Abstract
The intergenerational transmission of political orientations has been the topic of considerable research over the past few decades, but much of the evidence remains limited to two-party systems. In this study, we use data from the first wave of the Parent–Child Socialization Study conducted among 3,426 adolescents and their parents in the Flemish region of Belgium. Even in this multiparty system, we find a strong correspondence between voting intentions of parents and children, enhanced by the degree of politicization within the family. Talking about politics among parents and children has a significant positive effect on parent–child party correspondence, and more particularly political discussion with one’s father seems to have a stronger effect on father–child party correspondence than discussion with one’s mother does on mother–child correspondence.
Keywords
Introduction
The intergenerational transmission of political preferences has been the subject of numerous studies, particularly in the United States. Seminal studies by Jennings and Niemi (1981), Westholm and Niemi (1992), or Achen (2002) have led to the conclusion that parents have a strong influence on adolescents’ party preferences. Recent developments in political behavior, however, such as increasing volatility and the decline of partisanship, have changed the basic characteristics of electoral outcomes. Where party identification traditionally was considered to be a stable attitude, developed at a young age and shaped by socialization agents, party identity has become increasingly weak and hardly different from party preference in many European countries (Thomassen & Rosema, 2009). With regard to political preferences, there is now a clear distinction between older and younger age cohorts, leading to the question whether intergenerational transmission still contributes to patterns of stability (Hooghe, 2004). Therefore, the main research question in this article is to what extent parents can still be seen as important socialization agents when it comes to voting intentions.
Although quite some research has been conducted in this area over the last few decades, the bulk of this research was focused on the two-party system of the United States, or as Virginia Sapiro (2004) puts it in her review on political socialization research, “Clearly, we have learned too little about political socialization outside the United States” (p. 5). In this article, our aim is to apply the basic concepts of this research line within the context of a fragmented multiparty system. The Belgian case is a theoretically relevant “hard case,” which allows us to ascertain whether the strength of intergenerational transmission of party preferences that has been found in political systems with high levels of party identification is also present in a very fragmented multiparty system, with a coming and going of new political parties that often focus on one issue (Deschouwer, 2009). While the identification with one party traditionally has been quite strong in the 20th century in Belgium, this linkage between citizens and political parties has been weakened, leading to a decrease of party membership (Van Biezen, Mair, & Poguntke, 2012), an increase of electoral volatility (Dalton, McAllister, & Wattenberg, 2000; Lachat, 2007), and an ongoing process of dealignment (Dalton, Farrell, & McAllister, 2011).
There are four reasons why an analysis of this process in Belgium can offer an interesting contribution to the existing literature. First, in a fragmented and volatile multiparty system like Belgium, vote choices are less clearly delineated than they are in a two-party system. If we find evidence for intergenerational transmission even in this context, this would imply that this mechanism is still very important. Second, the ideological and policy differences between different parties in Belgium are not as large as they are in most two-party systems. Third, social and political dealignment has led to a weakened identification with a single party in Belgium, leading to high levels of electoral volatility. Fourth, because of these three considerations, we can expect less parental consistency, less clear and stable parental preferences and, therefore, less accurate perceptions by the adolescent, leading to fewer opportunities for parents to have an impact on the transmission of their vote choice to their children (Knafo & Schwartz, 2003; Okagaki & Bevis, 1999). This means that if we find the same mechanism of intergenerational transmission in Belgium as we do in stable two-party systems, we can establish in a convincing manner that not only strong party identifications, but also more volatile and temporary vote choices can be transmitted from parents to children.
Previous research has identified a number of facilitators for an effective transmission of party preferences between generations. One of the most important facilitators is family politicization. Next to a limited number of basic control variables, we analyze the effect of this variable on the correspondence between parents and children using structural equation modeling. The analysis is based on the results of the first wave of the Parent–Child Socialization Study (PCSS), conducted between January and September 2012 among 3,426 15-year-old Belgian adolescents and their parents. The major advantage of this dataset is that children and each of their parents were surveyed independently about their voting intentions, which leads to reliable information on the voting intentions of parents and their children.
