Introduction
The income participation gap is a stylized fact in political science. Citizens with higher income, on average, tend to be more politically engaged (Dalton, Reference Dalton2017; Gallego, Reference Gallego2015; Schlozman et al., Reference Schlozman, Verba and Brady2012). Recent evidence shows, however, that differences between the rich and the poor are not due to changes in income among adults but are likely the result of different socialization experiences during childhood and adolescence (Jungkunz and Marx, Reference Jungkunz and Marx2022, Reference Jungkunz and Marx2024). Parental socioeconomic status is therefore often considered a prime reason for differences in political involvement.Footnote 1 These findings based on mean differences neglect, however, that variation in political traits is strongly influenced by genetic heritage (Alford et al., Reference Alford, Funk and Hibbing2005; Fowler et al., Reference Fowler, Baker and Dawes2008; Klemmensen et al., Reference Klemmensen, Hatemi, Hobolt, Skytthe and Nørgaard2012b). In fact, genetic influences often explain a larger share of variance in various political attitudes and orientations than siblings’ shared environmental experiences, for example, characteristics in the parental household (Dawes et al., Reference Dawes, Cesarini, Fowler, Johannesson, Magnusson and Oskarsson2014; Hatemi and McDermott, Reference Hatemi and McDermott2012). Furthermore, parental background might provide conditions that make it more or less favorable for genetic influences to affect the development of political involvement (Manuck and McCaffery, Reference Manuck and McCaffery2014; Shanahan and Hofer, Reference Shanahan and Hofer2005). Not accounting for genetic influences might therefore inflate the effect of circumstances in the parental household during political socialization.
In this paper, we investigate the interplay of (parental) income and genetic dispositions in shaping of political involvement. Concretely, we address two question. First, how does the influence of genetic dispositions change across the income scale? And, second, does income retain any explanatory power after controlling for these dispositions? To do so, we use data from the German TwinLife study that sampled children, adolescents, and young adults along with information about their parents and siblings. Our results regarding the first research question show that the influence of genetic heritage on political interest increases substantially with parental income. This indicates that a favorable socioeconomic environment can enhance the impact of genetic dispositions to develop political involvement. Conversely, poverty may be such a powerful negative environmental influence that the effect of genes is suppressed. To address the broader implications of this result for the socialization process, we ask in a second step whether income differences matter above and beyond genetic disposition. This is done with family fixed-effects models among early adults. They support previous research and show no significant effect of income differences on political interest after controlling for family background and genetic influences. We therefore conclude that unequal political interest is primarily shaped by preadulthood influences.
The relationship between genetic influences, parental income, and political involvement
Genetic influences account for substantial variation in a large number or political traits such as interest in politics (Klemmensen et al., Reference Klemmensen, Hatemi, Hobolt, Skytthe and Nørgaard2012b; Kornadt et al., Reference Kornadt, Hufer, Kandler and Riemann2018; Weinschenk et al., Reference Weinschenk, Dawes, Kandler, Bell and Riemann2019; van Ditmars and Ksiazkiewicz, Reference van Ditmars and Ksiazkiewicz2024), political efficacy (Klemmensen et al., Reference Klemmensen, Hatemi, Hobolt, Skytthe and Nørgaard2012b), political knowledge (Arceneaux et al., Reference Arceneaux, Johnson and Maes2012), political ideology (Alford et al., Reference Alford, Funk and Hibbing2005; Funk et al., Reference Funk, Smith, Alford, Hibbing, Eaton, Krueger, Eaves and Hibbing2013; Hufer et al., Reference Hufer, Kornadt, Kandler and Riemann2020), strength of party identification (Hatemi et al., Reference Hatemi, Alford, Hibbing, Martin and Eaves2009a), and social and political participation and voting (Jæger and Møllegaard, Reference Jæger and Møllegaard2022; Kornadt et al., Reference Kornadt, Hufer, Kandler and Riemann2018; Weinschenk et al., Reference Weinschenk, Dawes, Kandler, Bell and Riemann2019, Reference Weinschenk, Dawes, Klemmensen and Rasmussen2023). Typically, estimates range between 30 and 60% of explained variance of the respective attitude or behavior. In contrast, most studies find only a limited impact of shared environmental factors such as the parental household. At first sight, this contradicts many empirical findings in political science and sociology that assign a substantial role to parental background in shaping political socialization (e.g., Janmaat and Hoskins, Reference Janmaat and Hoskins2022; Jungkunz and Marx, Reference Jungkunz and Marx2024; Lahtinen et al., Reference Lahtinen, Erola and Wass2019).
