European Journal of Political Research 60: 225–238, 2021 225

doi: 10.1111/1475-6765.12384

Change and continuity in the ideological gender gap a longitudinal analysis

of left-right self-placement in OECD countries

RUTH DASSONNEVILLE
Département de science politique, Université de Montréal, Canada

Abstract. When asked to place themselves on a left-right scale, men and women tend to take different
positions. Over time, however ideological gender differences have taken a different form. While women
were traditionally more right-leaning than men, from around the mid-1990s onwards they have been found
to take positions to the left of men. Using an originally constructed dataset that includes information on
the left-right self-placement of more than 2.5 million respondents in 36 OECD countries between 1973 and
2018, I empirically verify how the ideological gender gap has evolved since. The results show, first, that while
women have shifted to the left since the late 1970s, the pace of this change has strongly diminished since the
late 1990s. Second, there is important between-country variation in the size of the reversal in the ideological
gender gap. Third, with the exception of the Silent generation and the Baby-boomers, newer generations of
women have not taken more left-leaning positions than generations before them.

Keywords: gender; generations; Ideological gender gap; left right

Introduction

At least since the publication of the work of Inglehart and Norris (2000), scholars have been
aware of a remarkable shift in men’s and women’s ideological leanings. Traditionally, survey
researchers found women to position themselves to the right of men, and women were also
found to be more inclined to vote for right-wing parties. This ‘ideological gender gap’ has
been reversed, however. Women are now placing themselves to the left of men and are more
likely to vote for left-wing parties compared to men (Abendschön & Steinmetz 2014; Giger
2009; Inglehart & Norris 2003).
In this research note, I make use of an exceptionally comprehensive and longitudinal
dataset with information on 2.5 million individuals in 36 OECD countries between 1973
and 2018 to verify whether the modern ideological gender gap is growing further, remains
stable, or is decreasing (again) in the more recent period.
To preview the results, I find that even though the shift of women to the left was originally
fairly strong, the pace of this change has diminished since the turn of the twenty-first century.
Second, even though the overall pattern is one of women shifting to the left, there remains
substantial between-country variation in the pace of this change and in the size of the gender
gap. Finally, while female Boomers are significantly more left-leaning than women of older
generations, the positions of younger generations of women have not continued to shift to
the left. These results offer little signs of a ‘real’ gender cleavage in ideology taking shape.

Replication materials for this paper can be accessed on Harvard Dataverse: [Link]


C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
226 RUTH DASSONNEVILLE

From a traditional to a modern ideological gender gap

Different socio-demographic variables shape citizens’ political preferences and serve as the
basis of strong cleavages in party systems (Lipset & Rokkan 1967). In contrast to what
holds for class or religion, however, gender differences have never developed into a stable
political cleavage (Inglehart & Norris 2003). Focusing on the limited ideological differences
between men and women in cross-sectional surveys, though, masks an important over-time
change in ideological position taking – among women in particular. In the United States,
scholars have drawn attention to a shift in women’s and men’s partisan leanings, with women
increasingly sorting into the Democratic party while men turn to the Republican party
(Box-Steffensmeier, De Boef & Tse-Min 2004; Harsgor 2018; Kaufmann & Petrocik 1999).
Inglehart and Norris (2000, 2003) furthermore pointed out that by the late 1990s a reversal
of men’s and women’s ideological positions was visible in established democracies more
generally.
From previous work, we know that the trend and the size of the ideological gender
gap is associated with structural variables that have fundamentally changed women’s
‘role and position in society and family life’ (Abendschön & Steinmetz 2014: 318),
such as the increased participation of women in the labor market and their growing
economic independence, improved educational opportunities for women, or the process of
secularization. As women participate more in the labor market and often experience pay
disparities, as they become higher educated, and as they become more secularized, their
ideological position appears to be more left wing (Abendschön & Steinmetz 2014; Giger
2009; Inglehart & Norris 2000; Iversen & Rosenbluth 2006).
Furthermore, the reversal of the ideological gender gap has been linked to attitudinal
changes. According to Inglehart and Norris (2000), the growth of post-materialist values in
advanced democracies is correlated to the reversal of the ideological gender gap. Men’s and
women’s positions differ on issues that are related to ‘freedom, self-expression and gender
equality’ (Inglehart & Norris 2000: 446), and as such issues have gained salience over time
– at the expense of economic issues – women have turned to the left (Giger 2009; Inglehart
& Norris 2000).
The – sometimes implicit – expectation of much of this work is that a modern gender
gap would gradually appear across democracies. While the process can proceed more or
less quickly, it is expected that after a period of ‘gender dealignment’ (i.e., women shifting
to no longer holding more right-wing or conservative views than men), follows a period
of ‘realignment’, implying women shift further to a more left-leaning ideological position
(Inglehart & Norris 2000). Even in those countries where women are still more right-leaning
than men, it is expected that with time, they will not only move to a similar ideological
position as men, but move further to the left (Giger 2009).
The mechanism explaining the over-time change, previous work has shown, is a
process of generational replacement. Those generational differences originate in the fact
that ideological change reflects changing values in society and a trend towards more
post-materialism. Such value change is often argued to be driven to a large extent by
generational replacement (Harsgor 2018). Inglehart and Norris (2003: 38), for example, state
that


