Sociology 100 : 972 – 96 . Google Scholar CrossRef Search ADS Ghitza , Yair , and Andrew Gelman . 2014 . “ The Great Society, Reagan’s Revolution, and Generations of Presidential Voting .” Working Paper. Gidengil , Elisabeth . 2007 . “ Beyond the Gender Gap: Presidential Address to the Canadian Political Science Association, Saskatoon, 2007 .” Canadian Journal of Political Science 40 : 815 – 31 . Google Scholar CrossRef Search ADS Giger , Nathalie . 2009 . “ Towards a Modern Gender Gap in Europe? A Comparative Analysis of Voting Behavior in 12 Countries .” Social Science Journal 46 : 474 – 92 . Google Scholar CrossRef Search ADS Glenn , N. D . 2005 . Cohort Analysis . Thousand Oaks, CA .: Sage Publications . Google Scholar CrossRef Search ADS Green , Donald , Bradley Palmquist , and Eric Schickler . 2002 . Partisan Hearts and Minds . New Haven, CT : Yale University Press . Hanmer , Michael J. , and Kerem Ozan Kalkan . 2013 . “ Behind the Curve: Clarifying the Best Approach to Calculating Predicted Probabilities and Marginal Effects from Limited Dependent Variable Models .” American Journal of Political Science 57 : 263 – 77 . Google Scholar CrossRef Search ADS Inglehart , Ronald , and Pippa Norris . 2000 . “ The Developmental Theory of the Gender Gap: Women’s and Men’s Voting Behavior in Global Perspective .” International Political Science Review 21 : 441 – 63 . Google Scholar CrossRef Search ADS Inglehart , Ronald , and Pippa Norris . 2003 . Rising Tide: Gender Equality and Cultural Change around the World . Cambridge : Cambridge University Press . Google Scholar CrossRef Search ADS Iversen , Torben , and Frances Rosenbluth . 2006 . “ The Political Economy of Gender: Explaining Cross National Variation in the Gender Division of Labor and the Gender Voting Gap .” American Journal of Political Science 50 : 1 – 19 . Google Scholar CrossRef Search ADS Junn , Jane . 2017 . “ The Trump Majority: White Womanhood and the Making of Female Voters in the U.S .” Politics, Groups, and Identities 5 : 343 – 52 . Google Scholar CrossRef Search ADS Kaufmann , Karen M . 2002 . “ Culture Wars, Secular Realignment, and the Gender Gap in Party Identification .” Political Behavior 24 : 283 – 307 . Google Scholar CrossRef Search ADS Kaufmann , Karen M. , and John R. Petrocik . 1999 . “ The Changing Politics of American Men: Understanding the Sources of the Gender Gap .” American Journal of Political Science 43 : 864 – 87 . Google Scholar CrossRef Search ADS Knuckey , Jonathan . 2015 . “ The Survival of the Democratic Party Outside the South: An Update and Reassessment .” Party Politics 21 : 539 – 52 . Google Scholar CrossRef Search ADS Luks , Samantha , and Laurel Elms . 2005 . “ African-American Partisanship and the Legacy of the Civil Rights Movement: Generational, Regional, and Economic Influences on Democratic Identification, 1973–1994 .” Political Psychology 26 : 735 – 54 . Google Scholar CrossRef Search ADS Mannheim , Karl . 1952 . “ The Sociological Problem of Generations .” In Essays on the Sociology of Knowledge , edited by Paul Kecskemeti , pp. 276 – 322 . New York : Routledge and Kegan Paul . Manza , Jeff , and Clem Brooks . 1998 . “ The Gender Gap in US Presidential Elections: When? Why? Implications ?” American Journal of Sociology 103 : 1235 – 66 . Google Scholar CrossRef Search ADS Mason , William M. , and Stephen Fienberg . 1985 . Cohort Analysis in Social Research: Beyond the Identification Problem . New York : Springer-Verlag . Google Scholar CrossRef Search ADS Mattei , Laura R. Winsky , and Franco Mattei . 1998 . “ If Men Stayed Home... The Gender Gap in Recent Congressional Elections .” Political Research Quarterly 51 : 411 – 36 . Google Scholar CrossRef Search ADS Miller , Warren E . 1992 . “ Generational Changes and Party Identification .” Political Behavior 14 : 333 – 52 . Google Scholar CrossRef Search ADS Nadeau , Richard , and Harold W. Stanley . 1997 . “ Age and Partisanship, 1952–1992 .” In Southern Parties and Elections: Studies in Regional Change , edited by Robert P. Steed , Laurence W. Morleand , and Tod A. Baker , pp. 78 – 94 . Tuscaloosa : University of Alabama Press . Neundorf , Anja , and Richard G. Niemi . 2014 . “ Beyond Political Socialization: New Approaches to Age, Period, Cohort Analysis .” Electoral Studies 33 : 1 – 101 . Google Scholar CrossRef Search ADS Norpoth , Helmut . 1987 . “ Under Way and Here to Stay: Party Realignment in the 1980s ?” Public Opinion Quarterly 51 : 376 – 91 . Google Scholar CrossRef Search ADS Norrander , Barbara . 1997 . “ The Independence Gap and the Gender Gap .” Public Opinion Quarterly 61 : 464 – 76 . Google Scholar CrossRef Search ADS Norrander , Barbara . 1999 . “ The Evolution of the Gender Gap .” Public Opinion Quarterly 63 : 566 – 76 . Google Scholar CrossRef Search ADS Norrander , Barbara . 2008 . “ The History of the Gender Gaps .” In Voting the Gender Gap , edited by Lois Duke Whitaker , pp. 9 – 32 . Urbana : University of Illinois Press . Schreckhise , William D. , and Todd G. Shields . 2003 . “ Ideological Realignment in the Contemporary US Electorate Revisited .” Social Science Quarterly 84 : 596 – 612 . Google Scholar CrossRef Search ADS Shorrocks , Rosalind . 2016 . “ Modernisation and Government Socialisation: Considering Explanations for Gender Differences in Cohort Trends in British Voting Behaviour .” Electoral Studies 42 : 237 – 48 . Google Scholar CrossRef Search ADS Shorrocks , Rosalind . 2018 . “ Cohort Change in Political Gender Gaps in Europe and Canada: The Role of Modernization .” Politics & Society . https://doi.org/10.1177/0032329217751688 Smets , Kaat , and Anja Neundorf . 2014 . “ The Hierarchies of Age-Period-Cohort Research: Political Context and the Development of Generational Turnout Patterns .” Electoral Studies 33 : 41 – 51 . Google Scholar CrossRef Search ADS Tilley , James . 2002 . “ Political generations and partisanship in the UK, 1964–1997 .” Journal of the Royal Statistical Society: Series A 165 : 121 – 35 . Google Scholar CrossRef Search ADS Tilley , James , and Geoffrey Evans . 2014 . “ Ageing and Generational Effects on Vote Choice: Combining Cross Sectional and Panel Data to Estimate APC Effects .” Electoral Studies 33 : 19 – 27 . Google Scholar CrossRef Search ADS Welch , Susan , and John Hibbing . 1992 . “ Financial Conditions, Gender, and Voting in American National Elections .” Journal of Politics 54 : 197 – 213 . Google Scholar CrossRef Search ADS Yang , Yang , and Kenneth C. Land . 2008 . “ Age-Period-Cohort Analysis of Repeated Cross-Section Surveys: Fixed or Random Effects ?” Sociological Methods and Research 36 : 297 – 326 . Google Scholar CrossRef Search ADS Yang , Yang , and Kenneth C. Land . 2013 . Age-Period-Cohort Analysis: New Models, Methods, and Empirical Applications . Boca Raton, FL : CRC Press, Taylor and Francis Group . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the American Association for Public Opinion Research. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Public Opinion Quarterly Oxford University Press

The Partisan Gender Gap in the United States: A Generational Replacement?