Literature
Parents, individually or as part of a social network, can have a powerful and consistent impact on the political preferences of young citizens (Cross & Young, 2008; Gidengil, O’Neill, & Young, 2010; Levine, 2005). The transmission of political attitudes is dependent on several characteristics, with socioeconomic status (SES) of the parents as one of the most important determinants (Verba, Schlozman, & Burns, 2005).
Jennings and Niemi (1981) made one of the most important contributions to this line of research. They found that parents have a direct and strong effect on their children’s political preferences, with mediating effects of the degree of “politicization” within the family and the stability and homogeneity of parents. They confirmed earlier findings of—inter alia—Campbell, Converse, Miller, and Stokes (1960), and Hyman (1959), who stated that parents are indeed one of the most important socialization agents when it comes to party preferences. Furthermore, party identification and political orientations that are transmitted between generations seem to be less susceptible to change than preferences and values developed independently from parents (Kroh & Selb, 2009).
Although several authors have offered evidence for this hypothesis of intergenerational transmission, others suggested that these effects may well be overestimated or should at least be supplemented with other explanations. Kraut and Lewis (1975), for example, found that parents have a considerable influence on their children’s preferences, but that they can also contribute to the formation of a conflicting ideology, caused by personal conflicts. Connell (1972), on the other hand, points out that socialization researchers tend to overestimate parental impact, while in practice older and younger generations have developed their opinions in parallel, rather than in series, due to similar experiences and a common way of life. Glass, Bengtson, and Dunham (1986) also point out that status inheritance processes account for a substantial proportion of parent–child similarity. Furthermore, they emphasize the possibility of reciprocal or trickle-up influence. This particular hypothesis was elaborated by McDevitt and Chaffee (2002), who propose an integrated family communication model, in which child-initiated discussion can have an influence on parental political attitudes and behavior (Zuckerman, Dasović, & Fitzgerald, 2007). Achen (2002), finally, interprets the resemblance of parent–child party identification from a rational choice perspective: Children learn something about their own most likely future benefits by looking at their parents’ situation as a prior belief and they update their party identification likewise.
To summarize, this review of the literature makes clear that there are two main views on the political socialization process. The first is a traditional view that defines the transmission between parents and children as a direct socialization effect. Other scholars suggest that the correspondence between parents and children is not due to party identification inheritance, but is indirectly caused by status inheritance.
Another ongoing debate focuses on the strength of parental socialization by fathers and mothers. Early work suggested that the father had the strongest impact on transmission of party identification, while other empirical studies showed the opposite: Mothers can have a stronger impact on the choices of their children, for example, because they tend to spend more time with them (Coffé & Voorpostel, 2010). However, another frequently used framework is the gender-specific socialization theory, which states that mothers have more influence on the party preferences of daughters, while fathers have more influence on the party preferences of sons (Nieuwbeerta & Wittebrood, 1995). This pattern can be explained by the fact that for girls it is easier to identify with their mother than with their father, and vice versa. Already at a young age, girls and boys develop gender-specific orientations toward politics, and it can be expected that the same-gender parent functions as a role model in this regard (Hooghe & Dassonneville, 2013; Hooghe & Stolle, 2004).
Research in multiparty systems has shown that even in countries where voters are confronted with a wide range of parties, ideological orientation and party preferences can be successfully transmitted between generations (Westholm & Niemi, 1992). However, it should be clear that in this case, transmission is not as clear and one dimensional as it was shown to be in the American two-party system. Percheron and Jennings (1981) were among the first to investigate the transmission of political attitudes outside the United States. They introduced the hypothesis that the structure of the party system might determine the degree to which children acquire partisanship and ideological positions (Percheron & Jennings, 1981; Sapiro, 2004). The transmission manifests itself not only and certainly not mainly in an accordance of party preferences (Westholm & Niemi, 1992). Ventura (2001) found for example that Dutch children resemble their parents in their self-placement on a left–right continuum, rather than in their specific party affiliation. Other scholars did find clear evidence that also in multiparty settings, a party preference can be successfully transmitted (see Coffé & Voorpostel, 2010, on the transmission of radical right SVP support in Switzerland). It has to be noted that within this line of research, various dependent variables have been used, depending on the specific characteristics of investigated party system. In political systems where party identity can be considered as important, mostly identification with a party has served as the main dependent variable. In multiparty systems, however, the concept of party identity is less obvious and mostly political preference or voting intention serves as the dependent variable (Rosema, 2006). Given the fragmented nature of the Belgian party system, voting intention seemed to be the most reliable measurement in this specific study. Therefore, we will focus exclusively on voting intentions as a dependent variable in the remainder of this article.