C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE IDEOLOGICAL GENDER GAP 227

the theory of value change ( …) suggests that secular social trends have only a glacial
effect on cultural norms but that, through the socialization process, the conditions
experienced during the formative years of childhood and early adolescence make an
indelible impression on people. As a result, the values held in later life continue to be
shaped by these seminal experiences.

Structural changes of societal modernization as well likely have a generational

component, because younger birth cohorts ‘have particularly experienced modernization’
(Shorrocks 2018: 136). In fact, it is argued that below the surface of the over-time changes
that are being observed, there are more pronounced differences between generations. In
other words, it is assumed that generational differences are more important than period
effects when studying the ideological gender gap. This expectation, that originates in
the work of Inglehart and Norris (2003), receives empirical support by a number of
recent publications. Shorrocks (2018), who estimates age-period-cohort models to study the
ideological gender gap in Europe and Canada, finds strong evidence in favour of Inglehart
and Norris’ ‘gender-generation hypothesis’. Her findings lead her to conclude that ‘studies
that focus on the aggregate-level gender gap ( …) underestimate the magnitude of political
gender differences since the largest gender gaps are to be found within particular birth
cohorts’ (Shorrocks 2018: 159). Similarly, Harsgor (2018) finds that generational differences
are a big part of what explains over-time changes in the partisan gender gap in the United
States.
To the extent that new generations of women – who are socialized in a different context
than older generations and as a consequence thereof hold more left-wing attitudes –
continue to enter the electorate, the result will be a gradual and continued increase of
the ideological gender gap. Such changes might even contribute to the appearance of a
real gender cleavage in politics. However, with the exception of the longitudinal analyses
of Inglehart and Norris (2000, 2003) – that covers the period until the late 1990s – few
publications offer a view of the long-time trends in ideological differences between men and
women. And even less contributions do so comparatively. As a result, we do not know if the
modern ideological gender gap can be found in a growing group of countries, and whether
that gap continues to grow stronger over time, stabilizes, or perhaps shrinks in size in the
more recent period. This research note fills this void in our knowledge of the differences in
men’s and women’s ideological preferences.

Data and methods

To evaluate in more depth how women’s and men’s ideological self-placements have
changed over time, I make use of an originally constructed dataset with information
on the left-right self-placement of over 2.5 million respondents in 36 OECD countries.
The dataset is organized in a country-year format and covers the time period between
1973 and 2018. This formidable dataset was constructed by combining the data from
the Eurobarometer surveys, the Latinobarometro, the European Social Survey, the World
Values Study, the Comparative Study of Electoral Systems and national election studies
for particular countries. A complete list of countries and datasets included can be found
in Appendix A in the Supporting Information.