Public Opinion Quarterly , Volume Advance Article (2) – Jun 1, 2018

Loading next page...
 
/lp/ou_press/the-partisan-gender-gap-in-the-united-states-a-generational-cM7sKXBcXl
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the American Association for Public Opinion Research. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0033-362X
eISSN
1537-5331
D.O.I.
10.1093/poq/nfy013
Publisher site
See Article on Publisher Site

Abstract

Abstract To what degree have generational differences contributed to partisan changes in the American electorate, and what role did they play in the emergence of the gender gap in party identification? This paper sheds light on parallel and contradicting partisan trends among subgroups of the American electorate that affected the partisan gender gap over the past decades. By unpacking the gap by region, race, and generations, the analysis reveals that the effect of generational replacement on the gender gap varied in terms of size and direction between different subgroups. While in the South newer generations of white women diverged from the New Deal generation, consequently having a greater effect on the gender gap, in the rest of the country shifts among white men affected the gap to a greater extent than shifts among white women. Among African Americans, a decline in Democratic support was shown among newer generations of men, but less so among women. The findings highlight the importance of such political and historical contexts, and raise questions about the future of the partisan gender gap as the New Deal generation is replaced. Gender differences in voting and partisanship—a gender gap—have characterized US elections since at least 1952. Although women today identify as Democrats more so than men (the “modern gender gap”; see Inglehart and Norris [2000]), the pattern was precisely the opposite in the 1950s (the “traditional gender gap”). Scholars have shown that the modern gender gap resulted from structural, economic, and cultural changes that affected the lives of women, shifting their political preferences toward more liberal and prosocial policies generally identified with the Democratic Party (e.g., Manza and Brooks 1998; Edlund and Pande 2002). But these female-centered explanations for the gap tell only part of the story. A closer look shows that it was men, especially white men, who changed their preferences and moved to the Republican Party in higher numbers (Kaufmann and Petrocik 1999; Norrander 1999). The emergence of the gender gap, it has been shown, belongs to a set of broader partisan changes and political realignments that took place in the United States in the twentieth century. One powerful mechanism for political changes of this kind is generational replacement, in which newer cohorts with distinctive political preferences enter the electorate and replace older cohorts. Indeed, most explanations for the current modern gender gap suggest that women of younger generations should have been more affected by those external conditions. Although young generations played a significant role in shaping political trends in the United States (e.g., Carlson 2008), the generational perspective of the gender gap in the United States has largely been overlooked, leaving lacunae in our understanding of how the gap originated and is manifested. This paper seeks to address these lacunae by exploring how generational differences contributed to partisan-related trends in the United States. The paper unpacks the gap by region and over time. The analysis shows how generational trends varied between gender, region, and race. The contribution of this study is therefore twofold. First, it connects the gender gap literature to not only the literature on party realignment and political socialization in the United States, but also the impact of generational difference on political outcomes. Second, the study moves from analyzing the gender gap to analyzing the intergenerational dynamics of partisanship within each gender. Utilizing the ANES dataset between 1952 and 2016, the study emphasizes the role played by the New Deal generation and, more importantly, by the newer generations in influencing the partisan gender gap in the United States. The Partisan Gender Gap in the United States The gender gap in voting and in party identification has been a stable phenomenon in American politics for a few decades now, with some fluctuations over the years (Mattei and Mattei 1998; Box-Steffensmeier, De Boef, and Lin 2004). The gap, which started out as a traditional gender gap reflecting those in other Western countries, changed during the 1960s with a shift among women toward the Democratic Party, resulting in a small modern gap. Similar to patterns seen in Europe (Inglehart and Norris 2000; Giger 2009), men’s support for the Democratic Party declined during the 1970s and 1980s, widening an already growing gender gap (Kaufmann and Petrocik 1999; Norrander 2008). Most studies on the United States and elsewhere focus on the period of the modern gap. Explanations are mostly women centered (Gidengil 2007), attributing the emergence of the modern gender gap to changing structural, economic, and cultural factors that affected women’s policy and party preferences. For example, the decline in marriage and higher divorce rates (Edlund and Pande 2002), changes in women’s participation in the labor market (Manza and Brooks 1998), and inequality in gender division of household work (Iversen and Rosenbluth 2006) all served to make women more independent but also more economically vulnerable (Abendschön and Steinmetz 2014). Other studies have emphasized cultural change such as the decline in religiosity (Emmenegger and Manow 2014) as well as the rise of feminism (Giger 2009). The literature on the US gender gap focuses on how it is shaped by issue preferences and cultural factors, for example, social welfare issues (Erie and Rein 1988) and their salience for each gender (Kaufmann and Petrocik 1999), evaluation of the national economy (Welch and Hibbing 1992; Chaney, Alvarez, and Nagler 1998), and values such as reproductive rights and female equality (Conover 1988; Kaufmann 2002). Studies also show that the modern gender gap emerged as a result of men deserting the Democratic Party, while women’s partisanship remained more stable over time (Kaufmann and Petrocik 1999; Norrander 2008). The emergence of the partisan gender gap therefore can be examined in light of overall partisan trends and political realignments, especially in southern states, where much of the gap was attributed to differences in partisan realignment between genders (Norrander 1999). What previous studies on the voting gender gap in the United States tended to overlook was the generational component of political trends, which over time can change overall preferences in society. The effect of generational replacement on the electoral gender gap was suggested by Inglehart and Norris (2000, 2003), who argued that younger women were more affected by modernization-related processes (see also Shorrocks [2018]). However, in the United States, a modern gender gap is pronounced within both young and old birth cohorts who were born many years before modernization processes took place.1 Yet, as this study claims, generational differences have played a role in fostering partisan changes within each gender, consequently affecting gender gap trends. Generations in the United States as a source of political change Any aggregate change in attitudes can be the product of two mechanisms: period effects and generational replacement. In the former, people of all cohorts change their preferences over time. In this case the Zeitgeist, the spirit of the times, affects all individuals alike, regardless of age. Where generational replacement occurs, older generations continue to hold on to old attitudes, while newer generations with distinctive preferences enter the electorate and cause an overall change. Here, the circumstances of each period have a powerful influence on young people, and