Making Mothers Stay at Home? Analyzing the Impact of Partisan Cueing on Attitudes Toward Maternal Employment

. After decades of value change toward more favorable views of maternal employment, the trend has slowed down and even reversed in some Western countries. This article argues that political parties play a crucial yet neglected role in shaping the trajectories of value change: the dynamics resulting from interparty competition place parties in a position to provide cues to the electorate and to actively shape attitudes toward maternal employment. Partisan cueing should become particularly relevant with a declining relevance of party competition on economic issues. The results from multilevel regression models provide empirical support for this perspective.


Introduction
Over the last decades, we have witnessed a trend in public attitudes in most industrialized countries toward more egalitarian views about family and gender roles. Paid employment of mothers with small children has been increasingly accepted and supported (Davis and Greenstein 2009). Several studies argue, and find, that progressive value change has provided incentives for political parties to take more favorable positions on policies supporting dual earners and the reconciliation of work and family life (Blome 2017;Morgan 2013;Schwander 2018). However, more recently, the trend toward more gender-egalitarian attitudes has slowed down and even reversed in some countries (Braun and Scott 2009;Cotter, Hermsen, and Vanneman 2011). The reasons for this trajectory of attitudinal change and why it varies across countries are still poorly understood.
Recent contributions argue that rising opposition to egalitarian gender norms can be understood as "cultural backlash" among those upholding more traditional attitudes (e.g., Norris and Inglehart 2019). While some studies locate this attitudinal shift within the younger birth cohorts (Pepin and Cotter 2018;Shorrocks 2018), Norris and Inglehart (2019) argue that lowereducated men from older generations who were once well-represented by the dominant cultural values, in particular, tend to feel left behind by progressive societal value change. Accordingly, these feelings would find their political expression in a higher likelihood among those individuals to vote for right-wing populist parties, which most forcefully defend more traditional gender roles (Akkerman 2015;Campbell and Erzeel 2018;Mudde and Kaltwasser 2015;Norris and Inglehart 2019;Van Hauwaert and Van Kessel 2018).
However, a problematic feature of many of those studies is that they consider attitudinal change as exogenous. In this article, I challenge this view. My central claim is that political parties play an important role in shaping public attitudes toward family and gender roles. Building on an extensive but largely disconnected literature on partisan cueing and elite leadership, I argue that the positions and communication emanating from political parties provide cues to the electorate and thus shape public attitudes. The more parties emphasize egalitarian family positions and abstain from propagating traditional gender roles, the more public attitudes should become egalitarian. Partisan discourse on family and gender roles does not operate in a void, but rather is embedded in the wider context of party competition. Building on contributions that highlight the role of economic liberalization and inequality (e.g., Hopkin 2020; Manow 2018), I expect that a declining relevance of partisan competition on economic issues (Beramendi et al. 2015) amplifies the impact of partisan cueing for gender and family issues. A convergence of parties toward more economically right-leaning positions should open up the space for partisan discourse centering on gender and family issues. This discourse may evoke hopes and fears regarding how gender-egalitarian work-family policies affect the well-being of families.
I run multilevel regression analyses to test these propositions using data on party positions from the Comparative Manifesto Project (CMP; Volkens et al. 2019) and on public attitudes toward maternal employment from three waves of the International Social Survey Programme (ISSP 1994(ISSP , 2002(ISSP , 2012. The findings support my arguments outlined above. First, I find that change in party positions on family issues precedes a subsequent change in corresponding attitudes toward maternal employment; and, second, that this partisan cueing effect is particularly pronounced in party systems where parties have converged to positions leaning economically more to the right. The results challenge explanations that interpret attitudinal change as an original explanatory factor for cultural backlash and its implications for electoral politics (e.g., Norris and Inglehart 2019; Van Hauwaert and Van Kessel 2018). Attitudinal change is, at least to some extent, politically constructed by elite partisan discourse. In addition, partisan competition on economic issues matters. My findings contribute to the debate on the importance of material concerns vis-à-vis cultural factors in shaping cultural backlash (cf. Hopkin 2020; Manow 2018), and to the study of right-wing populist parties' electoral appeal and of the political viability of welfare state reforms supporting maternal employment (cf. Ferragina and Seeleib-Kaiser 2015;Garritzmann et al. 2017;Morel, Palier, and Palme 2012).