Facilitating Conditions
In the transmission of voting intentions, a number of facilitators enhancing this process can be distinguished. A first one is the degree of parental homogeneity. Politically homogeneous couples are believed to be more effective in transmitting their own political attitudes to their children than politically heterogeneous parents (Nieuwbeerta & Wittebrood, 1995). If both parents have different political preferences, it can be assumed that children will be exposed to competing or even contradictory information, and this might neutralize the effect of any socialization experience.
A second facilitator is the degree of communication, interaction, and salience of politics within families (Jennings, Stoker, & Bowers, 2009; Kroh & Selb, 2009; McDevitt & Chaffee, 2002). Talking about politics and emphasizing the importance of certain issues have a positive effect on a successful transmission. A third facilitating condition, operating on a macrolevel, is the party system. Westholm and Niemi (1992) stated that parent–child similarities will be less frequent in countries with less stable and more fragmented party systems. One of the main explanations is that in these countries, party preference is not based on a stable party identification (Brynin & Sanders, 1997; Thomassen & Rosema, 2009).
Hypotheses
When we investigate the intergenerational political transmission between parents and children in a multiparty setting, there are a number of reasons to assume that the resemblance between parents and children will be weaker than in some of the traditional research. First, there are the ongoing processes of dealignment, decreasing party membership, and increasing volatility, which have been weakening the linkage between voters and political parties (Dassonneville, 2012; Lachat, 2007). Second, one of the characteristics of contemporary multiparty systems is that the traditional-form party identification does not play the same dominant role as it does in countries such as the United States. As party preference is not a stable and salient personal attitude, the odds that this attitude will be effectively transmitted to the future generation are smaller. Third, we are confronted with a younger generation with a declining interest in electoral politics that expresses itself in lower levels of interest in parties and political institutions (Dalton, 2008; Henn, Weinstein, & Forrest, 2005; Putnam, 2000; Quintelier, 2007; Zukin, Ketter, Andolina, Jenkins, & Delli Carpini, 2006).
Based on this review of the literature, our main hypothesis is:
Next, we take into account the effects of politicization within the family and the fact that the influence of parents will be different according to the gender of the adolescents with two additional hypotheses:
Data and Method
PCSS
The analysis will be based on the results of the first wave of the PCSS, conducted in the spring of 2012 among adolescents and their parents in Belgium (Hooghe, Quintelier, Verhaegen, Boonen, & Meeusen, 2012). In this survey, a representative sample of 3,426 adolescents was interviewed during school hours using a written self-administered questionnaire, while they received a similar questionnaire for their parents that could be filled out at home (separately by both parents if the adolescent still had contact with both parents). In total, for 60.8% of all adolescents, both parents returned the questionnaire, while for 72.7% of the adolescents, at least one parent returned the questionnaire.
The 3,426 14- to 15-year-old respondents were selected using a stratified random sample of 61 Dutch language schools in Belgium. The sample was drawn based on location (province) and the education track being offered at the school. All third-grade pupils filled out a survey and received two copies of a similar survey for their parents, who were reminded by telephone if they did not respond spontaneously. Looking at gender and educational track, the sample closely mirrors the distribution in the population and therefore can be considered as representative for that age group in the Flemish region of Belgium.
In the Flemish high school system, there are three main educational tracks: general secondary education, technical, and vocational secondary education. Analyzing the response rates of the complete sample of adolescents and parents, we find a slight overrepresentation of general secondary education pupils and their parents and an underrepresentation of vocational secondary education pupils and their parents. To test Hypothesis 3, it is crucial that we can compare the effect of discussing politics with mother or father. Therefore, the analysis was limited to the 2,085 adolescents of whom we had information from a mother and a father. This amounts to 60% of all respondents. While it would be interesting to determine intergenerational transmission patterns in families where the adolescent only has contact with one parent, or where both parents have the same gender, this would require a different method of analysis and this will therefore be investigated in future analyses.