these effects persist over the lifecycle of a generation’s members (Mannheim 1952; Neundorf and Niemi 2014). In political research specifically, scholars have acknowledged the impact of major political actors during the period when voters come of age. This impact on political preferences lasts throughout the voters’ lives (Tilley 2002). As voting for the same party over time strengthens attachment to that party (Dinas 2014), we would expect to identify generational effects on political preferences according to the prevailing major political actors during the critical years of each birth cohort (Ghitza and Gelman 2014). In times of change, political events are more likely to affect voters who have not yet had the chance to participate in political acts such as voting or are still not committed to a particular party (Miller 1992) than older people whose political preferences have been firmly established (Green, Palmquist, and Schickler 2002). This mechanism suggests that new generations could be agents of political change. While there are numerous ways to define generations depending on the context and goals of the research, this paper specifically follows the literature on partisanship in the United States, which distinguishes generations according to the period in which they politically came of age. Here, in brief, I discuss the sociopolitical circumstances underpinning the paper’s classification of generations. The literature on partisanship further develops these distinct circumstances. A vast body of literature emphasizes the 1930s and 1960s as periods of substantial turnover in mass partisanship. The partisan realignment of the 1930s, which occurred in light of Franklin D. Roosevelt’s presidency and the New Deal, started a period of Democratic dominance that lasted until the 1960s. The South, of course, outnumbered other regions in its Democratic support, and was in practice a one-party system (Black and Black 1987). As debates on segregation and other issues began to polarize the American electorate, and as the Democratic Party moved toward supporting antisegregation policies, white southerners started deserting the party. The partisan transition of the white South was a gradual process, in which Johnson’s Civil Rights Act (1964) and Voting Rights Act (1965) constituted a turning point. Outside the South, Democratic identification declined as well, especially during the 1980s, which raised questions about a new realignment toward a Republican-dominated era (Norpoth 1987). The next two decades showed that this was not the case, as the Democratic Party regained power during the 1990s and 2000s (Knuckey 2015). Based on the critical periods of partisan realignment in the United States, scholars treated people who started voting before 1932 as the pre–New Deal generation, and those who started voting between 1932 and 1963 as the New Deal generation. A third generation, usually comprising people who started voting in 1964 and afterward, is referred to as the post–New Deal generation (e.g., Miller 1992; Nadeau and Harold 1997). I split the members of the post–New Deal generation into two, calling those who started voting in 1980 and afterward the post-Sixties generation.2 GENDERED GENERATIONAL EFFECTS: RESEARCH STRATEGY AND GOALS This paper suggests that generational differences may differ according to gender. As Mannheim (1952) claimed, historical circumstances may affect subunits of the same birth cohort differently. Thus, women and men may have reacted differently to political circumstances such that generational differences among them were not identical (Shorrocks 2016). Another obvious distinction that needs to be made in the US context is between white and African American voters, as they derive from different political contexts and experiences and therefore may exhibit different generational effects (Luks and Elms 2005). The same can be said about regions—the political context of the South in particular requires that this region be analyzed separately. Previous studies deal mostly with the electoral gender gap as a dependent variable. Focusing on women, these studies treat men as the benchmark to which women’s behavior was compared. But if we acknowledge that women and men are different social groups in terms of political behavior, it would be just as reasonable to compare newer generations of women to their female predecessors as to males of their own generation. Women today are different from men of their age (hence the gap between them), but they are also different—in a different way—from women of older generations in terms of education, divorce rates, values, and so on. Of course, the same can be said about men, although only a few studies on the gender gap elaborate on the changes in the male side of the equation (e.g., Kaufmann 2002; Iversen and Rosenbluth 2006). Therefore, the analysis of this paper estimates generational trends among each gender separately in order to learn how each generation’s preferences changed over the years and compared to other generations within the same gender. Methods I utilize the ANES cumulative dataset for the period 1952–2012, to which I added the 2016 preelection survey. The dependent variable, party identification, was originally measured on a seven-point scale, in which respondents were asked their partisanship identification, ranging between Democrat, Independent, and Republican. The answers on this scale included: strong Democrat; weak Democrat; Independent–Democrat; Independent; Independent–Republican; weak Republican; and strong Republican. This variable was recoded into a dummy variable (1 = strong, weak, or leaning Democrat; 0 = strong, weak, or leaning Republican). Pure Independents were excluded from the sample. Some studies excluded leaners and focused on weak and strong partisans only. However, Norrander (1997) showed that there is an independence gap between men and women, according to which men display a greater probability of identifying as leaners while women identify as weak partisans. It is thus important to include both categories in the study. The sample is divided into the southern and nonsouthern United States. The South includes 11 exfederation states: Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Texas, Tennessee, and Virginia. The sample is also divided by race, with the analysis first focus on white respondents, and a complementary analysis of African Americans. METHOD OF ANALYSIS Given the absence of panel data for an extended period of time, this study, like most age-period-cohort (APC) studies, makes use of cross-sectional setups. Such setups make it possible to track people who belong to the same birth cohort over time and to treat them as if they represent a sample of this birth cohort at each point in time (“synthetic cohorts”) (Alwin and Krosnick 1991). The current study employs a descriptive approach side by side with APC regression models. The descriptive method, which presents the total outcome for each generation at each point in time, is the most straightforward way of telling the story of each generation over its life cycle. It helps in identifying the periods in which generations behaved similarly or dissimilarly, as well as critical periods of major change within generations (Firebaugh 1997; Glenn 2005). Yet, this method does not control for the two main confounders—period and age3—or for other compositional differences that result in observed variation between cohorts (Yang and Land 2013). Therefore, I use APC regression models that provide cohort effects that are estimated over all periods and ages4 for three purposes: first, to look for empirical evidence for distinct tendencies that characterize each cohort at all ages and in all periods (Yang and Land 2008); second, to examine whether compositional differences in demographic characteristics between cohorts account for the differences found in partisan preferences (Smets and Neundorf 2014); and third, to evaluate generational effects in a more meaningful way using a simulation of generational