Both findings can, however, coexist if we consider the influence of genes not deterministically, but as dispositions interacting with environments. Genetic dispositions make it more likely to develop a certain political attitude or behavior (Suhay and Jayaratne, Reference Suhay and Jayaratne2013). However, they usually require a favorable environment to unfold their impact (Manuck and McCaffery, Reference Manuck and McCaffery2014; Rabin, Reference Rabin, Sinnott and Rabin2021; van Ditmars and Ksiazkiewicz, Reference van Ditmars and Ksiazkiewicz2024). Similar genes may thus cause different political involvement depending on the social environment in which children are raised. Although previous research indicated that the effect of genetic influence may vary by context (Dinesen et al., Reference Dinesen, Dawes, Johannesson, Klemmensen, Magnusson, Nørgaard, Petersen and Oskarsson2016; Fazekas and Littvay, Reference Fazekas and Littvay2015; Klemmensen et al., Reference Klemmensen, Hatemi, Hobolt, Petersen, Skytthe and Nørgaard2012a), so-called gene–environment interactions (GxE) have rarely been used in political science thus far (but see van Ditmars and Ksiazkiewicz, Reference van Ditmars and Ksiazkiewicz2024; Hatemi, Reference Hatemi2013; Fazekas and Littvay, Reference Fazekas and Littvay2015; Smith et al., Reference Smith, Alford, Hatemi, Eaves, Funk and Hibbing2012; Verhulst and Hatemi, Reference Verhulst and Hatemi2013). For instance, Hatemi (Reference Hatemi2013) found that life events can moderate the impact of genetic dispositions on economic attitudes like support for capitalism or socialism.
The few existing adoption studies further emphasize that both prebirth (i.e., genetic influences) and postbirth factors (i.e., environmental influences) play a substantial role in shaping the likelihood to vote. In particular, they find that the impact of the postbirth environment on turnout is higher among adoptees whose biological mothers do not vote, indicating that there are interactions between the pre- and postbirth influences (Cesarini et al., Reference Cesarini, Johannesson and Oskarsson2014; Oskarsson et al., Reference Oskarsson, Ahlskog, Dawes and Lindgren2022). As a result, the significance of environmental factors was probably understated in many previous studies that solely looked at the environment’s direct effects and ignored how they may have a moderating role on genetic dispositions.
In theory, there are multiple possibilities how the environment can interact with genetic influences (Shanahan and Hofer, Reference Shanahan and Hofer2005; Reiss et al., Reference Reiss, Leve and Neiderhiser2013). For the present research question, there are two broad possibilities. First, the relevance of genetic disposition might increase with socioeconomic status. This could be the case if the effect of genetic dispositions requires an enhancing context to unfold, for example, through the political stimulation in an educated and affluent household. Relatedly, poverty might be a constraining context that crowds out any favorable disposition, for example, through the stress that comes with economic worries. Also in this case, the weight of genetic factors should increase with income. Put differently, such an environment could then be a precondition for variation to occur. Second, socioeconomic status could have a compensatory function that corrects unfavorable dispositions. In this case, genetic differences would explain variation among the poor, but their influence should decline as income increases.
A priori, the first perspective seems a more reasonable expectation. In a related study, van Ditmars and Ksiazkiewicz (Reference van Ditmars and Ksiazkiewicz2024) find that genetic factors matter more for boys’ than for girls’ political interest, arguably because gendered socialization creates more favorable contexts for the former. For other dependent variables, the moderating influence of parental socioeconomic status also often takes the form of an enhancement process. Previous research has shown, for instance, that favorable socioeconomic conditions in childhood can provide an enriched environment in which children can better realize their genetic potential for IQ development (Scarr-Rowe hypothesis, see Rowe et al., Reference Rowe, Jacobson and Van den Oord1999; Scarr-Salapatek, Reference Scarr-Salapatek1971; Bates et al., Reference Bates, Lewis and Weiss2013) and other mental abilities (Tucker-Drob et al., Reference Tucker-Drob, Rhemtulla, Harden, Turkheimer and Fask2011).