Determinants and Political Relevance of Attitudes toward Maternal Employment reversed (Braun and Scott 2009;Cotter, Hermsen, and Vanneman 2011). Existing explanations of these attitudinal developments diverge considerably.
The first set of explanations refers to the incomplete nature of the revolutionary rise of female labor force participation. While governments regularly expanded family policies to facilitate female employment, they devoted much less attention to remaining gender inequalities (Daly 2011;Lewis, Campbell, and Huerta 2008). The gendered division of household and care duties did not adjust correspondingly to the rising levels of female labor force participation. As a result, the combination of paid employment and disproportional shares of household and care work results in a double burden for many women and is associated with time and role conflicts (Cotter, Hermsen, and Vanneman 2011;Lewis, Campbell, and Huerta 2008). In addition, in many countries family policy has been increasingly discussed in a context of enhancing "free choice" and individual responsibility of parents in how they allocate time for paid work and care within the household (Daly 2011). Such interpretations neglect that parental choice always takes place within the constraints of the parents' social and economic situation and that the experience of such constraints may contribute to more skepticism toward dual-earner models.
In this context, the impact of family policies shaping the trajectories of attitudinal change appears not to be unidirectional. Some of the more recent family policy reforms strengthened a more traditional division of labor between parents, and such reforms have the potential to reinforce more traditional gender role attitudes (Gangl and Ziefle 2015;Jozwiak 2021). In addition, the benefits of expanded family policies need to be sufficiently accessible to contribute to a substantial change in attitudes (Neimanns 2021a). Thus, in a context of persistent gender inequalities and often still insufficient support by family policies, continuing tensions in parents' attempts to combine work and family life may contribute to a slowing down of attitudes becoming more egalitarian. Such tensions may be particularly strong among individuals from the younger birth cohorts that are directly experiencing those difficulties and that are being socialized under a corresponding public discourse (Pepin and Cotter 2018;Shorrocks 2018).
The second set of explanations links the resurgence of more traditional attitudes to reactions among the older birth cohorts and the rising electoral importance of right-wing populist parties, with those parties tending to promote more traditional gender and family values (Akkerman 2015;Campbell and Erzeel 2018;Mudde and Kaltwasser 2015). While some scholars highlight cultural factors in the form of increasing resistance of individuals from older generations toward progressive value change as the source of this cultural backlash (Norris and Inglehart 2019), others emphasize the importance of increased economic insecurity and inequality for attitudinal change and associated electoral implications (Hopkin 2020;Manow 2018).
Notwithstanding the controversies surrounding the drivers of attitudinal change, a growing body of literature has documented that individual attitudes toward maternal employment are an important determinant of family policy reform (e.g., Ferragina and Seeleib-Kaiser 2015). Parties on the center-left and sometimes also on the center-right have increasingly tried to attract female votes by promoting gender-egalitarian family policies (Blome 2017;Hieda 2013;Morgan 2013;Neimanns 2021b;Schwander 2018). Such findings are in line with broader evidence in the political science literature that parties tend to be responsive to public opinion (Stimson, MacKuen, and Erikson 1995), even if responsiveness can be highly selective (e.g., Gilens 2012) or contingent on a range of conditioning factors (Busemeyer, Garritzmann, and Neimanns 2020;Culpepper 2011;Hobolt and Klemmensen 2008).
While various studies examining party responsiveness analyze changing attitudes toward maternal employment as an independent variable, many of those studies neglect that attitudes are hardly exogenous. Notable exceptions are a few historical-institutionalist studies that take into account the possibility that parties, in top-down rather than bottom-up approaches, may try to actively influence attitudes toward family and gender roles (Bonoli 2013;Fleckenstein 2011;Morgan 2006;Naumann 2012). Such a perspective implies that the strategies parties adopt in political competition can become highly relevant: party discourse may shape public attitudes; parties then respond to these politically constructed attitudes when reforming family policies, possibly affecting the evolution of female labor force participation patterns. However, to date, no quantitative evidence exists on the importance of partisan cueing effects on attitudes toward maternal employment. In the next section, I will elaborate on arguments that account for such an elite leadership perspective of political parties shaping public attitudes toward maternal employment.