replacement. Results FROM THE GENDER GAP TO GENDERED PARTISAN TRENDS What did the partisan gender gap look like among white Americans in each region? Figure 1 presents the percent of Democratic identifiers out of all partisans among white men and women between 1952 and 2016. The gender gaps portrayed are quite similar. In both regions, there was a period of traditional gender gap, of more women than men being Republican, which was followed by a transformation of the gap into its modern form. Figure 1. View largeDownload slide Partisan trends among white men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 1. View largeDownload slide Partisan trends among white men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. However, when moving from the gap to describing men’s and women’s partisan trends in each region, two different stories emerge. Prior to 1960, more white men than women in the South were Democratic, but the rapid decline in Democratic identification, which was less steep among women, caused the gender gap to reverse. Thus, after 1960, women identified with the Democratic Party in higher numbers than men. Outside the South, the partisan gap among white voters emerged through the two following transitions. The first revolved around women, who exhibited an increase in their Democratic support from the 1950s to the 1960s. Before 1960 the average Democratic support among white nonsouthern women was 51.8 percent, while in the years 1960–1968 the average was 55.3 percent. Yet a greater impact on the emergence of the modern gender gap was that of men’s defection from the Democratic Party. Their Democratic support, which was 55.4 percent on average during the 1970s, declined to the average level of 46.5 percent during the 1980s. This description of partisan trends, however, treats each group as composed of the same people, who might or might not have changed their preferences between periods. The next section examines an alternative portrayal of the electorate as composed of different generations that enter and leave the electorate over the years. Figure 2 unpacks partisan trends of southern white men and women according to the four generations described above: Pre–New Deal generation, New Deal generation, post–New Deal generation, and post-Sixties generation. The figure shows that at the beginning and among both genders, the New Deal generation exhibited trends similar to those of its predecessor, the pre–New Deal generation. During the 1950s and 1960s, these two older generations experienced a decline in Democratic support, while the next generation exhibited even lower levels of Democratic support. The members of this post–New Deal generation started voting when southern white voters massively left the Democratic Party. The young voters of the 1960s and afterward might have acquired their partisanship in Democratic families, but by the time they reached adulthood their parents’ generation had already started abandoning this political home. The graphs also show that the members of the newest generation, who came of age during the Republican years of the 1980s, were more Republican than all other cohorts in the beginning, although they did not become more Republican with time. Figure 2. View largeDownload slide Partisanship among white southern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long dash light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 2. View largeDownload slide Partisanship among white southern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long dash light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. While the graphs show similar generational stories for both genders, men in all cohorts were less likely to identify as Democrats compared to women. One explanation suggested by Schreckhise and Shields (2003) was that men were more affected by racial issues compared to women, who were more influenced by socioeconomic issues that made them more pro-redistribution and therefore more Democratic. Regarding white women outside the South (figure 3), the analysis clearly shows a Republican tendency of the oldest pre–New Deal generation. The following New Deal generation was much more Democratic than its predecessor at that time, and remained Democratic over most of its life cycle. Interestingly, though, this generation showed a decline in Democratic support in the years 2012 and 2016. Figure 3. View largeDownload slide Partisanship among white nonsouthern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long-dashed light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 3. View largeDownload slide Partisanship among white nonsouthern men and women in four generations. Source.—ANES data 1952–2016. Generational trends are presented for the pre–New Deal generation (long-dashed dark line), New Deal generation (solid line), post–New Deal generation (short-dashed line), and post-Sixties generation (long-dashed light line). Each trend line is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. The post–New Deal generation, which came of age in the second half of the 1960s and 1970s, was the main suspect for creating the modern gender gap. This new generation of women matured into adulthood during the second wave of feminism, when both women’s level of education and their share in the labor force rose. Indeed, at first, this generation was more likely to identify as Democrats compared to women of previous generations. However, the percentage of women identifying as Democrats among the post–New Deal generation continually declined and reached average levels among women during the 1990s and 2000s. The newest generation, which started voting in a rather right-wing period, was more Republican at the beginning, but then exhibited an increase in Democratic support over the 2000s. Regarding the overall decline in Democratic support among men, figure 3 provides some evidence for generational differences. The defection from the Democratic Party is pronounced by members of the New Deal generation, but this change was not great compared to the following generations. The post–New Deal generation, which started as more Democratic compared to other generations at that time, experienced a decline in Democratic support through the 1970s and 1980s. The next generation, who started voting in Reagan’s era, was even more Republican than previous generations. Interestingly, in the 2000s this generation showed a considerable increase in Democratic identification. BEYOND TIME: ESTIMATING NET GENERATIONAL EFFECTS In order to more precisely measure generational differences while controlling for period and age, I ran four logistic regression models separately for each of the subgroups described in the previous section: white southern men, white southern women, white nonsouthern men, and white nonsouthern women. The dependent variable in all models is Democratic identification.5 Independent variables include age and its second and third polynomials, three dummy variables for generations with the New Deal generation as the reference category, and survey-year dummies. The regression equation for each model is as follows: Logit(Democrat)=β0+β1Age+β2Age2+β3Age3+β4PreNewDeal+                                      β5PostNewDeal+β6PostSixties+∑j=127γjYearj As shown in table 1, on average, the New Deal generation was more Democratic than both its predecessors and followers, as indicated by the negative coefficients for most generational effects. However, only some of these generational effects are statistically significant. Table 1. Generational effects on democratic partisanship White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 Note.—Logistic regression models in which the dependent variable is Democratic identification. Reference categories are as follows. Generations: New Deal generation. Survey year: 1952. Full results including year coefficients are presented in table A1 in the Online Appendix. Standard errors in parentheses. Source.