Thus, we would expect that higher parental income should also increase the effect of genetic dispositions on political interest, whereas unfavorable socioeconomic circumstances weaken the effect. The latter is particularly plausible in light of existing evidence. Much research has shown that childhood poverty is a highly constraining context for political socialization (see Jungkunz and Marx, Reference Jungkunz and Marx2024, for a discussion and an overview of the literature).
Research design
To study the genetic foundations and contextual influences of adolescents, we use data from the German TwinLife project (Diewald et al., Reference Diewald, Kandler, Riemann, Spinath, Andreas, Baier, Bartling, Baum, Deppe, Eichhorn, Eifler, Gottschling, Hahn, Hildebrandt, Hufer, Instinske, Kaempfert, Klatzka, Kornadt and Weigel2023), which is increasingly used in political psychology (van Ditmars and Ksiazkiewicz, Reference van Ditmars and Ksiazkiewicz2024; Hufer et al., Reference Hufer, Kornadt, Kandler and Riemann2020; Weinschenk and Dawes, Reference Weinschenk and Dawes2019). TwinLife is a panel study of same-sex twins that started in 2014 with four cohorts with an average age of 5, 11, 17, and 23 years. The study includes responses from twins and first-degree family members like parents and siblings, and the main questionnaire is conducted via face-to-face interviews every 2 years. The TwinLife study and the respective experimental protocols received ethical approval from the German Psychological Association (protocol numbers: RR 11.2009 and RR 09.2013). Informed consent was obtained from all subjects and/or their legal guardian(s). Detailed information is provided by Mönkediek et al. (Reference Mönkediek, Lang, Weigel, Baum, Eifler, Hahn, Hufer, Klatzka, Kottwitz, Krell, Nikstat, Diewald, Riemann and Spinath2019) and Lang and Kottwitz (Reference Lang and Kottwitz2020).
What makes TwinLife such a unique and valuable twin study for political socialization research is the age composition of the sample. It samples a large number of twins at very young age and follows them throughout adolescence into early adulthood. Prior research has shown that adolescence is the time in which children’s attitudes are particularly malleable before crystallizing in prime age (Deth et al., Reference Deth, Abendschön and Vollmar2011; Dinas, Reference Dinas2013; Hatemi et al., Reference Hatemi, Funk, Medland, Maes, Silberg, Martin and Eaves2009b). Thus, it is particularly important to study the influence of genetic dispositions at a time when they can (potentially) have the greatest effect on children’s attitudes.
We capture political involvement through political interest, which was measured on a 4-point scale from no to very strong interest in politics. To investigate the moderating role of parental income on the development of political interest, we use the equivalized household income. A detailed summary of all variables and question wording can be found in the Supplementary Material.
We proceed in two steps. First, we decompose the variance of political interest of adolescents (cohort 2 and 3, starting age 11 and 17 years) into genetic and environmental influences using the Classical Twin Design (CTD; Medland and Hatemi, Reference Medland and Hatemi2009). The CTD is based on a comparison between monozygotic twins (MZ) that share 100% of their genes and dizygotic twins (DZ) that share 50% of their genes. It allows us to decompose the variance of political interest into additive genetic influence (A), shared or common environment influences (C), and nonshared or unique environmental influences (E)Footnote 2 (see Verhulst et al., Reference Verhulst, Prom-Wormley, Keller, Medland and Neale2019). Common environmental factors affect both twins similarly (e.g., parental education or wealth), whereas unique environmental influences are experiences that are individual to each twin (e.g., selection of partners or participation in social activities).