The Belgian Party System
The Belgian party system is one of the most fragmented party systems in Europe (Deschouwer, 2009). In the current federal parliament, 14 different parties have elected representatives. This is partly due to the fact that the traditional (Christian-Democrat, Liberal, and Socialist) parties split up into two regional parties during the 1960s and 1970s. This evolution has led to the formation of two segregated party systems: one for the Dutch-speaking part of Belgium and one for the French-speaking part of the country. Particularly in Flanders, the party system became strongly fragmented with the rise of a number of successful new parties. Next to the traditional Christian-Democrats, Liberals and Socialists, the Greens, the extreme-right Vlaams Belang, and the Flemish-Nationalist New Flemish Alliance all succeeded in capturing a large part of the electoral market.
The PCSS was conducted among Dutch language high school students. As a result, they could only choose their preferred Flemish party, as these are the only political parties competing in the Dutch language part of Belgium. In the survey, the respondents were presented with the following question: “If you could vote in an election for the Belgian parliament today, which party would you vote for?” The respondents could chose between the Christian-Democrats (CD&V), Greens (Groen), Flemish Nationalists (N-VA), Liberals (Open VLD), Socialists (Sp.a), extreme-right (Vlaams Belang), Libertarians (LDD), and extreme leftist socialists (PvdA), with an additional open response category for “other party.” This question taps voting intentions as this intention most clearly captures electoral preference in a fragmented party system as Belgium (Dassonneville, 2012). Due to their young age, the respondents in our sample have no real-life experience with voting. We did, however, investigate whether the party distribution among the adolescents reflects real-life electoral results (Table 1).
Adolescents’ Party Preferences and 2010 Election Results (Percentages).
Source. PCSS (2012), Ministry of the Interior (2010).
Note. PCSS = Parent−Child Socialization Study.
We compared the voting intentions of the full sample of adolescents (N = 3,426) with the most recent federal election results in Belgium (June 2010). Self-evidently, both results should not be directly compared as the election results are valid for the entire population. Nevertheless, it seems clear that adolescents follow the general pattern of the election results. Among adolescents too, Christian-Democrats and Nationalists clearly are the largest parties. What is equally important is that we find the same differences with regard to voting preferences as elections studies have found among the general population. Boys have higher scores for extreme-right and nationalist parties; girls are more likely to express a preference for the Greens and the Christian-Democrats. Pupils in vocational schools are also strongly overrepresented among the voters of the extreme-right party. Although we cannot be certain that the adolescents report a real party preference, we can be confident that the familiarity with the concept of political parties is sufficient to answer the question in a way that is not different from the way in which adults actually vote.
Dependent Variables: Correspondence in Voting Intentions
For the measurement of our dependent variable, we have constructed a number of possible combinations of the voting intentions of children, mothers, and fathers. We have measured voting intention in the same manner in the children’s, father’s, and mother’s surveys. Based on this question, we have constructed a series of dummy variables, which will function as dependent variables in our final model.
The recoded dummy variables are constructed as follows. PC stands for Party of the Child, PM for Party of the Mother, and PF for Party of the Father. Full correspondence (PC = PM = PF) is obtained when father’s, mother’s, and child’s voting intentions are equal. We obtain partial correspondence when the child’s voting intention matches its mother’s (PC = PM) or its father’s (PC = PF) voting intention—not both. The reference category is no correspondence, when there is no correspondence at all with mother or father.
Given the fact that our third hypothesis deals with gender-specific transmission patterns, in the final structural equation model, we will introduce a distinction between correspondence with the father (PC = PF) or correspondence with the mother (PC = PM).
Independent Variables
Family politicization
To measure the extent to which talking about politics within the family effects transmission of voting intentions, we constructed a latent variable “family politicization,” consisting of fathers’ and mothers’ report on the intensity of discussions about politics within the family (Jennings et al., 2009). We asked both parents, “How often do you talk about political and societal issues with your partner?” and “How often do you talk about political and societal issues within your family?” Both items were measured on a 4-point Likert-type scale. In the measurement model, these items were included in one latent construct, using a confirmatory factor analysis.