—ANES data 1952–2016. * p < 0.05. View Large Table 1. Generational effects on democratic partisanship White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 White southernmen White southernwomen White nonsouthernmen White nonsouthernwomen African Americanmen African Americanwomen  Age –0.011 0.001 –0.093* –0.082* 0.066 0.280* (0.041) (0.036) (0.023) (0.021) (0.092) (0.084)  Age2 0.000 0.000 0.002* 0.002* –0.001 –0.006* (0.001) (0.001) (0.000) (0.000) (0.002) (0.002)  Age3 –0.000 –0.000 –0.000* –0.000* 0.000 0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Generation  Pre–New Deal –0.173 –0.118 –0.334* –0.338* -0.907* –0.192 (0.160) (0.135) (0.082) (0.074) (0.345) (0.307)  New Deal – – – – – –  Post–New Deal –0.327* –0.388* –0.054 0.049 –0.473 –0.439 (0.122) (0.114) (0.068) (0.062) (0.305) (0.271)  Post-Sixties –0.245 –0.370* –0.258* –0.153 –0.820 –0.660 (0.197) (0.178) (0.110) (0.101) (0.435) (0.416)  Survey years + + + + + +  Constant 2.401* 1.648 1.956* 1.675* 0.199 –2.342 (0.698) (0.585) (0.360) (0.328) (1.350) (1.266) N 4,498 5,391 13,353 15,995 2,305 3,606 Note.—Logistic regression models in which the dependent variable is Democratic identification. Reference categories are as follows. Generations: New Deal generation. Survey year: 1952. Full results including year coefficients are presented in table A1 in the Online Appendix. Standard errors in parentheses. Source.—ANES data 1952–2016. * p < 0.05. View Large In the South (models 1 and 2), generational replacement has been more pronounced among women compared to men. While for white women the coefficients for the two youngest generations are statistically significant, among men, only the post–New Deal is significantly less Democratic than the New Deal generation (p < .05). This difference between women and men probably has to do with the greater loyalty of the New Deal generation to the Democratic Party on the female side. On average, southern white women of the New Deal generation did not defect from the Democratic Party as much as men from the same generation, at least up to the beginning of the 2000s (see their moderate slope in this period in figure 2). Next, the idea of a generational replacement accounting for men’s decline in Democratic support outside the South is only partially supported (models 3 and 4). That is, the post-Sixties generation among white men has significantly lower Democratic support compared to the New Deal generation, but the post–New Deal does not. Nevertheless, the picture of the average effect hides the trend of decreasing Democratic support among the post–New Deal generation during the 1980s, which can be detected in figure 3. Among women, the New Deal generation emerges as significantly more Democratic than its predecessor. The younger generations, however, show mixed results, depending on gender. Among white women, the post–New Deal generation shows higher levels of Democratic support compared to the New Deal generation (their coefficient is positive), but this result is not significant. The post-Sixties generation was on average less Democratic than the New Deal generation, but again this result is not significant. This finding confirms the idea of generational convergence among nonsouthern white women. Do these generational differences stem from intergenerational differences in demographic composition, or do they hold when controlling for demographic factors? To answer this question, I ran the same four models of table 1 with additional controls for college degree, income, work status, and marital status, four demographic variables that are available in the ANES throughout the period of this research. The results, which can be found in table A3 in the Online Appendix, show that the pattern of generational effects that emerged from table 1 persists. AFRICAN AMERICANS, GENERATIONS, AND THE PARTISAN GENDER GAP While some studies do not find gender gap in party identification among African Americans (e.g., Conway 2008), Junn (2017) emphasizes the role of women of color in fostering and stimulating the gender gap in voting in the United States. This does not mean that men of color are not Democratic overall, but that increasing turnout levels and demographic changes made women of color an important subgroup of voters affecting voting gap between women and men in the United States. Although sample sizes of African Americans in the ANES do not allow us to separate this subgroup into regions, the analysis of overall partisan trends among black women and men, and in particular its generational aspect, carry important insights about the gender gap in the United States. The partisan gender gap among African Americans since 1980 was 3.7 percent on average according to the ANES data, but it is not statistically significant, probably resulting from the small number of respondents in each election year. Figure 4 shows that an increase in Democratic support occurred among both genders during the 1960s and 1970s. Similarly to nonsouthern whites, a decline in Democratic support among black men widened the gender gap. Figure 4. View largeDownload slide Partisan trends among African American men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. Figure 4. View largeDownload slide Partisan trends among African American men and women 1952–2016. Source.—ANES data 1952–2016. Solid lines represent women. Dashed lines represent men. Each trend line (accompanied by 95 percent confidence intervals) is based on the percentage of identifiers as Democrats in each survey year, using a kernel-weighted local polynomial regression for better presentation. To estimate generational differences among each gender, I ran an APC analysis (models 5 and 6 in table 1), which shows that all coefficients are negative, thus all generations are less Democratic than the New Deal generation, but only the pre–New Deal generation among men is significantly significant (p < 0.05). As the generational categorization suggested above is not necessarily relevant for African Americans, I conducted a robustness check of generational trends using Generalized Additive Models (see table B1 and figure B3 in the Online Appendix). The results show that newer cohorts of black women were a bit less likely to be Democratic, but the confidence intervals around these effects are quite large. The picture of generational effects among black men is clearer, with members of older cohorts more likely to be Democrats, and members of younger cohorts less likely. BACK TO THE GENDER GAP: A GENERATIONAL REPLACEMENT? The last empirical section estimates to what degree generational replacement alone affected the total gender gap among white Americans over the past decades.6 This analysis examines the impact of generational replacement on party identification, using a simulation in which newer generations hold the same political preferences as older generations.7 In particular, the simulation here investigates what partisan trends would look like if the post–New Deal and post-Sixties generations had party identification levels identical to the New Deal generation. Figure 5 presents the results of the simulation for southern whites on the left and for nonsouthern whites on the right. The bars represent the difference between the predicted percent of Democratic identifiers based on the original model and the predicted percent of Democratic identifiers based on the simulation, for each decade. The dark gray bars stand for women, and light gray bars for men. Whiskers mark 95 percent confidence intervals.8 Figure 5. View largeDownload slide The difference in partisan trends in the absence of generational replacement. The bars represent the difference between the predicted percent of Democratic identifiers based on the original APC model and the predicted percent of Democratic identifiers based on the simulation (i.e., no generational replacement), for each decade (table 1, models 1–2 for the left panel and models 3–4 for the right panel). Positive gaps indicate by how much the level of Democratic support would have been higher in the case of no generational replacement. Dark gray bars stand for women, and light gray bars for