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
228 RUTH DASSONNEVILLE

The dataset includes information from all current OECD member states. This results in a
focus on 36 countries worldwide, which makes for a more diverse sample than what has been
used in most previous research that has studied the ideological gender gap. This also implies
that not all countries in the dataset have the same level of democratic development. All
country-years for which data are included, however, are from settings that were considered
at least ‘partly free’ by Freedom House.1
Each of these surveys that were used to construct the dataset includes a question asking
respondents to place themselves on a left-right scale.2 However, not all surveys measure left-
right self-placements in exactly the same way. The number of answer categories differs, and
some scales have a mid-point while others do not (Kroh 2007). To allow combining these
datasets, all scales were first harmonized to 1–10 scales (see also Ferland (2017) and Powell
(2000)).3 On this scale, 1 corresponds to the most left-wing position while 10 corresponds
to the most right-wing position. Given that the number of response options, and other
characteristics of the different surveys that are used might influence the results, I verify
whether the results are robust to controlling for the number of answer options and the
original surveys the data are from.
To evaluate the evolution of the ideological gender gap over time, and trends in men’s
and women’s ideological positions, I first make use of a nonparametrical technique and
estimate local polynomial smoother lines (Fah 2018). I also examine the over-time trend
in a parametric way by means of regression analyses – with a focus on the effects of time,
gender and the interaction between both. To account for the possibility that the change
might be non-linear, the time trend is accounted for by means of a series of decade dummies.
I also include country-fixed effects, implying the models serve to analyse within-country
variation over time. I complement this pooled analysis with analyses for each of the countries
separately, to get a sense of the extent to which patterns are general or context-specific.
To evaluate the extent to which gender realignment is driven by generational
replacement, in a second step, I estimate models that include parameters to differentiate
between the left-right placements of members of different generations. Members of the
same birth cohort grow up in the same socio-historical context. These shared experiences,
in particular what happens during their ‘formative years’, are thought to leave a lasting
imprint and affect their political attitudes and behaviour (Inglehart 1990; Mannheim 1928).
Given the importance of socio-political events for the development of generations, most
work that studies the role of generations in politics relies on country-specific categorizations
of different generations. Nevertheless, there is a large amount of overlap between the
generational categories that have been developed for different established democracies
(Grasso 2016). The Second World War, for example, is thought of as a demarcation
for distinguishing generations in most categorizations that are used to study political
attitudes and behaviour (Grasso 2016; Van den Broek 1996). Given the strong similarities
in generational categories that are developed for different countries, I rely on a single and
fairly general categorization that distinguishes between six broad generations: the Greatest
generation (1910–1924), the Silent generation (1925–1945), the Baby-boomers (1946–1964),
Generation X (1965–1980), Generation Y (1981–1996) and Generation Z (1997- present).
In addition, to account for the possibility that this general categorization does not capture
well the reality in several of the countries in the dataset, I verify whether results hold when
estimating age-period-cohort models that focus on differences between 5-year birth cohorts.

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE IDEOLOGICAL GENDER GAP 229

When investigating generational effects in this way, it is important to take into account
the role of age and period effects as well. Given that these three time-related variables are
linearly dependent, we cannot attribute differences between groups that are born in different
years to generational effects if the models do not account for the fact that members of
different birth cohorts also have a different age or have been interviewed in different time
periods (Glenn 2003; Neundorf & Niemi 2014). While it is important to account for the
three time effects, their perfect linear dependency implies that models cannot be estimated
without restricting at least some of the parameters. One way of restricting parameters is to
create categories for a certain time-effect rather than estimating its effect in a continuous
way (Neundorf & Niemi 2014). The time period-effect is already restricted by the reliance
on decade dummies, and the same holds for the generational effects – for which I distinguish
six generational groups. To account for life-cycle effects I add respondents’ age at the time
of the survey, and I also include age squared to account for the possibility that the relation
between age and ideology is nonlinear.

Results

Change over time

How has the ideological gender gap evolved since the late 1990s? As a first step, I focus
on a select number of countries for which information on the ideological self-placement
of respondents is available from 1973 onwards. More specifically, I use data from Belgium,
Denmark, France, Germany, Ireland, Italy, Luxembourg, the Netherlands and the United
Kingdom. Given that the full dataset is unbalanced, with some countries only included since
the 1990s or even later (see the Supporting Information for details), this focus on a select
number of European countries allows for comparability over time.
Figure 1 presents trends in men’s and women’s ideological positions in these nine
countries. The graph shows a picture that is consistent with much of the earlier research
on the ideological gender gap. At the start of the time series, there is evidence of a (small)
‘traditional’ gender gap. That is, around 1970 women – on average – had a higher score
on the 1–10 ideological self-placement scale, suggesting they positioned themselves to the
right of men. Around the middle of the 1980s, this difference between men and women is
strongly reduced and by the middle of the 1990s the gender difference had disappeared
while a ‘new’ gender gap emerged. Furthermore, the data also suggest that the dynamics in
the ideological gender gap are mostly the result of women shifting their ideological position
to the left. This shift was fairly strong between 1980 and ca. 1995. The over-time analysis
that is presented in Figure 1 thus replicates the findings of previous work – that has relied
on the same data to study changes in the ideological gender gap in European countries
(Abendschön & Steinmetz 2014; Giger 2009; Inglehart & Norris 2003).
More importantly for the purposes of this research note, the over-time analysis that is
presented in Figure 1 also offers insights in how the ‘new’ gender gap has evolved since the
turn of the century. From this perspective, Figure 1 shows a fairly stable gender gap. Women’s
ideological position does not appear to have changed much since the early 2000s. And the
contrast in the pace of change before and after 2000 is in fact quite stark. In this set of nine
countries, women’s position on the left-right axis has shifted from 5.60 to 5.16 between 1980