Political discussion with parents
We also asked the adolescents how often they talk about politics with their parents. We use two separate variables in our model, “Talking about politics with father” and “Talking about politics with mother.” The items are introduced separately to analyze the differences in influence between fathers and mothers on the voting intentions of their children.
Education level
In the Flemish educational system, there are three main types of high school education. As a preliminary analysis suggested that vocational, art, and technical training are closely related with regard to socioeconomic characteristics, these education forms were grouped. This results in a dummy variable, with general schooling coded “1” and technical, art, and vocational training coded “0.”
Party size
The odds that one will have the same party preference as one’s parents are automatically larger in dealing with large political parties. To control for this effect, we also included the size of the vote share within the PCSS sample.
Analysis
In the empirical part of this article, we first assess whether there is indeed a correspondence between voting intentions, before moving on with a more elaborated structural equation model. In Table 2, we present the frequencies of the main correspondence measures. It can be observed that 42.9% of all adolescents have the same voting intention as their mother, and we do not observe any difference in this regard between boys and girls. Agreement with the father is at a slightly lower level (38.0%), and here, we observe that boys tend to agree more often with the father than girls do, although it has to be noted that this difference is not significant. Generally, slightly more than half of all adolescents have the same voting intention as at least one parent (whether this is the mother or the father), and 29.8% agrees with both parents. Again, this percentage is slightly higher among boys than it is among girls.
Voting Intention Correspondence Between Parents and Children.
Source. PCSS (2012). The total number of respondents differs for every group, due to the fact that we analyze dyads of respondents and the N for fathers and mothers is not equal.
Note. PC = Party of the Child; PM = Party of the Mother; PF = Party of the Father.
These preliminary results clearly suggest that there is more political homogeneity occurring within families than one would expect given a purely random distribution. In a next step, we use SEM to explain the occurrence of this phenomenon. In the model, we use two dependent variables: agreement with the mother (PC = PM) and agreement with the father (PC = PF).
In a first analysis, we estimate a general model for all respondents. We present two models, a first one with and a second one without party size (Figure 1). In a second analysis, we split the respondents up according to their gender in a multigroup analysis with gender as a grouping variable.
Structural equation model for intergenerational transmission of voting intentions.
We use the intensity of political discussions within the family as a latent variable (oval form in the diagram). To optimize the fit of this measurement part of the model, we estimate the error covariances between the two measurements for the father and the mother (Byrne, 2011; Kline, 2011). In Table 3, we present all unstandardized and standardized coefficients for the measurement and structural parts of this model.
Full Structural Equation Model Predicting Correspondence with Mother and Father.
Note. Reported measures are obtained through a structural equation model analysis in Mplus 5.21 with WLSMV estimation (outcome variables are categorical). Results are unstandardized coefficients (B), standard errors (SE), standardized coefficients (β) and p-values. PC = Party of the Child; PM = Party of the Mother; PF = Party of the Father; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation.
p < .05. **p < .01. ***p < .001.
The most important conclusion we can draw from these results is that talking about politics within the family has a strong significant effect on correspondence in voting intention, for mother−child and father−child dyads. We find a positive significant effect of family politicization (talking about politics as reported by the parents) on the correspondence of voting intentions between father and child, and mother and child. Second, we find a positive significant effect of talking about politics with the father on the correspondence between fathers and children. Discussion with the mother, on the other hand, does not have the same effect on correspondence. There is no significant effect of talking about politics with one’s mother on correspondence in voting intentions. However, differences in frequency cannot be held responsible for these results (descriptive statistics in appendix).
In Model II of Table 4, the effect of party size is included. We can observe that this variable indeed has a positive effect, implying that correspondence is more likely if the adolescent prefers a larger party. Nevertheless, including this variable does not change the main effects, which means that the results we obtained in Model I are not due to party-size effects.
Multigroup Structural Equation Model Grouped by Gender.