Four assumptions underlie such ACE models (Plomin et al., Reference Plomin, DeFries, Knopik and Neiderhiser2013): First, it is assumed that genetic effects are additive and do not interact with each other. The equal environment assumption assumes, second, that MZ and DZ twins are treated equally by their environment (Scarr and Carter-Saltzman, Reference Scarr and Carter-Saltzman1979). While this has been debated in the literature (Horwitz et al., Reference Horwitz, Videon, Schmitz and Davis2003), current research assumes that different treatment between MZ and DZ twins is empirically either not the case or unrelated to the development of political attitudes (Conley et al., Reference Conley, Rauscher, Dawes, Magnusson and Siegal2013; Littvay, Reference Littvay2012; Smith et al., Reference Smith, Alford, Hatemi, Eaves, Funk and Hibbing2012). Third, the model assumes that there is no assortative mating between parents of twins. Nonrandom mating would potentially increase the shared environmental influences in DZ twins. However, assortative mating is widespread, for example, based on education (Blossfeld, Reference Blossfeld2009). We therefore use a correction formula
$ 0.5+0.5\times {h}_0^2\times {r}_p $
where
$ {h}_0^2 $
indicates the relative share of variance of the genetic influences without assortative mating and
$ {r}_p $
the correlation between the political interest of the mother and the father.Footnote
3 Fourth, the basic model assumes that there are no gene-environment interactions or correlations. This would imply that the genetic influence on political interest is independent of common or individual environmental circumstances. As we are interested in such a GxE, we relax this assumption throughout the analysis.
ACE models generally tell us how much variance can be attributed to genetic and environmental influences—and how the weight of these components varies across parental income. This is important information, for example, on the size of a potential omitted-variable bias in analyses of political socialization not accounting for genetic disposition. What ACE models do not tell us directly is how much explanatory power socioeconomic background, as a specific environmental factor, retains after accounting for genetic variation. In other words, we still do not know how changes in income affect changes in political interest when controlling for genetic dispositions. If we want to identify, with a higher degree of certainty, the causal influence of income or other socioeconomic factors, this ultimately is the key question. In a second step, we then use the cohort of early adults (starting age 23) to further test whether income differences result in different degrees of political interest after controlling for shared parental environment in childhood and shared genetic dispositions. Early adulthood is an interesting stage of the life course. It is the time when many young adults have already entered the labor market and earn their own personal income, but where political attitudes are still relatively malleable (Emmenegger et al., Reference Emmenegger, Marx and Schraff2017).
We run family fixed-effects models to estimate the impact of personal income on political interest, which can more credibly be interpreted as causal. These models control for all aspects that are shared by twins and non-twin siblings, for example, their family background. Since previous research has shown that influencing factors of political participation (e.g., education) are not just confounded by characteristics in the family household but also substantially confounded by genetic dispositions (Ahlskog, Reference Ahlskog2021; Dinesen et al., Reference Dinesen, Dawes, Johannesson, Klemmensen, Magnusson, Nørgaard, Petersen and Oskarsson2016), we then restrict the sample further to DZ and MZ twins. This is sometimes called discordant twin design or co-twin control design (McGue et al., Reference McGue, Osler and Christensen2010; Weinschenk et al., Reference Weinschenk, Dawes, Oskarsson, Klemmensen and Nørgaard2021). Using MZ twins as counterfactuals eliminates potentially confounding genetic influences more effectively. We acknowledge that fixed-effects models have limitations of their own. They are still a correlational approach based on variation that is not shared across the siblings, even when they address common familial factors. Moreover, because the discordant twin design is a fixed-effects model, any issues with attenuation bias resulting from measurement error would be exaggerated. Therefore, attenuation bias, family confounding (genes, socialization, etc.), or a mix of the two may be responsible for the observed decline in effect estimates when comparing the naive ordinary least squares (OLS) models to the discordant twin models. That said, we want to underscore that our findings below are in line with a long list of previous evidence that points toward a strong influence of parental income on political socialization during adolescence (see further Jungkunz and Marx, Reference Jungkunz and Marx2024). Technically, we then assume that each twin is nested in a given twin pair and serves as control for the respective other twin. Thus, we can identify the effect of twin differences in unique environmental factors on political interest.