men. Whiskers mark 95 percent confidence intervals. Figure 5. View largeDownload slide The difference in partisan trends in the absence of generational replacement. The bars represent the difference between the predicted percent of Democratic identifiers based on the original APC model and the predicted percent of Democratic identifiers based on the simulation (i.e., no generational replacement), for each decade (table 1, models 1–2 for the left panel and models 3–4 for the right panel). Positive gaps indicate by how much the level of Democratic support would have been higher in the case of no generational replacement. Dark gray bars stand for women, and light gray bars for men. Whiskers mark 95 percent confidence intervals. As can be seen, almost all bars show positive differences, which means that if the two newer generations had the same preferences as the New Deal generation, white women and men in both regions would have supported the Democratic Party in higher numbers. The graph clearly shows the differences between regions. In the South, generational differences were larger among white women, while outside the South white men exhibited larger intergenerational differences compared to women. In particular, white women in the South would have been more Democratic by six percentage points in the 1990s had no generational replacement taken place. This difference was even larger in the following years and reached the level of an additional eight percentage points in the 2010s. Generational replacement affected white southern men as well, with six percentage points higher Democratic support in the 2000s had no generational replacement taken place. However, not all these differences are statistically significant. Outside the South, the picture is the opposite. White men’s partisanship was more affected by generational replacement compared to women’s. If newer generations had political preferences identical to the New Deal generation, white men’s support for the Democratic Party would have been higher by three percentage points in the 2000s and by four percentage points in the 2010s. White nonsouthern women would also have been a bit more Democratic if no replacement had taken place, but these differences are smaller and not statistically significant. The last part of the analysis asks what the partisan gender gap among whites would look like if no generational replacement had taken place. To answer this question, I went back to the raw data and calculated the partisan gender gap found in the ANES for each decade since 1980—with and without generational replacement.9 Findings show that if generational replacement had not taken place, outside the South the gap between white women and men would have been consistently smaller. The entrance of newer generations of men who did not identify with the Democratic Party as their predecessors had done enlarged the gap between white women (who remained loyal to the Democratic Party) and men. In the 1990s, for example, the gender gap could have been smaller by 5.5 percentage points if generational replacement had not taken place. The impact of generational differences diminished in the 2000s due to the increase in Democratic support among white men of the youngest generation (see their trend line in figure 3). Interestingly, a sharp decline in Democratic support among white women of the New Deal generation in the 2012 and 2016 elections, which was shown in figure 3, caused a reverse gender gap among this generation in that period. This finding demonstrates how cohorts of voters can change their political preferences at older ages, even if only temporarily. The results concerning southern whites are mixed. Up to 2000, the gap among the New Deal generation was smaller than the total gap. In the 2000s, the gender gap among the New Deal generation increased as its men departed from the Democratic tendencies prevalent among their female counterparts. At the same time, the gender gap among younger generations declined due to younger women becoming less Democratic and more similar to their male counterparts. The result of these trends was a smaller gender gap in the 2000s compared to what could have happened to the gap had newer generations held the preferences of the New Deal generation.10 Conclusions This paper analyzes several aspects of the origins of the partisan gender gap in the United States, asking to what degree generational replacement contributed to the emergence of the gap. The analysis, which unpacked the gap by gender, region, race, and generation, reveals differences between these subgroups that affected the gender gap within each region/race. In particular, the analysis highlights the role of the New Deal generation, which came of age during a period of Democratic dominance, and the extent to which generations that came afterward diverged (or not) from this generation’s Democrat tendency. The analysis shows that among white women, younger generations in the South were significantly less Democratic than women of the older New Deal generation. Outside the South, white women of the New Deal generation were clearly more Democratic than their predecessors, but their followers did not show much different party preferences and were only a bit less Democratic. Nonsouthern white men were the ones who exhibited a decline in support for the Democrats, shared by newer generations as well, which narrowed the gender gap among whites in that region. Among African Americans, and similarly to nonsouthern whites, generational differences affected men to a greater extent. Younger generations of black men were less Democratic than older black voters, but among black women there was no substantial decline in support for the Democrats along cohorts. Three main points arise from these findings. First, they demonstrate how important it is to unpack the quantity called the gender gap into men’s and women’s separate preferences, especially when studying transformations therein over time. Although the gap is important in itself, it might conceal variations within each gender—and other subdivisions—that explain central trends in American politics. Second, although a lot of attention has been given to the role of women, and to changes that occurred on the female side of the equation, the results show that some trends are attributed to men, and those that are attributed to women are in the opposite direction from the one that might be expected, as women of newer generations showed diminished Democratic support compared to women of older generations. Third, some political transitions are attributable to the entrance of newer generations and some are not. Only by separating those trends over time and within subgroups can we identify when generational replacement mattered and when it did not. This kind of perspective is missing for the youngest birth cohorts, some of which show very high levels of Democratic identification. If these are meaningful trends representing generational phenomena or a short-term effect, only time will tell. Appendix This article utilizes the ANES Cumulative Time-Series data set from 1952 to 2012 and the ANES pre-election survey for 2016. The most common ANES study design is a cross-section, equal-probability sample, although in some surveys they oversampled certain subpopulations. Additional information can be found online at http://www.electionstudies.org/studypages/anes_timeseries_cdf/anes_timeseries_cdf.htm. I utilize the weights provided by the ANES [variable: VCF0009z] and the weighting variable V160101 for the 2016 survey. Most of the surveys were carried out face-to-face, except for 2012 and 2016 in which an internet based survey was added. The response rates are as follows (http://www.electionstudies.org/overview/dataqual.htm): 1952: 77.2; 1956:85.5; 1958: 78.1; 1960 and 1962: not ascertained; 1964: 80.6; 1966: 77.1; 1968: 77.4; 1970: 76.6; 1972: 75.0; 1974: 70.0; 1976: 70.4; 1978: 68.9; 1980: 71.8; 1982: 72.3; 1984:72.1; 1986: 67.7; 1988: 70.5; 1990: 70.6; 1992: 74.0; 1994: 72.1; 1996: 59.8; 1998:63.8; 2000: 60.5; 2002: 55.3; 2004: 66.1; 2008: 59.5; 2012: 38.0 for the face-to-face component and 2.0 for the Internet component; 2016: 50.0 for the face-to-face component and 44.0 for the Internet component. Measures 1952-2012 cumulative file Partisanship (VCF0301). Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or what? (IF REPUBLICAN OR DEMOCRAT) Would you call yourself a strong (REP/DEM) or a not very strong (REP/DEM)? (IF INDEPENDENT, OTHER [1966 AND LATER: OR NO PREFERENCE; 2008: OR DK) Do you think of yourself as closer to the Republican or Democratic party? (Strong Democrat, Weak Democrat, Independent–Democrat, Independent–Independent, Independent–Republican, Weak Republican, Strong Republican) Race (VCF0106). What race or ethnic group or groups best describes you? (white; black; other) Education (VCF0110). What is the highest grade of school or year of college you have completed? (grade school or less (0–8 grades); high school (12 grades or fewer, incl. non-college training); some college (13 grades or more but no degree); college or advanced degree) Household income (VCF0114). Information about income is very important to understand how people are doing financially these days. Your answers are confidential. Would you please give your best guess? [responses recoded by ANES in the Cumulative File] (0 to 16 percentile; 17 to 33 percentile; 34 to 67 percentile; 68 to 95 percentile; 96 to 100 percentile) Male (VCF0104). Do you identify as a man or a woman? (male, female) Age (VCF0101). What is your date of birth? (age 17 and below and 97 and up were excluded). South. Using variable (VCF0901), individuals living in states included in the original Confederacy were coded 1 (Virginia, Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Texas, Tennessee); otherwise, 0. Work status (VCF0118). 1968-1970: Are you presently employed, or are you unemployed, or retired, (a housewife), (a student), or what? 1972-1978: (1972: We’d like to know if you are looking for work, working now) (1974- 1978: We’d like to know if you are working now, or are you unemployed,) retired, (a housewife) a (student), or what? (IF HOMEMAKER OR STUDENT) Are you doing any work for pay at the present time? (IF R IS HOMEMAKER OR STUDENT AND R IS WORKING FOR PAY:) About how many hours do you work on your job in the average week? 1980 AND LATER: We’d like to know if you are working now, temporarily laid off, or are you unemployed, retired, permanently disabled, (a homemaker), (a student), or what? Marital Status (VCF0147). 1952: Are you married? 1956-2004: Are you married now and living with your husband/ wife (2002: spouse)-- or are you widowed, divorced, separated, or have you never married? 2008: Are you married now and living with your husband/wife-- or are you widowed, divorced, separated, or have you never married? / Are you married, divorced, separated, widowed, or have you never been married? 2012: Are you now married, widowed, divorced, separated or never married? Are you currently living with a partner, or not? 2016 pre-election survey List of variables: Age (V161267), gender (V161342), race (V161310x), region (V161010e), partisanship (V161158x), income (V161361x), work status (V161276x), education (V161270), marital status (V161268). Supplementary Data Supplementary data are freely available at Public Opinion Quarterly online. The research for this article was completed while the author was a doctoral candidate in the Department of Political Science at Hebrew University of Jerusalem. For helpful comments and suggestions, the author thanks Sylvia Bashevkin, André Blais, Elisabeth Gidengil, Guy Grossman, Orit Kedar, Einat Lavy, Odelia Oshri, Michael Shalev, and Rosalind Shorrocks, as well as the editors and three anonymous reviewers. Footnotes 1. A modern gender gap is found among white nonsouthern voters within most cohorts who were born after 1906, while among white southerners this gap is observed within cohorts born after 1926. Among African Americans, the modern gender gap is found only among cohorts born after 1946 (with the exception of the 1956–1965 cohorts). 2. The relevant birth years for each generation according to their chronological order are: 1880–1907, 1908–1939, 1940–1958, and 1959–1990. I use different generational categorizations for robustness checks, which generally support this categorization. These are Generalized Additive Models (table B1) and 10-year interval cohorts per Firebaugh and Chen’s (1995) categorization (table B2). For more details see the Online Appendix, Section B. 3. Since age (life cycle) effects generally create political continuity at the societal level, they cannot account for an aggregate political change. Methodologically, age effects are confounded with cohort at each point in time and therefore need to be controlled for. 4. Every APC analysis encounters an inherent problem of identification (Mason and Fienberg 1985) due to the perfect collinearity between the three components. Various methods for dealing with this problem have been proposed over the years. Nonetheless, as Tilley and Evens (2014) noted, all APC methods carry assumptions that may or may not fit the data in a given case. 5. For robustness checks, I ran the analyses with the original seven-point scale variable. The main results were replicated, except for African American men, whose generational effects become significant using OLS model. See table A2 in the Online Appendix. 6. Sample sizes of African Americans are too small for this procedure. 7. I conducted this analysis by calculating the predicted probability to identify as a Democrat of each respondent based on the coefficients from the relevant regression model (table 1) and the observed values for that respondent (Hanmer and Kalkan 2013). Then I calculated another set of predicted probabilities; this time, newer generations are awarded the coefficient of an older cohort instead of the coefficient of their original one. 8. The standard errors of the difference were calculated via bootstrap over 1,000 draws. 9. The first was calculated as the total gender gap in each decade, and the second as the gender gap among the New Deal generation only. 10. Similarly to nonsouthern whites, a tremendous decline in Democratic support took place in the 2012 and 2016 elections among white southern women of the New Deal generation. This trend led to an inverse partisan gender gap among that generation. References Abendschön , Simone , and Stephanie Steinmetz . 2014 . “ The Gender Gap in Voting Revisited: Women’s Party Preferences in a European Context .” Social Politics 21 : 315 – 44 . Google Scholar CrossRef Search ADS Alwin , Duane F. , and Jon A. Krosnick . 1991 . “ Aging, Cohorts, and the Stability of Sociopolitical Orientations over the Life Span .” American Journal of Sociology 97 : 169 – 95 . Google Scholar CrossRef Search ADS Black , Earl , and Merle Black . 1987 . Politics and Society in the South . Cambridge, MA : Harvard University Press . Box-Steffensmeier , Janet M. , Suzanna De Boef , and Tse-Min Lin . 2004 . “ The Dynamics of the Partisan Gender Gap .” American Political Science Review 98 : 515 – 28 . Google Scholar CrossRef Search ADS Carlson , Elwood . 2008 . The Lucky Few: Between the Greatest Generation and the Baby Boom . London : Springer . Chaney , Carole K. , R. Michael Alvarez , and Jonathan Nagler . 1998 . “ Explaining the Gender Gap in U.S. Presidential Elections, 1980–1992 .” Political Research Quarterly 51 : 311 – 39 . Conover , Pamela Johnston . 1988 . “ Feminists and the Gender Gap .” Journal of Politics 50 : 985 – 1010 . Google Scholar CrossRef Search ADS Conway , M. Margaret . 2008 . “ The Gender Gap: A Comparison across Racial and Ethnic Groups .” In Voting the Gender Gap , edited by Lois Duke Whitaker , pp. 170 – 83 . Champaign : University of Illinois Press . Dinas , Elias . 2014 . “ Does Choice Bring Loyalty? Electoral Participation and the Development of Party Identification .” American Journal of Political Science 58 : 449 – 65 . Google Scholar CrossRef Search ADS Edlund , Lena , and Rohini Pande . 