Intergenerational Transmission and Support for EU Membership in the United Kingdom : The Case of Brexit

Euroscepticism is increasingly important to the shaping and understanding of contemporary European public opinion and politics. The origins of the trait, however, particularly the values that predispose individuals to view the European Union (EU) as a legitimate (or otherwise) political institution, remain poorly understood. Literature on political socialization identifies the family as a vital influence on the development of many social and political attitudes. This study explores the role of the family in the development of Euroscepticism by examining evidence of intergenerational transmission of hostility towards membership of the EU between parents and children in the United Kingdom during its ‘Brexit referendum’. The study shows that the attitudes of parents during one’s politically formative years can be an important factor in shaping support for EU membership. It also finds that this intergenerational transmission is different for mothers and fathers: while there is a greater likelihood of a child’s attitudes being affected by those of their father, if they are affected by their mother’s views they are more likely to eventually share their mother’s position on EU membership. This identifies the family as a key source of the values that shape support for European integration, potentially accelerating or opposing other social trends that have resulted in successive generations typically being more supportive of EU membership.


Introduction
Euroscepticism-that is, hostility towards European Union (EU) membership and/or integration-has become increasingly significant in shaping European public opinion and politics over the last decade.The financial and Euro crises of 2007/8, and 'refugee crisis' of 2012, in particular, have raised the salience of debates regarding the costs and benefits of EU membership for European citizens.Euroscepticism can also be seen as a response to the increasing impact of EU membership (i.e.further integration, globalization, increased migration, and technological development) on political issues traditionally shaped by citizens' core political and social values, such as understandings of national identity and state sovereignty, attitudes towards cultural change, and perceptions of the role of the state in relation to citizens and the economy (McLaren, 2002;Hooghe and Marks, 2005;Ford and Goodwin, 2014;Hobolt, 2016Hobolt, , 2018)).The growing popularity of political parties that blend hostility towards the EU with scepticism towards mass migration, cultural change and open market economies (such as Lega Nord, the Front Nationale or Fidesz), and the successful 'leave' campaign in the United Kingdom's (UK) referendum on EU membership in 2016 (Clarke et al., 2017;Hobolt, 2018), are but the latest indications of the rising importance of these issues and their relationship with Euroscepticism.
Understanding the origins of Euroscepticism is, therefore, increasingly salient for contextualizing contemporary public opinion and electoral politics in Europe.Previous research has shown that Euroscepticism reflects both short-term utilitarian assessments about the costs and benefits of EU integration, and more deeply-held affective values regarding the perceived legitimacy of the EU and its use of power, which originate in the early years of socialization (Boomgaarden et al., 2011;Down and Wilson, 2017).This means that key influences during socialization-such as the family-have considerable potential to shape the way individuals view and assess the legitimacy of the EU's power over their lives and community (Hyman, 1959;Jennings andNiemi, 1968, 1981;Plutzer, 2002;Jennings, Stoker and Bowers, 2009;Quintelier, Verhaegen and Hooghe, 2014).It is possible, therefore, for beliefs and values that shape Euroscepticism to be 'transmitted' from parents to their children during the socialization process, identifying the family as a potentially key source of support for (or opposition towards) EU integration.The scope of this potential is, however, poorly understood, as is the extent to which the transmission of values that underpin affective support for or opposition to the EU from parents to children can persist and still be apparent, or whether they can be overridden entirely, by shorter-term utilitarian assessments.
Research on how socialization (and socializing agents) affect citizens' views of the EU is limited.Previous studies have examined how the broader political climate and historic development of European integration affects Euroscepticism (Down andWilson, 2013, 2017;Fox and Pearce, 2018), and considered how familial influence is related to the development of 'European' identity (Quintelier, Verhaegen and Hooghe, 2014).The role of the family in shaping an individual's propensity to view the EU as a legitimate and valued institution, however, remains largely unstudied.This study addresses this deficiency and examines how intergenerational transmission-between parents and their children-is related to the development of Euroscepticism, using the opportunity presented by the United Kingdom's 'Brexit' referendum.Using the UK Household Longitudinal Survey (UKHLS), it examines the relationship between individuals' support for leaving the EU, and the attitudes expressed by their parents during the 'formative years' of their political socialization.It also considers how the transmission of Euroscepticism may vary between mothers and fathers, and how the impact of parents' attitudes could differ as a reflection of their political characteristics and different household roles.
The study finds evidence of intergenerational transmission, with some citizens being disproportionately likely to share the Euroscepticism expressed by their parents during their formative years and to support Brexit, despite being in an age group dominated by hostility towards the United Kingdom's withdrawal from the EU.Only a minority of respondents exhibit such a replication of 'hard Euroscepticism', however (i.e. the view that the United Kingdom should leave the EU [Lubbers and Scheepers, 2007]).This is partly due to the fact that, while the United Kingdom is a largely Eurosceptic country, only a minority of parents within our sample held such passionately Eurosceptic beliefs during their children's socialization to the extent that they were clearly transmitted and replicated in their offspring.We also find that transmission varies between parents: maternal influence seems more dependent on mothers' level of engagement with politics when compared with paternal influence (and a sizeable minority of women have little interest in politics in the United Kingdom).That said, transmission from mothers is potentially stronger, with passionately Eurosceptic and politically engaged mothers being more likely to instil Euroscepticism in their children than similar fathers.The study concludes, therefore, that while social change generally means younger generations are less Eurosceptic than older generations (Down andWilson, 2013, 2017;Fox and Pearce, 2018), a citizens' propensity to support EU membership is, to varying degrees, likely to be shaped by the attitudes and values of their parents.This means that the family can foster particular views towards the EU, even to the extent of acting as a buffer against the pro-EU tendencies promoted by social change.This suggests that deep-rooted experiences of previous political contexts are continuing to shape European public opinion regarding EU membership, national identity, the role of the state, mass migration, and globalization.