Results
Table 1 shows the results from the ACE models by cohort. The estimates in the left column are based on respondents under the age of 18 in the entire TwinLife data clustered in families and wave. The remaining columns show the results from the first wave, which provides the largest sample size of all waves and which is not subject to potential panel attrition. Overall, we find that about 38% of the variance of political interest of adolescents can be attributed to genetic dispositions, around 9–14% to shared environmental circumstances, and about 48–53% to unique environmental experiences. These estimates are quite similar across cohorts.Footnote 4
ACE model for political interest by cohort
Note. ACE model of political interest corrected for assortative mating. A indicates genetic influences, C influences of the shared environment, and E unique environmental experiences. The mean age is 11 for cohort 2 and 17 for cohort 3. Entries for A, C, and E refer to the percentage of variance as part of the total variance of political interest. r(MZ) and r(DZ) indicate the correlation in political interest among MZ and DZ twins. N refers to the respective number of twin pairs. Pooled results based on respondents under the age of 18, that is cohort 2 from all waves (starting age 11) and cohort 3 from the first wave (starting age 17).
These patterns have important implications for research on political involvement, namely that a substantial share of political interest in adolescents can be attributed to genes. For example, our genes may influence curiosity about complex processes or the willingness to engage in social exchange, which are essential for the acquisition of political sophistication. Although the larger share of variance can be attributed to environmental factors, researchers should keep in mind that the unique environment E is a residual category that includes measurement error and therefore might be inflated. Against this background, the influence of genetic factors is substantively large.
But our main question concerns the interaction of genetic factors with income. Figure 1 shows the results of the ACE decomposition by parental income for the pooled cohorts 2 (starting age 11) and 3 (starting age 17). We report full estimates and model comparison test in the Supplementary Material. In line with our expectation, genetic influences account for little variation when parental income is low (
$ {\beta}_{\mathrm{A}}=0.446 $
, 95% CI [0.369, 0.523];
$ {\beta}_{\mathrm{AxIncome}}=0.085 $
, 95% CI [0.022, 0.148]). For instance, when parental income is one standard deviation below the mean, genes account for only about 23% of the total variance in political interest. The shared/unique environment accounts for 24/53% (
$ {\beta}_{\mathrm{C}}=-0.258 $
, 95% CI [−0.372, −0.143];
$ {\beta}_{\mathrm{CxIncome}}=0.108 $
, 95% CI [0.015, 0.200]). This observation is consistent with the argument that poverty is such a detrimental environment for political socialization that favorable dispositions are constrained. The pattern changes substantially for children with parents of high income (one standard deviation above the mean). Here, genes account for 51% of the variance in political interest and the shared environment only for 4% (the share of the unique environment decreases slightly to 45%). Genetic dispositions for political engagement thus seem to require an enhancing socioeconomic environment to unfold their effect.
ACE decomposition by parental income. Note: Share of variance in twins’ political interest that is explained by genetic (green), common environmental (orange), and unique environmental (purple) factors. Corrected for assortative mating. Pooled results based on respondents under the age of 18, that is, cohort 2 from all waves (starting age 11) and cohort 3 from the first wave (starting age 17). The model is based on 1,111 MZ twin pairs and 1,512 DZ twin pairs.
The results are fairly similar when analyzing only the first wave. However, while the effect on genetic dispositions is quite similar (
$ {\beta}_{\mathrm{A}}=0.424 $
, 95% CI [0.304, 0.544];
$ {\beta}_{\mathrm{AxIncome}}=0.129 $
, 95% CI [0.031, 0.228]), the negative effect on the shared environment is slightly stronger (
$ {\beta}_{\mathrm{C}}=-0.259 $
, 95% CI [−0.431, −0.087];
$ {\beta}_{\mathrm{CxIncome}}=0.176 $
, 95% CI [0.057, 0.296]). Socioeconomic circumstances in the household hence seem to matter more in early adolescence (see Figure B.2 and Tables B.11 and B.12 in the Supplementary Material). Although the differences are not statistically significant, this is in line with previous research. Both social and biological factors, such as parental socioeconomic position and cognitive or personality traits (that are related to political interest), have heterogeneous impacts across the life cycle that are most noticeable in childhood and adolescence (see Reiss et al., Reference Reiss, Leve and Neiderhiser2013).