Role of Partisan Cueing for Attitudes toward Maternal Employment
A sizable literature has dealt with the influence of party discourse on individual-level attitudes (e.g., Jacobs and Shapiro 2000;Lenz 2009;Schneider and Ingram 1993;Zaller 1992), but this research has often remained disconnected from studies on democratic responsiveness, which have studied to what extent public opinion influences the positioning of political parties and public policy (e.g., Hobolt and Klemmensen 2008;Stimson, MacKuen, and Erikson 1995 ). A prominent assessment of elite influence on public attitudes is Zaller's (1992) model of information processing that explains how public opinion changes in response to new information. Elite communication, for example from politicians via the media, provides cues for the public. Individuals receive this new information, do or do not accept it, and sample it against their individual background of norms, values, and recent experiences.

Partisan Cueing and Attitudes Toward Maternal Employment
How individuals react to new information depends on the ideological proximity to the messenger. An individual is likely to be more influenced by the political articulation of a party he or she feels close to. As a consequence, citizens tend to change their opinion in response to information from media and election campaigns to stay consistent with their preferred political party or candidate (Lenz 2009). Policymakers can use this leverage to influence public opinion before engaging in policy reforms (Jacobs and Shapiro 2000;Schneider and Ingram 1993). Thus, the influence of public opinion on policies will be overestimated if parties are perceived as simply "pandering" to public demands. While many of these studies are concerned with short-term influences on public opinion, there is also evidence that political signals influence attitudes that are believed to be more stable and deeply embedded (Kumlin and Svallfors 2007;Schmidt and Spies 2013).
Elite cueing should be particularly relevant for issues relating to gender roles and related policies. Work-family policies, in particular those addressing the issue of female employment, have a strong value-based component as they critically touch upon potentially conflicting responsibilities between the family and the state, where there have been fierce conflicts in the past (Morgan 2006). Political elites shape such value-based judgments by ascribing positive or negative images to the issue of female employment (cf. Schneider and Ingram 1993). Parties may either proactively promote a dual-earner model or denounce progressive family policies as illegitimate state involvement in the sphere of the family and dual-earner models as an attack on traditional family values and well-being.
The incentives for political leaders to influence public opinion and to potentially deviate from responsiveness toward the median voter or their core electorate may vary (and are discussed in more detail below). From a policyseeking perspective, partisan activists may try to implement policies in line with their vision of society (Bonoli 2013;Morgan 2006;Naumann 2012). In several cases, reforms of family and labor market policy toward a dual-earner model originated not from voter pressure, but from (female) groups within the political parties (Naumann 2012). From a vote-and office-seeking perspective, partisan elites may try to attract specific voter groups to broaden their electoral base. In such instances, political entrepreneurship within the parties becomes particularly important because party leaders may pursue modernizing family policy reforms against substantial opposition from inside the party to attract the votes of employed women (cf. Blome 2017;Fleckenstein 2011;Morgan 2013). In addition, parties are constrained in their positioning by contextual factors such as the countr ys socioeconomic situation and the nature of party competition. For electoral reasons, they may want to increase or decrease the salience of value issues (Tavits and Potter 2015), and they may take into account the behavior and positioning of relevant competitor parties for their own positioning (Schwander 2018).
Irrespective of the particular motives, partisan discourse on family issues should leave its imprint on public attitudes toward maternal employment. H1: The more partisan discourse emphasizes egalitarian positions on family values, the more voters' attitudes to maternal employment develop in an egalitarian direction.
A central constraining factor of party competition is the socioeconomic context in which a country is located. Against the background of recent economic, societal, and political changes, scholars have identified a declining relevance of political contestation on economic issues relative to those of social values across the advanced Western democracies (Beramendi et al. 2015;Kriesi et al. 2006). Party competition has become increasingly twodimensional, with party positions on economic and social value issues having become increasingly independent from each other over time (Hieda 2013;Kriesi et al. 2006). In many countries, parties and party systems have shifted to the right in economic terms (e.g., Ward, Ezrow, and Dorussen 2011). Governments often reduced their efforts to combat rising levels of socioeconomic inequality and engaged in welfare state retrenchment and economically liberalizing reforms (Baccaro and Howell 2017;Hopkin 2020;Manow 2018;Pontusson and Rueda 2010). It is thus of interest to evaluate how such a contextual setting shapes the opportunity structures for parties' positioning on family values and the implications for partisan cueing.
Where parties continue to differ on socioeconomic positions, one can expect partisan conflict on economic and redistributive issues to dominate party competition, with secondary cleavages being subordinated to traditional class conflict (Kriesi et al. 2006). In line with this expectation, Kumlin and Svallfors (2007) have found class conflict in attitudes toward redistribution to be particularly pronounced in countries that have a strong legacy of class conflict. Partisan cueing on value issues, including party positions on the role of the family, should play less of a role in such contexts of marked partisan differences on economic issues.