2002 . “ Why Have Women Become Left-Wing? The Political Gender Gap and the Decline in Marriage .” Quarterly Journal of Economics 117 : 917 – 61 . Google Scholar CrossRef Search ADS Emmenegger , Patrick , and Philip Manow . 2014 . “ Religion and the Gender Vote Gap: Women’s Changed Political Preferences from the 1970s to 2010 .” Politics & Society 42 : 166 – 93 . Google Scholar CrossRef Search ADS Erie , Steven P. , and Martin Rein . 1988 . “ Women and the Welfare State .” In The Politics of the Gender Gap: The Social Construction of Political Influence , edited by Carol M. Mueller , pp. 173 – 91 . Newbury Park, CA : Sage Publications . Firebaugh , Glenn . 1997 . Analyzing Repeated Surveys . Thousand Oaks, CA : Sage Publications . Google Scholar CrossRef Search ADS Firebaugh , Glenn , and Kevin Chen . 1995 . “ Vote Turnout of Nineteenth Amendment Women: The Enduring Effect of Disenfranchisement .” American Journal of Sociology 100 : 972 – 96 . Google Scholar CrossRef Search ADS Ghitza , Yair , and Andrew Gelman . 2014 . “ The Great Society, Reagan’s Revolution, and Generations of Presidential Voting .” Working Paper. Gidengil , Elisabeth . 2007 . “ Beyond the Gender Gap: Presidential Address to the Canadian Political Science Association, Saskatoon, 2007 .” Canadian Journal of Political Science 40 : 815 – 31 . Google Scholar CrossRef Search ADS Giger , Nathalie . 2009 . “ Towards a Modern Gender Gap in Europe? A Comparative Analysis of Voting Behavior in 12 Countries .” Social Science Journal 46 : 474 – 92 . Google Scholar CrossRef Search ADS Glenn , N. D . 2005 . Cohort Analysis . Thousand Oaks, CA .: Sage Publications . Google Scholar CrossRef Search ADS Green , Donald , Bradley Palmquist , and Eric Schickler . 2002 . Partisan Hearts and Minds . New Haven, CT : Yale University Press . Hanmer , Michael J. , and Kerem Ozan Kalkan . 2013 . “ Behind the Curve: Clarifying the Best Approach to Calculating Predicted Probabilities and Marginal Effects from Limited Dependent Variable Models .” American Journal of Political Science 57 : 263 – 77 . Google Scholar CrossRef Search ADS Inglehart , Ronald , and Pippa Norris . 2000 . “ The Developmental Theory of the Gender Gap: Women’s and Men’s Voting Behavior in Global Perspective .” International Political Science Review 21 : 441 – 63 . Google Scholar CrossRef Search ADS Inglehart , Ronald , and Pippa Norris . 2003 . Rising Tide: Gender Equality and Cultural Change around the World . Cambridge : Cambridge University Press . Google Scholar CrossRef Search ADS Iversen , Torben , and Frances Rosenbluth . 2006 . “ The Political Economy of Gender: Explaining Cross National Variation in the Gender Division of Labor and the Gender Voting Gap .” American Journal of Political Science 50 : 1 – 19 . Google Scholar CrossRef Search ADS Junn , Jane . 2017 . “ The Trump Majority: White Womanhood and the Making of Female Voters in the U.S .” Politics, Groups, and Identities 5 : 343 – 52 . Google Scholar CrossRef Search ADS Kaufmann , Karen M . 2002 . “ Culture Wars, Secular Realignment, and the Gender Gap in Party Identification .” Political Behavior 24 : 283 – 307 . Google Scholar CrossRef Search ADS Kaufmann , Karen M. , and John R. Petrocik . 1999 . “ The Changing Politics of American Men: Understanding the Sources of the Gender Gap .” American Journal of Political Science 43 : 864 – 87 . Google Scholar CrossRef Search ADS Knuckey , Jonathan . 2015 . “ The Survival of the Democratic Party Outside the South: An Update and Reassessment .” Party Politics 21 : 539 – 52 . Google Scholar CrossRef Search ADS Luks , Samantha , and Laurel Elms . 2005 . “ African-American Partisanship and the Legacy of the Civil Rights Movement: Generational, Regional, and Economic Influences on Democratic Identification, 1973–1994 .” Political Psychology 26 : 735 – 54 . Google Scholar CrossRef Search ADS Mannheim , Karl . 1952 . “ The Sociological Problem of Generations .” In Essays on the Sociology of Knowledge , edited by Paul Kecskemeti , pp. 276 – 322 . New York : Routledge and Kegan Paul . Manza , Jeff , and Clem Brooks . 1998 . “ The Gender Gap in US Presidential Elections: When? Why? Implications ?” American Journal of Sociology 103 : 1235 – 66 . Google Scholar CrossRef Search ADS Mason , William M. , and Stephen Fienberg . 1985 . Cohort Analysis in Social Research: Beyond the Identification Problem . New York : Springer-Verlag . Google Scholar CrossRef Search ADS Mattei , Laura R. Winsky , and Franco Mattei . 1998 . “ If Men Stayed Home... The Gender Gap in Recent Congressional Elections .” Political Research Quarterly 51 : 411 – 36 . Google Scholar CrossRef Search ADS Miller , Warren E . 1992 . “ Generational Changes and Party Identification .” Political Behavior 14 : 333 – 52 . Google Scholar CrossRef Search ADS Nadeau , Richard , and Harold W. Stanley . 1997 . “ Age and Partisanship, 1952–1992 .” In Southern Parties and Elections: Studies in Regional Change , edited by Robert P. Steed , Laurence W. Morleand , and Tod A. Baker , pp. 78 – 94 . Tuscaloosa : University of Alabama Press . Neundorf , Anja , and Richard G. Niemi . 2014 . “ Beyond Political Socialization: New Approaches to Age, Period, Cohort Analysis .” Electoral Studies 33 : 1 – 101 . Google Scholar CrossRef Search ADS Norpoth , Helmut . 1987 . “ Under Way and Here to Stay: Party Realignment in the 1980s ?” Public Opinion Quarterly 51 : 376 – 91 . Google Scholar CrossRef Search ADS Norrander , Barbara . 1997 . “ The Independence Gap and the Gender Gap .” Public Opinion Quarterly 61 : 464 – 76 . Google Scholar CrossRef Search ADS Norrander , Barbara . 1999 . “ The Evolution of the Gender Gap .” Public Opinion Quarterly 63 : 566 – 76 . Google Scholar CrossRef Search ADS Norrander , Barbara . 2008 . “ The History of the Gender Gaps .” In Voting the Gender Gap , edited by Lois Duke Whitaker , pp. 9 – 32 . Urbana : University of Illinois Press . Schreckhise , William D. , and Todd G. Shields . 2003 . “ Ideological Realignment in the Contemporary US Electorate Revisited .” Social Science Quarterly 84 : 596 – 612 . Google Scholar CrossRef Search ADS Shorrocks , Rosalind . 2016 . “ Modernisation and Government Socialisation: Considering Explanations for Gender Differences in Cohort Trends in British Voting Behaviour .” Electoral Studies 42 : 237 – 48 . Google Scholar CrossRef Search ADS Shorrocks , Rosalind . 2018 . “ Cohort Change in Political Gender Gaps in Europe and Canada: The Role of Modernization .” Politics & Society . https://doi.org/10.1177/0032329217751688 Smets , Kaat , and Anja Neundorf . 2014 . “ The Hierarchies of Age-Period-Cohort Research: Political Context and the Development of Generational Turnout Patterns .” Electoral Studies 33 : 41 – 51 . Google Scholar CrossRef Search ADS Tilley , James . 2002 . “ Political generations and partisanship in the UK, 1964–1997 .” Journal of the Royal Statistical Society: Series A 165 : 121 – 35 . Google Scholar CrossRef Search ADS Tilley , James , and Geoffrey Evans . 2014 . “ Ageing and Generational Effects on Vote Choice: Combining Cross Sectional and Panel Data to Estimate APC Effects .” Electoral Studies 33 : 19 – 27 . Google Scholar CrossRef Search ADS Welch , Susan , and John Hibbing . 1992 . “ Financial Conditions, Gender, and Voting in American National Elections .” Journal of Politics 54 : 197 – 213 . Google Scholar CrossRef Search ADS Yang , Yang , and Kenneth C. Land . 2008 . “ Age-Period-Cohort Analysis of Repeated Cross-Section Surveys: Fixed or Random Effects ?” Sociological Methods and Research 36 : 297 – 326 . Google Scholar CrossRef Search ADS Yang , Yang , and Kenneth C. Land . 2013 . Age-Period-Cohort Analysis: New Models, Methods, and Empirical Applications . Boca Raton, FL : CRC Press, Taylor and Francis Group . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the American Association for Public Opinion Research. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Public Opinion QuarterlyOxford University Press

Published: Jun 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off