As discussed in the beginning of this paper, prominent perspectives in political science stipulate an influence of income and other socioeconomic factors on political involvement—be it directly or through unequal socialization experiences. What do our results mean for this assumption of a causal role of income in shaping political interest? While previous research found substantial differences between income groups, cross-sectional data usually does not allow controlling for a variety of aspects that shape political socialization. This includes parents’ socioeconomic situation, parenting styles, and in particular the genetic heritage that we showed to matter above. The twin data allow us to account for all of these aspects. We can, hence, greatly reduce omitted variable bias and approximate a causal test of income effects on political interest.
A particularly suitable way to isolate income effects are family fixed-effects models. By exclusively leveraging variation within siblings (e.g., within a twin dyad), such models eliminate all factors the respective siblings have in common as potential confounders. This includes the shared family environment for siblings in general; the timing of socialization for DZ twins; and, additionally, the genetic dispositions for MZ twins. Table 2 shows the family fixed-effects models from cohort 4 (early adulthood) across the pooled waves 1–3. We restricted the sample to families in which all siblings were employed, and none was currently in higher education or training (which might bias the effects of income differences).Footnote 5 The mean age of cohort 4 in the first wave is about 23 years in the first wave. Due to sample size, we combine all three waves and cluster respondents in families and waves. We report descriptive between- and within-variance of income in Table A.3 in the Supplementary Material. Furthermore, we omitted respondents with very low reported income (personal gross income below 450 euros), as these could stem from measurement error which could inflate within-variation. Robustness checks in Table C.2 including these respondents does not alter the main pattern of our findings.
Family fixed-effects models of political interest on income
Note. Family fixed-effects models indicating the effect of income on political interest with standard errors in parentheses. Income is measured as equivalized gross household income in 1000 Euro. Models “All” include siblings and all forms of twins, “DZ” models include only dizygotic twins, and “MZ” models include only monozygotic twins. Results based on cohort 4 (early adulthood) and pooled waves 1–3. Respondents are nested in families and waves *p < 0.05, **p < 0.01, ***p < 0.001.
Family fixed-effects certainly produce conservative models, as much variance in income is disregarded. Twins, and siblings in general, tend to make similar socio-economic experiences. A within-between decomposition of the variation in income (Table C.1 in the Supplementary Material) shows that the contribution of differences across families is, unsurprisingly, considerably larger than of difference within families. The standard deviation of the former is 922.17 euros compared to 359.20 euros for the latter. On the one hand, this is a restriction of our analytical leverage, which would be higher had we more identical twins with vastly different socioeconomic circumstances in our data set. On the other hand, this is a finding with real-world relevance in its own right. It reminds us that income and other socioeconomic factors, which are prominent variables in political behavior research, have been shaped by preadulthood forces that we often cannot include in our models.
Starting with the naive OLS model as a benchmark, we replicate the commonly observed income gap. Political interest is greater for respondents with higher income (b = 0.113, p < 0.001). In the most basic family fixed-effects model, twins and siblings are nested in families so that we can control for parental socioeconomic background. Restricting the model to only DZ twins further allows us to control for parental upbringing, whereas restriction to MZ twins also controls for genetic dispositions. We find no significant effects of income differences on political interest in early adulthood in any of the models. This likely means that the often-observed income gradient in political involvement is already caused by socioeconomic circumstances in childhood and adolescence. These results are basically identical when we operationalize socioeconomic factors differently, for example, as subjective income satisfaction or as being employed (see Tables C.3 and C.4 in the Supplementary Material).