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
230 RUTH DASSONNEVILLE

Figure 1. Men’s and women’s ideological self-placement in 9 European countries over time, 1973–2018.
Note: Local polynomial smoother line of men’s and women’s ideological self-placement (on a 1–10
ideological scale) and 95% confidence intervals are presented. Based on yearly information on respondents’
ideological self-placement in Belgium, Denmark, France, Germany, Ireland, Italy, Luxembourg, the
Netherlands and the United Kingdom.

and 2000. In the next 18 years, it has only shifted 0.08 points, to an estimated average of 5.08.
In other words, the pace of the change was more than fivefold before 2000 than it has been
after 2000.
These patterns are confirmed when estimating over-time trends in the ideological self-
placement of men and women parametrically. In Table 1, I present the results of six
ordinary least squares (OLS) regression models that serve to estimate changes in ideological
placements over time. To do so, the models include a set of dummy variables to distinguish
between the decades that are covered in the dataset. To estimate differences in the trend
of men’s and women’s ideological positions over time, I add to the models interactions
between respondents’ gender (female = 1, male = 0) and the decade dummies. As can be
seen from the estimates in Table 1, this parametric estimation approach shows a picture that
is consistent with the results in Figure 1. Note that Models 1 and 2 rely on data from the nine
countries that were used in Figure 1.
Looking at the estimates in Table 1, it can be seen, first, that ideological positions were
more left wing from the 1990s onwards than they were in the 1970s and 1980s (Model 1).
Second, the inclusion of the interactions with gender in Model 2 suggests – like Figure 1 – a
strong contrast in the pace of change in men’s and women’s ideological positions until the
1990s, after which the continued increase in the ideological gender gap appears to flatten
off. Importantly, these results appear to hold when broadening the scope of the analyses
and also including countries for which we have data since 19854 (Models 3 and 4) or all 36
countries in the dataset (Models 5 and 6).5
In a dataset of this size, ideological differences can be estimated very precisely. It should
be acknowledged, however, that the size of the effects is substantively rather small. Taking

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE IDEOLOGICAL GENDER GAP 231

Table 1. Gender differences and ideological self-placement over time, 1973–2018

(1) (2) (3) (4) (5) (6)

Countries with data Countries with data

since 1973 since 1985 All countries

Female −0.081*** 0.238*** −0.041*** 0.238*** −0.063*** 0.238***

(0.011) (0.029) (0.010) (0.029) (0.007) (0.029)
1980s (ref: 1970s) −0.049 0.017 −0.116 −0.062 −0.115 −0.060
(0.078) (0.084) (0.078) (0.085) (0.078) (0.085)
1990s (ref: 1970s) −0.262 ***
−0.096 −0.249 ***
−0.109 −0.265 ***
−0.116
(0.073) (0.079) (0.073) (0.079) (0.073) (0.079)
2000s (ref: 1970s) −0.275*** −0.088 −0.251*** −0.092 −0.251*** −0.092
(0.073) (0.079) (0.073) (0.079) (0.073) (0.078)
2010s (ref: 1970s) −0.325 ***
−0.120 −0.334 ***
−0.155 −0.320 ***
−0.140
(0.075) (0.081) (0.075) (0.081) (0.074) (0.079)
Female × 1980s −0.134*** −0.109** −0.111**
(0.036) (0.035) (0.035)
Female × 1990s −0.332 ***
−0.280 ***
−0.299***
(0.033) (0.033) (0.033)
Female × 2000s −0.372*** −0.317*** −0.317***
(0.033) (0.033) (0.031)
Female × 2010s −0.407*** −0.355*** −0.358***
(0.036) (0.035) (0.032)
√ √ √ √ √ √
Country FE
(N) individuals 1,255,958 1,255,958 1,653,829 1,653,829 2,587,118 2,587,118
(N) countries 9 9 17 17 36 36
2
R 0.016 0.017 0.022 0.022 0.027 0.028