In contrast, where parties converge on their economic positions, they should have stronger incentives to emphasize value issues in political competition. In particular, economically right-wing parties should have incentives to emphasize more traditional positions on family values. Proposing open retrenchment as a means to attract voters has proven to be no viable electoral strategy in times of "permanent austerity" even for economically right-wing parties (Pierson 1996). Instead, parties on the right can mobilize on traditional social values, which tend to be overrepresented among their supporters (Marks et al. 2006). Mobilizing on traditional social values helps economically right-wing parties to differentiate themselves from their competitors and to draw voters away from voting based on redistributive considerations guided by economic self-interest (Tavits and Potter 2015).

Partisan Cueing and Attitudes Toward Maternal Employment
While parties economically on the left might in principle have incentives to avoid the convergence of positions on economic issues and to emphasize proredistributive issues instead (Tavits and Potter 2015), the embeddedness of those parties in the contemporary liberalized political economies makes it more difficult to deliver on those positions. An electoral way out of this impasse for parties economically on the left has been to compensate a shift to the right on the economic dimension with more egalitarian positions on family values. In the context of rising levels of female employment, economically leftwing parties have often reacted relatively quickly to address the demands for modernized social and family policy, trying to reach out for new voter segments in the middle class (Abou-Chadi and Wagner 2019 ;Boix 1997;Bonoli 2013;Gingrich and Häusermann 2015;Neimanns 2021b).
Whether parties, on average, move toward more traditional or more egalitarian social value positions likely depends on further contextual factors (cf. Schwander 2018). On the one hand, parties with conservative positions on family values have great potential to dominate the political discourse with traditional value issues, which leads other parties to try to avoid sensitive moral issues. Where parties with dedicated anti-dual-earner positions exist, other parties may refrain from emphasizing progressive family policies, as they could easily be attacked by their political opponents and presented as a threat to families' well-being (Manow 2013;Morgan 2006). However, the opposite dynamic is also possible. In a climate that is less hostile to modern views on the family, parties may be pressed to concede their conservative stance on the family in order not to lose middle-class voters (cf. Blome 2017; Bonoli 2013; Morgan 2013; Schwander 2018). The central implication for the importance of partisan cueing effects on attitudes toward maternal employment is that cueing effects, in the one or other direction, should be most pronounced where parties have converged on economically more right-leaning positions.
H2: The more the party system is located to the right on economic issues, the stronger is the effect of partisan discourse related to family issues on voters' attitudes toward maternal employment.
The theoretical framework implies that a shift of economically left-wing parties toward centrist economic positions might provide only limited benefits to those parties. As parties offer fewer alternatives in socioeconomic terms, this exposes them to the risk of a dominant conservative value discourse. A shift of left-wing parties toward more egalitarian family policies with simultaneous neglect of the economic interests of their lower-income constituencies may thus have its limits in terms of political viability. The following empirical analysis tests the propositions elaborated in this section.

Empirical Analysis Data and Methods
The analysis uses data on individual attitudes toward female employment from three waves of the ISSP module on "Family and Changing Gender Roles" (ISSP 1994(ISSP , 2002(ISSP , 2012 and on party positions from the CMP (Volkens et al. 2019). ISSP data is matched to the CMP based on respondents' party preferences. I consider party positions for the election preceding the years in which the ISSP was fielded. The temporal lag between party positions during the election campaign and the measurement of attitudes in the subsequent wave of the ISSP ensures that the effect of party positions on attitudes is assessed, rather than in the opposite direction. The results do not change if the averages of positions from the two elections preceding measurement in the ISSP are used. This alternative model specification provides further confidence that the results are not driven by reverse causation. 1 I use those countries that were included in at least two consecutive ISSP waves. Overall, the sample comprises 31,234 individuals nested in seventeen countries (fortythree country-years), and ninety-four parties (233 party-years), respectively. 2 I refrain from including Eastern European countries in the main analysis. The instability of the party systems in these countries during the observation period complicates the matching of parties and party supporters over time. Electoral volatility and high numbers of newly founded parties running for elections and old ones disappearing have often been characteristic features in these young democracies (Tavits 2008). While the matching of parties and partisan supporters is more difficult in Eastern European countries, it should nevertheless be possible to observe evidence of partisan cueing at the country level. In an additional step of the analysis, I therefore add six Eastern European countries included in the ISSP in 2002 and 2012 (Czechia, Hungary, Latvia, Poland, Slovakia, and Slovenia) to the regression models to examine to what extent the arguments about partisan cueing might travel beyond the context of Western countries.