Discussion and conclusion
Previous research on the income–participation gap has paid little attention to genetic influences. Our results show that economic hardship undermines the influence of genetic dispositions on the development of political interest in adolescence, whereas favorable economic conditions in the parental household facilitate the relationship. These findings are in line with research that showed that parents’ economic worries have a negative impact on family life and parenting styles (Mauno et al., Reference Mauno, Cheng and Lim2017; Schenck-Fontaine et al., Reference Schenck-Fontaine, Lansford, Skinner, Deater-Deckard, Giunta, Dodge, Oburu, Pastorelli, Sorbring, Steinberg, Malone, Tapanya, Tirado, Alampay, Al-Hassan, Bacchini, Bornstein and Chang2020; Voydanoff, Reference Voydanoff2004), which can hamper the development of cognitive and noncognitive skills in the long run (Clark et al., Reference Clark, D’Ambrosio and Barazzetta2021; Dahl and Lochner, Reference Dahl and Lochner2012; Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007). Because such skills have been shown to be crucial for the development of political involvement (Deary et al., Reference Deary, Batty and Gale2008; Holbein, Reference Holbein2017), we find it plausible that poverty is so detrimental for political socialization that it impedes genetic variation to unfold.
A normative implication of this finding—and one that is often emphasized by social scientists—is that reducing child poverty should translate into a desirable increase of democratic engagement in the population. A less often considered implication is that poverty reduction would, at the same time, increase political inequality by genetic background. To better understand this tension between socio-economic and genetic inequality, the social sciences will have to overcome the still widespread aversion to biological explanations. We still have a lot to learn about how genetic factors interact with material living conditions in shaping political inequality over the life course.
An empirical implication of our findings is that income is a questionable predictor of political engagement when used in cross-sections of adults. When one accounts for the joint influence of socialization and genes, as we do with twin fixed-effects, there is no remaining effect of income. The socioeconomic gradient can arguably not be fully understood when life cycle processes and genetic background are ignored. Future research on this topic will have to a) spell-out more clearly which causal pathway is assumed to link income and politics and b) use suitable methods to identify these mechanisms.
Our findings come with some limitations. Most importantly, although core social inequality and sociodemographic indicators in the TwinLife are comparable to the official German Microcensus figures (Lang and Kottwitz, Reference Lang and Kottwitz2020), twins might still be different compared to nontwins in terms of other, less easily measurable characteristics (Hagenbeek et al., Reference Hagenbeek, Hirzinger, Breunig, Bruins, Kuznetsov, Schut, Odintsova and Boomsma2023). Such differences are not necessarily problematic if they are unrelated to political involvement, and it has also been shown that twins do not differ from singletons in terms of internalizing or externalizing problem behavior (Pulkkinen et al., Reference Pulkkinen, Vaalamo, Hietala, Kaprio and Rose2003). It may, however, be the case that growing up with a sibling at the same age affects socialization behavior which can impact political socialization as well (Christensen and McGue, Reference Christensen and McGue2020). Future research should therefore connect findings from twin studies and larger, population-wide studies to increase the validity of the findings. We would like to emphasize, however, that our main finding highlighting poverty as a constraining context for political socialization is in line with previous results based on observational data on young citizens (Jungkunz and Marx, Reference Jungkunz and Marx2024; Abendschön and Tausendpfund, Reference Abendschön and Tausendpfund2017) and natural experiments (Akee et al., Reference Akee, Copeland, Holbein and Simeonova2020).
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/pls.2025.1.
Acknowledgments
Earlier versions of the manuscript were presented at the colloquium of the Mannheim Centre for European Social Research (MZES) and the 2024 conference of the German Political Science Association in Göttingen. We are grateful for helpful feedback from Thomas Bräuninger, Marius Busemeyer, Susanne Garritzmann, Thomas Gschwend, Staffan Kumlin, Arndt Leininger, Sabrina Mayer, Markus Tausendpfund, Nadja Wehl, and the three anonymous reviewers.
Data availability statement
This research article was awarded Open Data and Open Materials badges for transparent practices. All replication materials are available in the corresponding author’s Harvard Dataverse (https://doi.org/10.7910/DVN/VXYSPX). Data can be accessed freely at the GESIS Data Service (http://doi.org/10.4232/1.14186).
Financial support
This research was part of the project “The micro and macro dynamics of political inequality” (PI: Paul Marx), funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation – Project no. 490987363), and the project “Growing into politics under material hardship: The impact of socio-economic problems on political socialization,” funded by the Deutsche Forschungsgemeinschaft (DFG – Project no. 505158386).
Competing interest
The authors declare no competing interests exist.
Open access