Note: Estimates and standard errors (clustered by country-year) are reported. The models include country
fixed effects. Significance levels: * p < 0.05; ** p < 0.01; *** p < 0.001.

the estimates from Model 2, the gap between men’s and women’s ideological positions has
increased by 0.4 points (on a 1–10 left-right axis) in four decades of time (i.e., between the
1970s and the 2010s). Given the length of the time period that is covered by the analyses
(45 years), a half point change on a 10-point scale can be qualified as rather small. While
such small changes could theoretically build up to a more substantive effect, the fact that
the analyses suggest the change has halted in the more recent time period suggest the gap
will remain fairly modest in the foreseeable future.
A country-by-country analysis of gender differences in ideology, and how these have
changed over time, offers further nuance. First, as evident from the country-specific graphs in
Figure 2, whether or not there are differences between men and women varies substantially
between countries. These plots also show substantial heterogeneity in the over-time trends
of men’s and women’s left-right self-placement. Clearly, not all countries closely follow the
average trend that is depicted in Figure 1. This is also evident when parametrically estimating
the interactions between gender and time (decades) in each of the countries separately.

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
232 RUTH DASSONNEVILLE

Figure 2. Men’s and women’s ideological self-placement over time, 1973–2018.

Note: Local polynomial smoother line of men’s and women’s ideological self-placement (on a 1–10
ideological scale) and 95% confidence intervals are presented.

These analyses, that are summarized in Appendix B in the Supporting Information, suggest
that the change is in the expected, negative, direction in most (28) countries. In these
countries, the average marginal effect of the ‘female’ variable is either becoming negative or
growing more negative over time. However, the size of this over-time change is substantively
small in most countries. In fact, there are only 16 countries where the average marginal effect
of being female in the 2010s is significantly smaller (or more negative) than it was in the first
decade for which data are available in a certain country. To summarize, women have changed
their ideological position over time – while men have not. The pace of this change, however,
is not constant. While women’s ideological position quickly turned to the left between the
mid-1980s and the late 1990s, the trend is much reduced since the turn of the twenty-first
century. In addition, there are stark contrasts between countries, with very little evidence
of change in the ideological gender gap over time in about half of the 36 countries under
scrutiny here.

The role of generational replacement

The evidence of women’s ideological realignment is present, but fairly weak. So far, however,
I have studied average differences between men’s and women’s ideological positions –
irrespective of the generations they belong to. In doing so, I might have underestimated

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE IDEOLOGICAL GENDER GAP 233

Figure 3. The ideological gender gap by generation.

Note: Estimates and 95% confidence intervals are presented. Estimates from a model that includes
information on the nine countries for which there are ideological data since 1973. Full results can be
consulted in the Supporting Information.

the extent of change among certain generations (Shorrocks 2018). In addition, even if the
overall change is limited, if the ideological gender gap continues to grow larger among the
youngest generations, there is a real potential for the ideological gender gap to gain impact
as the newer generations continue to gain electoral weight.
To explore this possibility, I estimate a new series of OLS models that include a set of
dummy variables to distinguish between respondents who are members of the Greatest
generation, the Silent generation, Baby-boomers, and Generations X, Y and Z. In addition,
to distinguish generational effects from differences that are driven life-cycle differences, I
account for age and age2 in the models. As previously, the models include country fixed
effects and standard errors are clustered by country.
The full results can be consulted in Appendix C in the Supporting Information, but
Figure 3 summarizes the findings. To be consistent with Figure 1, the effects that are plotted
in Figure 3 are obtained from models that are restricted to countries for which there are
data since 1973.6 First, and importantly, Figure 3 shows that the main contrast is between the
Greatest and Silent generation on the one hand and Baby-boomers and younger generations
on the other. It is in the juncture between the Silent and the Baby-boom generations that
ideological gender realignment occurred, as women of the Silent generation were on average
still more right-leaning than men, while the opposite holds for women of the Baby-boomer
generation. Importantly, however, both men and women among the Silent generation and
the Boomer-generation were more left-leaning than members of the generations before
them. In fact, the general turn to the left – among both men and women – seems a more
important shift than the reversal in men’s and women’s ideological positions. Second,
Figure 3 shows that the ideological position of female members of more recent generations