Note. Reported measures are obtained through a multigroup structural equation model analysis in Mplus 5.21 with WLSMV estimation (outcome variables are categorical). Results are unstandardized coefficients (B), standard errors (SE), standardized coefficients (β), and p-values. PC = Party of the Child; PM = Party of the Mother; PF = Party of the Father; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation.
p < .10. *p < .05. **p < .01. ***p < .001.
In Table 4, we develop a multigroup structural equation model to assess whether effects are different for girls than for boys. As the sample is split up into half for the analysis of boys and girls, all coefficients in the split-file model become structurally less significant. Therefore, we also report the significance threshold of this model at the .10 level (†).
It can be observed that the effect of talking about politics with father or mother is very similar for boys and girls. So while there are some gender-specific effects with regard to the parents, we did not find a similar pattern with regard to the adolescents. When comparing the effect of talking about politics with the father, for instance, we find that this has the same positive effect on correspondence between father and son (.112, p = .056) as on the correspondence between father and daughter (.138, p = .060). The hypothesis that girls are more strongly influenced by their mothers, and boys are more strongly influenced by their fathers is therefore not supported by the analysis: for both groups, discussion with the father has the strongest effect on agreement with the father.
Discussion
We have investigated the intergenerational transmission of voting intentions between parents and adolescents in the Flemish region of Belgium. Before moving on to the discussion on the implications of these results, we briefly address a number of study limitations. A first limitation is the external validity. Although the Flemish region of Belgium is a theoretically relevant case of a fragmented multiparty system, we cannot assume that the mechanisms we found in this study can be generalized toward other multiparty systems. Particularly the results concerning the difference between discussion with one’s father and discussion with one’s mother should be investigated further and should be tested in other cultural contexts as well.
Second, there are a few data limitations. On the one hand, using data from 14- to 15-year-old adolescents provides us with very relevant information for young citizens in this stage of early political preference formation, but results should always be interpreted within this context as well. The adolescents in the sample are not politically experienced and often lack the experience of other political behavior opportunities, such as voting, discussions with other politically experienced peers, and so on. Therefore, the chances are higher that they will rely on the basic impulses they do receive, namely the experiences of direct family members. A second data limitation is that the paths we drew in our model are theoretically justified, but the analytical method and data used cannot suffice to clearly demonstrate that this is indeed a causal influence. A final data limitation is that these are purely survey results, and that they should be interpreted accordingly. We have, for instance, no firsthand substantive information on the content of political discussion and we do not know whether children consciously make the same decision as their parents. Gaining better knowledge of these precise mechanisms will require more in-depth observations within the family context.
Returning to the hypotheses, one of our main expectations was that we would only find a weak correspondence between parents and children. However, we could observe that 53.3 % of all adolescents had the same voting intention as at least one of their parents, with no significant differences between boys and girls. Contrary to Hypothesis 1, the conclusion has to be that the intergenerational homogeneity with regard to party preference is quite elevated, even in the highly fragmented Belgian political system. The fact that the mechanisms of intergenerational transmission do apply in Belgium indicates how strong these mechanisms still are in the current political setting. While some decades ago, it was argued that young generations would develop their political preferences as a way to develop their own identity, often in opposition to the political values supported by the generation of their parents, this is obviously not the case any longer. To a surprisingly high extent, the political preferences of adolescents are in line with the preferences of their parents. This implies that we can no longer assume that generational replacement will be the main driving force in social and political change. If younger generations apparently have the same political preferences as their parents—even in strongly fragmented political system—this rather suggests that all age groups are affected in the same manner by period effects.
In Belgium too, electoral volatility is quite elevated, and therefore it is quite likely that parents have switched between parties over the past years. Nevertheless, the adolescents in the PCSS sample still prefer the same party as their parents, especially in highly politicized families. This means that stable party identifications are not a necessary condition for intergenerational transmission to take place. No matter how fragmented the system is and how many options the adolescence have, adolescent children rely on the cues they receive from their parents to make their own party choice. While in two-party systems, party electorates often confound with strong social cleavages (e.g., on ethnic, religious, or class lines), thus blurring the precise causal mechanism for intergenerational transmission, this is to a lesser extent the case in multiparty systems, suggesting a more direct, distinctly political transmission mechanism. This suggests that in the Belgian case, the high level of correspondence cannot be explained solely by a mechanism of status inheritance as there is no strong relation between party preference and social class. For political parties, and the party system in general, this implies that the family should not be neglected as a sphere of political socialization and recruitment. In line with earlier research (Verba et al., 2005), the current analysis demonstrates clearly that the family does remain an important determinant for political attitudes and behaviors. While it can be observed that political parties increasingly target individual potential supporters, for example, by means of new electronic media, it should be kept in mind that within families, quite some political discussion does take place, with a direct impact on vote choice, as we have demonstrated.