Attitudes toward maternal employment are the dependent variable. This variable is taken from a battery of items included in the ISSP tapping attitudes toward family and gender roles. More specifically, it is asked: Do you think that women should work outside the home 1: full-time, 2: part-time or 3: not at all under the following circumstances? When there is a child under school age.
I select this item because it represents a relatively direct measure of support and skepticism regarding maternal employment (cf. Kangas and Rostgaard 2007). In addition, the item explicitly mentions the employment of mothers with children below school age. It is here that public attitudes have continued to be highly divided and the role of the state in facilitating maternal employment or making it more difficult has been highly contested (Morgan 2006, Partisan Cueing andAttitudes Toward Maternal Employment 2013). In this context, partisan discourse should play a decisive role in shaping citizens' assessment of maternal employment.
On average, 40 percent of respondents think that mothers with pre-school children should stay at home and 48 percent that they should work part-time. Because the share of respondents supporting full-time employment of mothers with pre-school children (12 percent) is rather low, and to ease the presentation of the results by running logistic regression models, I recode the variable and assign a value of 1 to full-time and part-time employment and 0 to preferences to "stay at home." The results stay the same if the original categorical coding of the variable is used in ordered logistic or probit models.
Supplementary table A.5 reports country-average values of attitudes toward maternal employment for the three waves of the ISSP. On average, individuals have become more supportive of maternal employment (1994: 45 percent of individuals; 2002: 58 percent of individuals; and 2012: 69 percent of individuals), but there has been substantial variation across countries. In some countries, attitudes barely changed between the ISSP waves, and in Austria, the Netherlands, and Japan, individuals became less supportive of maternal employment between 2002 and 2012. The analysis aims to uncover to what extent changes in party positions may account for these varying trends in attitudes.
I include a range of individual-level control variables that are standard in studies on the determinants of attitudes toward maternal employment (e.g., Neimanns 2021a; Sjöberg 2004). Including these control variables is particularly important in the context of this study to control for possible changes over time in the composition of parties' electorates. I control for gender, age, educational attainment, household income, employment status, having children, living together with a partner, and partner's employment status. 3 The detailed operationalization of the variables is described in Supplementary table A.1.
The central independent variable captures party positions on the role of the family based on CMP data. The CMP measures party positions on the ideological left-right continuum based on parties' statements on fifty-six issue categories coded from party manifestos (Volkens et al. 2019). Not least due to its broad coverage, the CMP has become one of the most widely used sources to measure party positions. I use scores on party statements toward a positive and a negative view on traditional morality (per 603 and per 604) as a measure of party positions on the role of the family. On the egalitarian side (per 604), the measure covers support for divorce and abortion, support for a modern family composition, and a separation of the church and the state. On the conservative side (per 603), it includes suppression of immorality, stability of the traditional family, and support for the role of religious institutions. The scale is constructed in a way that higher values represent more egalitarian positions (egalitarian family values ¼ per 604 -per 603). Lowe et al. (2011) similarly propose using these two items as reflecting a "traditional morality" policy dimension. They validate this scale by comparing it with expert survey data on party positions by Benoit and Laver (2006), which has been proven to provide valid and reliable estimates. I construct this variable at the party-year level and as country-year averages. For the latter measure, parties are weighted by vote share to avoid the excessive influence of small parties with extreme positions.
On average, parties shifted slightly toward more egalitarian positions over time (Supplementary table A.5). However, as has been the case with individuallevel attitudes, there is substantial and persistent variation in positions both across and within countries, implying that party positions on family values continue to differ in substantial ways (Supplementary table A Obviously, the positional statements in party manifestos should be considered a rough proxy of party positions and discourse related to family values and, more specifically, maternal employment. While a distinct advantage of the CMP data is its wide temporal and spatial coverage, a drawback is that it also comes with the potential for serious measurement error (e.g., Gemenis 2013). Measurement error might be a particularly relevant issue in this study because the overall salience of the two items on family values is relatively low. On average, parties devote only 2.55 percent of their manifestos to family values, compared with 41.18 percent to economic issues (see below for the definition of economic issues). The relatively low salience of family issues in party manifestos may underestimate their actual relevance in partisan positioning and communication, as several studies have found that these issues have often clearly mattered in electoral competition (Morgan 2013;Schwander 2018). As cross-validation, I compare my CMP-based measure of party positions on family values to data by Schwander (2018), who reports party statements toward women and families in electoral campaigns coded from newspaper articles for three countries for various elections in the early 2000s. The comparison of both types of data in Supplementary figure A.4 reveals a similar positioning of parties and how it varies across countries, which supports using the CMP data as a suitable proxy. I also discuss two exemplary country cases (Austria and the Netherlands) in greater detail in the Supplementary Appendix (table A.12 and the associated discussion), which furthermore demonstrates that party positions as coded from the manifestos reflect the broader political discourse in those countries.