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
234 RUTH DASSONNEVILLE

(Generations X, Y and Z) is not substantively more to the left than that of the female
Boomers. Across these generations, women’s ideological position is consistently to the left
of that of men and always hovers around 5.2. Third, from the Generation X onwards, the gap
between men’s and women’s ideological positions appears fairly stable. Fourth, changes in
the ideological gender gap among younger generations seem to be driven by generational
differences among men, with men of Generation X, Y and Z being somewhat more right
leaning than male Boomers.
The results in Figure 3 confirm that gender realignment is a result of generational change
(Shorrocks 2018). Importantly, however, they also show that the main shift is now behind us.
The turning point in men’s and women’s ideological positions seems to have been when the
Boomers entered the electorate. In addition, the slight increase in the gap that has followed
appears to be driven by men. Among newer generations of citizens, differences between men
and women appear fairly stable. In this way, Figure 3 also gives a hint at what is to be expected
for the future: while newer generations of citizens will continue to join the electorate and
replace older generations, the growing impact of Generations X, Y and Z will do little to
affect the ideological gender gap.
Importantly, these basic results hold when enlarging the dataset and estimating
generational differences in the 17 countries for which there are data since 1985 or in all
36 countries (see Appendix C in the Supporting Information). In addition, to categorize
different generations, I relied on a fairly general distinction between six groups. In different
countries in the dataset, however, this categorization might reflect more or less well the
differences between groups of citizens that are based on their lived experiences. As an
alternative and more flexible approach, I therefore also estimated models that account for
generational differences through a distinction between 5-year birth cohorts. As can be seen
from the results that are reported in Appendix D in the Supporting Information, doing so
does not substantively alter the conclusions.
The results suggest that the ideological gender gap has been largely stable since the
middle of the 1990s. In addition, the generational patterns that are observed offer little
reason to assume that the gap would increase much in the near future. Given the nature of
the dataset that is used in this note, that combines information from different data sources,
it is important to verify that the inclusion of certain data sources, or the use of a particular
answer scale for left-right ideology, does not drive the results. As clear from the results that
are reported in Appendix F in the Supporting Information, the coefficients of interest are
largely unaffected by controlling for such survey-related elements. That is, the estimates of
the time effects, and those of the interaction between gender and time (i.e., decades), are very
similar to those reported in Table 1. Second, given that substantially more data are available
for some countries than for others (see Appendix A in the Supporting Information), there
is a risk that the overall trends are mostly driven by what happens in these countries. As
an additional robustness test, I therefore estimated the main models (Models 5 and 6 in
Table 1) when weighting all countries for which there are data equally, and when weighting
observations by the population size of their countries. The analyses of weighted data still
suggest a small over-time increase in the ideological gender gap (see Appendix G in the
Supporting Information).


C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE IDEOLOGICAL GENDER GAP 235