The Intergenerational Transmission of Political and Social Characteristics
The family is widely regarded as a key socializing agent, with parents, in particular, playing a pivotal role in providing their children with a framework for interpreting and traversing the social world.In political science the influence of parents is often said to be concentrated during the 'formative years' of adolescence, when children first begin to engage with and develop an awareness of politics (Jennings, Stoker and Bowers, 2009).The mechanism by which this influence is realized is described as social learning, whereby children receive cues about how to respond to political stimuli through 'observational learning, modelling, imitation and identification' (Jennings, Stoker and Bowers, 2009: p. 783).From a sociological perspective, it is argued that these explicit processes of political socialization are underpinned by more generalized, tacit and unconscious processes of primary domestic socialization in the family home (Bourdieu, 1977(Bourdieu, , 1986;;Moore, 2004).From this perspective, exposure to particular beliefs and values earlier in a person's life acts to frame their subsequent understanding and interpretation of political issues.

Factors Influencing Intergenerational Transmission
Research on social learning indicates that the likelihood of a particular characteristic being transmitted from parent to child is influenced by two factors: (1) the strength and consistency of cue giving, and (2) the nature of the trait.Traits that are of greater salience to the parent(s), (for example, an issue they feel passionate about) are more likely to be transmitted because there will be more consistent and clear cue giving (Jennings and Niemi, 1968;Tedin, 1974;Dinas, 2013;Quintelier, Verhaegen and Hooghe, 2014;Meeusen, 2014).According to this logic, a parent who is engaged with politics and a passionate advocate of membership of the EU, for example, is more likely to consistently and clearly express pro-EU sentiments than a parent who is not interested in politics and/or has no clear view on EU membership.Consequently, the offspring of the former are more likely to receive cues that influence the development of their attitudes than the offspring of the latter.
There is also typically a greater parent-child similarity in traits that are particularly value-laden or affectively oriented (such as ideological or moral beliefs), compared to those that are more utilitarian in nature (Hess and Torney, 1965;Jennings, Stoker and Bowers, 2009;Rico and Jennings, 2012;Dinas, 2013).For example, the greatest congruence between the parents and children studied by Jennings, Stoker and Bowers (2009) was for partisan political beliefs, which tend to hinge on citizens' moral and ideological values.Indeed, the successful transmission of other attitudes or beliefs depended on how strong a moral or affective component they had.This research suggests that the fact that moral or affective beliefs tend to be stable, unchanging, and frequently expressed, means that parents will provide their children with numerous consistent cues that increase the likelihood of transmission (Jennings, Stoker and Bowers, 2009).
Transmission can also vary between parents.Quintelier, Verhaegen and Hooghe (2014) found that the development of European identity was more strongly influenced by interaction with mothers than fathers.This could be due to gendered (rather than specifically parental) characteristics-such as men being more interested in politics than women, meaning that transmission is more likely to come from fathers than mothers.It could also reflect differences in mother-child and father-child relationships (Korupp, Ganzeboom and Van der Lippe, 2002;Flouri and Buchanan, 2004;Jaspers, Lubbers and De Vries, 2008;Jennings, Stoker and Bowers, 2009;Quintelier, Verhaegen and Hooghe, 2014;Quintelier, 2015).For example, Jennings and Langton (1969) found that children are more likely to be influenced by the values of the parent to whom they feel closest.Traditionally, mothers and fathers have occupied different household roles, with mothers more likely to be central to family politics and having closer emotional ties to their children, arguably resulting in a greater likelihood of transmission (Zuckerman et al., 2007;Coffe and Voorpostel, 2011).