As argued above, I expect the economic positioning of the party system to affect partisan cueing on family issues. I calculate the economic positioning of Rather than examining the influence of party positions, research on policy feedback has examined how the design of family policies affects attitudes toward maternal employment (e.g., Neimanns 2021a; Zoch and Schober 2018). I expect partisan cueing effects to exist independent of possible policy feedback effects. Evaluating the relationship between family policy generosity (as proxied by the level of public spending on childcare) and changes in party positions on family issues lends plausibility to this reasoning (Supplementary figure A.1). There is no association between levels of public childcare spending and shifts in party positions on family issues. Thus, party discourse may evolve independent of the status quo of family policy, and it may have distinct effects on individual-level attitudes.
I run multilevel logistic regression models with random intercepts for country-years and party-years to account for the nested structure of the data. All models include period fixed effects to control for common time trends that are unrelated to cross-country and cross-party variation in party positions. In a first step, to probe Hypothesis 1, I examine the effects of party positions on family issues at the level of party-years. This takes into account that cueing effects should be particularly strong among the electorate of a given party. Because I am interested in variation over time in attitudes and positions and to control for idiosyncratic effects of parties, I include party fixed effects in this model specification. A major limitation of using repeated cross-sections of survey data is that, despite including an extensive list of individual-level control variables, it is impossible to rule out that the effects of party positions on attitudes of the electorate of a given party might be due to a changed composition of the electorate. For this reason, I specify additional models that include average party positions at the country-year level and replace party fixed effects with country fixed effects. If these additional models replicate cueing effects at the party system level, this ensures that cueing effects found at the party level are not simply due to changes in voting behavior. Hypothesis 2 postulates that at the party system level the effect of party positions related to family issues on attitudes is conditional on parties' positioning on economic issues. To evaluate this claim, both variables interact in this step of the analysis.
I run various alternative model specifications to assess the robustness of the results. This includes running ordered logistic and (ordered) probit models, and controlling for female employment rates to account for the possible influence of broader attitudinal trends that are not due to the effects emanating from political parties. I briefly report on these models at the end of the empirical section.
A limitation of the research design is that the causal effect of partisan positioning on individual attitudes cannot be examined directly. While qualitative studies underscore the plausibility of the postulated causal effect (Manow 2013;Morgan 2006), the strength of the analysis covering a wide range of time and countries is its external validity. Although this approach comes with limitations regarding internal validity, it has the potential to contribute significantly to existing research on partisan cueing effects for other issue areas. This research has often relied on cross-national data at one time point (Schmidt and Spies 2013), on panel studies in one country (Lenz 2009), or on evidence from survey experiments (Stoeckel and Kuhn 2018). Evidence of partisan cueing across time and countries would provide further external validity to these results.

Results
Model 1 in table 1 reports partisan cueing effects on attitudes toward maternal employment at the level of part-years. The results provide strong support for Hypothesis 1: the more parties emphasize egalitarian positions on family issues, the more egalitarian attitudes toward maternal employment become. The effect of partisan cueing is highly statistically significant and substantial in size. A shift in party positions of 3.28 points (which equals one standard deviation of change in party positions) is associated with a change in attitudes by 2.16 percentage points (0.31 of a standard deviation; cf. Supplementary table A.5). 4 The effect estimates of the individual-level control variables are in line with what could be expected based on earlier research on the determinants of maternal employment that reflect material self-interest and exposure to norms and ideas related to gender roles (e.g., Davis and Greenstein 2009;Neimanns 2021a). Women, more educated individuals, and those with higher income are more supportive of maternal employment. A stronger labor market attachment of respondents and partners is associated with more egalitarian attitudes, whereas having kids goes along with more traditional attitudes. Generational effects, which play a central role in Norris and Inglehart's (2019) work, do also matter, with younger birth cohorts having more egalitarian attitudes. Period dummies are included to control for general time trends unrelated to variation in party positions. The significant positive coefficients for the years 2002 and 2012 demonstrate that in comparison with the reference year (1994), attitudes have become more egalitarian over time. In addition to this independent time trend, however, party positions matter. 5 While the individual-level control variables included in table 1 keep the composition of partisan electorates constant, with the repeated cross-section data at hand it is difficult to rule out that partisan cueing effects could be the result of changes in voting behavior. To account for this possibility, Model 2 estimates partisan cueing effects at the country-year level. Because cueing effects can be expected to be strongest for the party an individual feels close to (cf. Zaller 1992), effects at the country level might be less pronounced compared with the party-level effects in Model 1.