Second, we have analyzed which factors could have an effect on this transmission. Using structural equation modeling, we constructed a latent variable for the measurement of family politicization and two endogenous dummy variables for correspondence with mother or father. The most important finding was the confirmation of previous research that talking about politics within the family context can indeed significantly increase the likelihood of adolescents sharing the same voting intention as (one of) their parents. However, there are some interesting differences between fathers and mothers. Talking about politics within the family and, more specifically, talking about politics with one’s father, has a strong effect on the correspondence between father and child. Talking about politics with one’s mother, however, did not seem to have the same significant effect on mother−child correspondence. This is surprising, particularly because we did not find any difference in the frequency of discussion with both parents. This means that it is not unreasonable to assume that fathers interact in a different manner about politics than mothers do. Additional analyses have indeed shown that the adolescents who discuss politics more with their father have a better perception of his voting intention, whereas this was not the case for mothers. Although this should be further investigated, these findings lead us to assume that children discuss more electoral politics with their father, while political discussions with their mother will be less focused on party politics, possibly explaining the difference in our models. This would be in line with other survey evidence, showing that while women and men in general might have equal levels of political participation, men still tend to specialize more often in electoral and party politics, and apparently this focus is being transmitted within the family context as well. For the overall composition of the electorate, this does imply that it is likely that gender differences in electoral behavior will continue across generations.
In Hypothesis 3, we proposed that political discussion with the parent of one’s own gender will have a stronger effect on correspondence of voting intentions with that parent. Based on our results, we should reject this hypothesis, as we did not find any clear differences for this process between girls and boys: for both groups, the effect of discussing with the father is more important.
We can safely conclude, therefore, that traditional political socialization mechanisms do still apply for new generations. Apparently, even among younger generations, traditional party politics might be less of a distant reality than one would assume, particularly when electoral politics is discussed within the family context. As was the case for earlier generations, contemporary adolescents take on the political preferences of their parents to a large extent. In the 1980s, it was routinely assumed that the rise of new political parties and cleavages, centered on postmaterialist issues, developed according to a process of generational replacement with, for example, the emerging Green parties catering mainly for a younger electorate. The occurrence of this process of generational replacement apparently can no longer be taken for granted and political parties have to adapt to a system with a high degree of stability between generations with the same family background.
On a more speculative note, we might expect that while generational replacement has become less important, demographic trends can have an impact on future power relations. Especially in the context of the United States, for instance, it has already been shown that the increasing demographic weight of ethnic minorities and Spanish speaking groups does have an impact on the balance of power between Democrats and Republicans. If future research would show that intergenerational transmission among ethnic minorities would be just as strong as among this general population sample, this would allow us to predict that in the future, too, ethnic minorities will be characterized by a distinct party preference, and thus it can be assumed that their impact on the political system as a whole will continue to increase. While it has been argued that traditional political cleavages no longer play an important role in the electoral process, our analysis offers a mechanism to explain the continued relevance of these cleavages in contemporary politics.
Footnotes
Appendix
Descriptive Statistics for “Talking About Politics With Mother” and “Talking About Politics With Father.”
| N | Minimum | Maximum | M | SD | Variance | Skewness | Kurtosis | |
|---|---|---|---|---|---|---|---|---|
| Talk politics with mother | 2,055 | 1.0 | 4.0 | 2.005 | 0.8143 | .663 | .386 | .108 |
| Talk politics with father | 2,002 | 1.0 | 4.0 | 2.166 | 0.8842 | .782 | .266 | .109 |
Source. Parent−Child Socialization Study (2012).
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.