Conclusion

The results that are presented in this note suggest that gender realignment, while present
in many countries, reflects a change that is modest in size, that varies strongly between
countries, and that is driven by a change among older generations. The change that is
observed is not a continuous process, as obvious from the fact that after a period of fairly
strong change in the late 1980s and early 1990s, gender differences in ideology have remained
rather stable.
What has caused this change in men’s and women’s ideological positions? The data do
not allow testing causal mechanisms but a look at the timing of the change allows some
speculation of what might and might not have led to the reversal in men’s and women’s
ideological positions. The aggregate-level pattern suggests a trend that coincides with the
end of the Cold War. Even though this historical event might well have contributed to
ideological change, the substantial heterogeneity in the timing and the speed of change that
is visible from the country-level patterns limit its potential impact. In fact, the widely varying
trends in different countries render a single historical event or time period an unlikely
explanation of the change that is observed. The data show a pattern that is more in line
with traditional explanations of ideological change being driven by gradually shifting values
and structural changes in society (Inglehart & Norris 2003). First, the imprint that such
changes leave on citizens match with the findings of strong differences between members
of different generations. Second, the timing of societal changes such as secularization, or an
increased gender equality in the labor market or in higher education – while fairly general
phenomena – varies considerably between countries as well (Iversen & Rosenbluth 2006;
Shorrocks 2018).
The analyses not only confirm the presence of generational differences, this note adds
novel insights in the nature of these generational effects. First, the data suggest a remarkable
contrast in gender differences between the Greatest and Silent generations on the one
hand and younger generations on the other. Second, from the generation of Baby-boomers
onwards, newer generations of women do not appear to take substantively different
ideological positions than generations before them. Third, a recent increase in the gender
gap is driven by male post-Boomers moving to the right.
Women’s ideological realignment appears to be the product of an important – but
substantively small – change between the Silent generation on the one hand and Baby-
boomers and younger generations on the other. These findings imply that most of the effects
of ‘gender realignment’ have probably already been absorbed by the system. As a result,
there is little reason to expect a ‘real’ gender cleavage to start taking shape in the near
future. Even though the ideological gender gap is substantively small, and even though
the results suggest it will likely remain small, those differences are not meaningless. In
most democracies, ideological self-placements are strongly correlated with the vote choice
(Dalton, Farrell, & McAllister 2011). Furthermore, this connection is equally strong for male
and female voters (Dassonneville et al. forthcoming). As a result, even small changes might
have important electoral effects, in particular when elections are very competitive.
The results that are presented in this research note have several important implications.
First, they confirm the role of generational replacement in bringing about social and
political change (Abramson & Inglehart 1986) and suggest that the context in which

C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
236 RUTH DASSONNEVILLE

citizens become politically socialized is key to explaining their political attitudes (Hooghe
2004). The change in men’s and women’s ideological positions occurred as one generation
replaced the other, and has been mostly stable since. Second, the results point to changes in
public opinion that are altogether small – hence testifying of the stickiness of aggregate
public opinion. Third, and finally, the results suggest a large amount of country-level
heterogeneity. Even though the shift of women towards the left can be discerned in all but
a handful of countries, the timing and the size of this shift vary a lot. As such, the results
suggest that future research that seeks to gain a good understanding of the sources of the
change in the ideological gender gap should take into account country-level factors and
conditions.

Acknowledgements

I am grateful to Semih Çakır, Fernando Feitosa, Thomas Galipeau and Eric Guntermann
for excellent research assistance. Previous versions of this paper were presented at the 2019
Conference of the Southern Political Science Association, January 17–19, Austin, Texas, at
a research seminar of the CERAPS at the University of Lille, at the June 2019 GESIS
Political Science Roundtable in Mannheim and at the Comparative Politics and Political
Economy Research Seminar at the University of Konstanz. I thank participants to these
events, and the anonymous reviewers of this journal for their suggestions and comments.
I am also grateful to Liran Harsgor and Catherine Bolzendahl for comments on previous
drafts. Any remaining errors are my own. This project received funding from the Canadian
Social Sciences and Humanities Research Council (Insight Development Grant, 430-2017-
0824).

Online Appendix

Additional supporting information may be found in the Online Appendix section at the end
of the article:

Appendix A. Countries and datasets included in the analyses

Appendix B. Left-right self-placement by gender over time (1973-2018), country specific
estimations
Appendix C. Generational differences, full results and full dataset
Appendix D. Age-period-cohort models with 5-year birth-cohorts
Appendix E. Accounting for autocorrelation in the data
Appendix F. Taking into account scale- and survey-effects
Appendix G. Weighted data
Appendix H. The meaning and relevance of left-right self-placements

Notes

1. Turkey is the only country in the dataset where democracy has declined to a ‘non-free’ status (since 2018).
However, the data from Turkey are restricted to the period before this decline in democracy.
2. See Appendix H for a lengthy discussion of the relevance of left-right self-placements.


C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE IDEOLOGICAL GENDER GAP 237

3. For example, for converting the median value from a 0–10 scale (γ) to a 1–10 value (κ), Powell (2000: 273)
suggests applying the following formula: κ = 5.5 – (.90 × 5 – γ).
4. Doing so implies that data from Canada, Greece, Israel, Japan, Portugal, South Korea, Spain, and the
United States of America are added.
5. The longitudinal nature of the data, with measures of citizens’ ideological self-placement measured in
different years, implies a risk of autocorrelation in the dataset biasing the estimates. To address this
issue, I also verified whether conclusions hold when estimating a series of year-specific models and
plotting the coefficient for the gender-variable over time. As can be seen from the graphs in Appendix
E, these coefficient plots show patterns that are largely consistent with the estimates from Table 1. In
addition, I removed serial correlation from the dataset by taking the difference between the ideological
self-placement of an individual and the mean left-right placement in her country-year. As can be seen
from results that are reported in Appendix E, using these deviations from the country-year means as the
dependent variable in regression models does not affect the conclusions.
6. The estimates for models based on data from countries that were included since 1985 and all countries
can be consulted in the Supporting Information.