The coefficient estimate of aggregate party positions on family issues in Model 2 is positively signed but falls slightly short of statistical significance. This implies that the association between aggregated party positions on family issues and Continued attitudes toward maternal employment in the electorate is less tight as compared with the cueing effects for parties and their respective supporters. Hypothesis 2 specified that cueing effects should be most pronounced where party systems have shifted to the right on economic issues. To test this hypothesis, Model 3 adds the positioning of the party system on economic issues to the model (and its interaction with party system positions on family Note: Model 1 includes party dummies; Models 2 and 3 include country dummies. ***P < 0.001; **P < 0.01; *P < 0.05. issues). It is important to note that all four constellations of shifts to the left and right on economic and family value issues occur empirically. As Supplementary table A.11 demonstrates, shifts below and above the median value of economic and social value positions at the party system level are distributed roughly equally. Thus, with a declining relevance of party competition on left-wing economic issues, party systems do not move toward more egalitarian positions on family values (Bonoli 2013), nor does a shift toward more traditional positions on family values dominate (Tavits and Potter 2015). This implies that even if parties might have a first-order preference to move toward more egalitarian positions (for economically left-wing parties) or more traditional positions (for economically right-wing parties) on family values, they may be unable to implement their preference due to constraints imposed by party competition. When accounting for positional shifts on the economic dimension (Model 3 in table 1), party system cueing effects increase in size and become statistically significant. The negative interaction term indicates that cueing effects are more pronounced in party systems that have shifted more to the right in economic terms. To illustrate this effect, figure 1 plots predicted probabilities of attitudes toward maternal employment, contingent on changes in party positions on economic and family issues. Figure 1 shows that where party systems have shifted to the left, aggregate positions on family issues do not affect attitudes toward maternal employment in the electorate. Attitudes are highly similar, independent of parties' positions on family issues. However, where parties have shifted to the right in economic terms, it matters crucially what positions parties take on family issues. If parties have at the same time emphasized more strongly egalitarian positions on family issues, attitudes have become more egalitarian. Conversely, a shift toward more traditional positions on family issues is associated with a more traditional trajectory of attitudes toward maternal employment. Wald tests indicate that these effects are statistically significant at the 95 percent level. The magnitude of the difference in attitudinal change is 3.74 percentage points (i.e., approximately half of a standard deviation; cf. Supplementary table A.5) between party systems that coupled an economic right shift with more egalitarian family positions (at the 75th percentile) as compared to more traditional positions (at the 25th percentile). Party competition on the economic dimension conditions partisan cueing effects on family issues: aggregate cueing effects are particularly pronounced where parties have shifted economically more to the right.

Robustness Issues
The results are robust to various alternative model specifications. This includes running random intercept ordered logistic and (ordered) probit models (Supplementary table A .2). Controlling for the female labor market participation rate (OECD 2020) to account for the possible influence of rising female labor force participation on both party positions and individual-level attitudes also does not alter the findings (Supplementary table A.3). To rule out the risk of reversed causation (i.e., the possibility that changes in attitudes could drive changes in party positions) I re-estimated the models using the averages of party positions from the two elections preceding the waves of the ISSP. The results do not change with this model specification (Supplementary  table A .4).
Previous research argued that cultural backlash should be driven in particular by lower-educated men from older birth cohorts because they should feel most threatened by progressive value change (Norris and Inglehart 2019). In a separate step of the analysis, I test whether those individuals might also be most susceptible to partisan cueing. To do so, I specify a category of men without tertiary education born before 1945 and interact this category with party positions. The results (Supplementary table A  employment. However, the interaction terms with party positions are insignificant. This suggests that partisan cueing matters for lower-educated men from older birth cohorts, and it does so similarly to the remaining population. As a final step, I briefly discuss the results of additional models adding six Eastern European countries. The effects of partisan cueing become less pronounced in these models, and they are insignificant for the country-year level effects in Models 2 and 3 (Supplementary table A

Conclusion
Over the past decades, most Western countries have witnessed progressive value change toward attitudes becoming increasingly supportive of egalitarian family values and gender roles. Governments reacted to these attitudinal changes and their electoral implications and implemented family policies supporting maternal employment and the reconciliation of work and family life (Blome 2017;Morgan 2013;Schwander 2018). Several scholars have noted that this trajectory of value change has slowed down and even reversed in some countries, but disagreement remains regarding the reasons for these developments. While some scholars emphasize the role of a cultural backlash as a counter-movement of those who were once at the center of the dominant value discourse but now feel left behind by progressive value change (e.g., Norris and Inglehart 2019), others highlight the role of persisting gender inequalities and deficiencies in family policies supporting maternal employment (e.g., Jozwiak 2021; Lewis, Campbell, and Huerta 2008), or of economic inequality and insecurity more generally that may amplify latent traditional attitude traits (e.g., Hopkin 2020; Manow 2018).
A shortcoming of many of those studies that consider value orientations as an explanatory factor is that they disregard that value orientations can hardly be considered exogenous. In this article, I have argued that political parties, by providing cues to the electorate, likely play an important role in shaping trajectories of change in attitudes toward maternal employment and that these partisan cues on family value orientations should be particularly important in contexts where parties have converged to the right in their economic positions.
The results of multilevel logistic regression models of party positions and attitudes toward maternal employment across seventeen Western countries observed up to three times over a period of nearly twenty years corroborated these claims. The more parties emphasized egalitarian family values, the more attitudes developed in an egalitarian direction. This evidence of partisan cueing has been particularly pronounced in contexts where parties have shifted to the right in economic terms.
A limitation of this study is that it has been unable to explicitly test the causal effect of party positions on individual attitudes. Nevertheless, the finding of robust statistical associations between shifts of lagged party positions and individual attitudes across time and countries lends further external validity to evidence of partisan cueing in other issue areas that relies on only one country or one point in time (Lenz 2009;Schmidt and Spies 2013;Stoeckel and Kuhn 2018). Future research would benefit from applying multi-methods approaches, including in-depth case studies and the analysis of panel and experimental data, to be able to draw robust causal inferences to corroborate the existence of partisan cueing effects in the realm of family and gender issues.
Notwithstanding such limitations, the findings in this article have important implications for our understanding of the trajectories of maternal employment, the politics of welfare state reform, and of political representation more generally. The literature on partisan cueing has to date often remained disconnected from studies on the politics of work-family policies (e.g., Blome 2017; Ferragina and Seeleib-Kaiser 2015;Morgan 2013;Schwander 2018). My findings of politically constructed attitudes highlight that the absence of family policy reforms supporting dual-earner arrangements does not necessarily respond to the interests of a traditionally minded electorate or to some sort of a "cultural backlash." Instead, it can signal party elites' failure to bring such issues into the political discourse (cf. Manow 2013). Thus, it appears problematic to place too much weight on attitudes as an explanatory factor on its own. Attitudes are to some extent politically constructed and as such reflect social, economic, and political context, filtered through dynamics at the level of political elites.
The finding that partisan cueing effects are particularly pronounced in contexts where parties have moved to the right on economic positions highlights the risk of cultural backlash being associated with economic liberalization (Hopkin 2020;Manow 2018), and the political vulnerability of parties' strategic reorientation toward the middle class. Where parties fail to offer alternatives to economic liberalization, the risk of a dominant traditional value discourse may counteract the attempts in particular of traditional left-wing parties to win voters in the middle class through social investment (e.g., Beramendi et al. 2015).
Some open questions remain for future research. While party positions on economic and family value issues have been the central independent variables in this analysis, future work could explore in more detail how the socioeconomic context or patterns of party competition constrain parties' positioning on family issues (cf. Schwander 2018). Future research could also apply the research design based on repeated cross-sections of survey data to study partisan cueing effects on issues such as attitudes toward redistribution or immigration to examine to what extent partisan cueing shapes popular attitudes in these issue areas.

Funding
This work was supported by the European Research Council (grant number 311769).