References

Abendschön, S. & Steinmetz, S. (2014). The gender gap in voting revisited: Women’s party preferences in a
European context. Social Politics 21(1): 315–344.
Abramson, P.R. & Inglehart, R. (1986). Generational replacement and value change in six West European
societies. American Journal of Political Science 30(1): 1–25.
Box-Steffensmeier, J.M., De Boef, S. & Tse-Min, L. (2004). The dynamics of the partisan gender gap.
American Political Science Review 98(3): 515–528.
Dalton, R.J., Farrell, D.M. & McAllister, I. (2011). Political parties and democratic linkage: How parties
organize democracy. Oxford: Oxford University Press.
Dassonneville, R., Nugent, M., Hooghe, M. & Lau, R.R. (Forthcoming). Do women vote less correctly?
The effect of gender on ideological proximity voting and correct voting. Journal of Politics.
[Link]
Fah, J. (2018). Local polynomial modelling and its applications: Monographs on statistics and applied
probability. New York: Routledge.
Ferland, B. (2017). Ideological congruence over government mandates under majoritarian and proportional
electoral systems. West European Politics 41(2): 350–383.
Giger, N. (2009). Towards a modern gender gap in Europe? Social Science Journal 46(3): 474–492.
Glenn, N.D. (2003). Distinguishing age, period, and cohort effects. In J.T. Mortimer & M.J. Shanahan (eds),
Handbook of the life course. Boston, MA: Springer.
Grasso, M.T. (2016). Generations, political participation and social change in Western Europe. London:
Routledge.
Harsgor, L. (2018). The partisan gender gap in the United States: A generational replacement? Public
Opinion Quarterly 82(2): 231–251.
Hooghe, M. (2004). Political socialization and the future of politics. Acta Politica 39(4): 331–341.
Inglehart, R. (1990). Culture shift in advanced industrial society. Princeton, NJ: Princeton University Press.
Inglehart, R. & Norris, P. (2000). The developmental theory of the gender gap: Women’s and men’s voting
behavior in global perspective. International Political Science Review 21(4): 441–463.
Inglehart, R. & Norris, P. (2003). Rising tide: Gender equality and cultural change around the world.
Cambridge: Cambridge University Press.
Iversen, T. & Rosenbluth, F. (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): 1–19.
Kaufmann, K.M. & Petrocik, J.R. (1999). The changing politics of American men: Understanding the sources
of the gender gap. American Journal of Political Science 43(3): 864–887.
Kroh, M. (2007). Measuring left-right political orientation: The choice of response format. Public Opinion
Quarterly 71(2): 204–220.


C 2020 European Consortium for Political Research
 14756765, 2021, 1, Downloaded from [Link] by Universitat Pompeu Fabra, Wiley Online Library on [30/09/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
238 RUTH DASSONNEVILLE

Lipset, S.M. & Rokkan, S. (1967). Party systems and voter alignments: Cross- national perspectives. New York:
Free Press.
Mannheim, K. (1928). The problem of generations. Essays on the sociology of knowledge. London: Routledge.
Neundorf, A. & Niemi, R.G. (2014). Beyond political socialization: New approaches to age, period, cohort
analysis. Electoral Studies 33: 1–6.
Powell, G.B., Jr. (2000). Elections as instruments of democracy – Majoritarian and proportional visions. New
Haven, CT: Yale University Press.
Shorrocks, R. (2018). Cohort change in political gender gaps in Europe and Canada: The role of
modernization. Politics & Society 46(2):135–175.
Van den Broek, A. (1996). Politics and generations. Cohort replacement and generation formation in political
culture in the Netherlands. Tilburg: Tilburg University Press.

Address for correspondence: Ruth Dassonneville, Département de science politique, Université

de Montréal, Pavillon Lionel-Groulx, 3150 rue Jean-Brillant, Montréal, H3T 1N8, Canada. Email:
[Link]@[Link]


C 2020 European Consortium for Political Research