The Intergenerational Transmission of Euroscepticism
Existing research indicates that there is a clear potential for a trait rooted in affective values and orientations to be expressed consistently by parents during the early years of their children's political socialization, to the extent that it can shape their children's subsequent political values, attitudes and behaviours in adulthood.Previous scholarship has highlighted the multidimensional nature of Euroscepticism-identifying it as a potential reflection of not only short-term assessments of whether EU membership is good for one's country or community, but also more deeply held, longer-term social and political values and priorities (Lindberg and Scheingold, 1970;Boomgaarden et al., 2011).This affective dimension of Euroscepticism links to individual beliefs about the legitimacy of the EU as a political institution and the political values it embodies: international cooperation, pooled sovereignty, international governance, and cross-border citizenship.It is also shaped by broader political and social values related to conceptions of national identity and state sovereignty, views of cultural identity, the impact of migration and social change, and the role of the state and government in the economy and daily life (McLaren, 2002;Boomgaarden et al., 2011;Ford and Goodwin, 2014;Hobolt, 2016;2018).
To the extent that Euroscepticism reflects affective sentiments regarding the EU and related socio-political values, it can be heavily influenced, therefore, by the political values and beliefs of parents during socialization.The importance of socialization in the development of Euroscepticism is underlined by studies of generational trends showing that people growing up in different social, economic, and political contexts (defined, for example, by different stages of European integration) have different likelihoods of being Eurosceptic that last throughout their lives (Down and Wilson, 2013;2017;Fox and Pearce, 2018).Similarly, Quintelier, Verhaegen and Hooghe (2014) found that parents had an important influence on their children's views on national and European identity, and that this could vary depending on the strength of identity held by mothers and fathers.There has been no study of how familial socialization is related to the development of the affective dimension of Euroscepticism, however.While holding or rejecting a European identity is likely to be related to an individual's views of EU membership, the two are quite different concepts.European identity refers to 'a specific form of social identity, expressed as a feeling of belonging to the European Union' (Quintelier, Verhaegen andHooghe, 2014: p. 1104) and focusses primarily on how an individual considers themselves to be associated with-and brings meaning to their relationship with-a wider society or group.Euroscepticism, on the other hand, is a political attitude that reflects an individual's support (or otherwise) for a particular use, distribution or manifestation of political power, in this case through their country's relationship with the EU and other European countries.The differences between the traits may also mean that processes of socialization and intergenerational transmission operate differently, or are influenced differently by mediating factors such as political engagement and the parent in question.

Research Design
This study uses the UKHLS, a household panel study of the United Kingdom population with data on a wide range of social, political, and economic traits, including Euroscepticism and respondents' preferences during the 2016 'Brexit referendum'.The survey includes a long running panel (with some respondents recruited in 1991), and recruits respondents' children when they turn 16.It also allows parent-child relationships to be identified.This means that expressions of Euroscepticism among respondents in recent waves can be matched with the Euroscepticism of their parents in earlier waves, to approximate the effect of respondents' parents' attitudes expressed during their formative years on their attitudes in later life.This panel design makes this the only study to date that enables the examination of the intergenerational transmission of traits related to the EU that takes account of how their expression in later life is related to attitudes respondents' were exposed to during their formative years.
The United Kingdom is not, of course, expected to be representative of public opinion throughout the EU, and even less so during a rare referendum on EU membership.The lack of appropriate cross-national data makes dependence on a single country instance unavoidable.The United Kingdom does provide a good case in which to look for evidence of intergenerational transmission, however, not least because of the unique opportunity to study the process using UKHLS panel data.Moreover, the United Kingdom is consistently identified as one of the most Eurosceptic Member States, meaning that if evidence of the intergenerational transmission of Euroscepticism can be found anywhere in the EU, it should be in the United Kingdom (Gabel, 1998;Nelsen, Guth and Fraser, 2001;Hooghe and Marks, 2005).There is also little reason to think that the specific context of the United Kingdom, or of its 2016 referendum, should alter the processes of intergenerational transmission or familial socialization to the extent that generalizations are impossible.Generalizing from the United Kingdom can also be justified on the basis that Euroscepticism is consistently shown to follow similar trends, and to be associated with similar characteristics and circumstances, throughout the EU (Gabel, 1998;Nelsen, Guth and Fraser, 2001;Hooghe and Marks, 2005).While a clear avenue for further research is to confirm this, this study assumes that the United Kingdom provides an appropriate example of an EU Member State in which to study the intergenerational transmission of Euroscepticism.
Questions about EU membership are infrequent in the UKHLS, meaning that only a few waves are available for this analysis and so a straight-forward operationalization of intergenerational transmission must be used.Respondents' preferences in the United Kingdom's referendum on EU membership in 2016 were matched with the responses of their parents to a series of questions about EU membership in 2006 i.e. during an earlier stage of their political socialization.It is not expected that parents' attitudes in 2006 will be wholly representative of their views of the EU, nor that there would be no change at all in the expression or intensity of that view over time.The fact that Euroscepticism is widely regarded to be a relatively stable trait amongst adults, however, suggests that using the 2006 data as an indication of whether respondents' were exposed to particularly Eurosceptic attitudes during their formative years is appropriate (Lubbers and Jaspers, 2010;Down andWilson, 2013, 2017;Fox and Pearce, 2018). 1 Moreover, if there is any instability in parents' expression of Euroscepticism, this actually strengthens the test of whether intergenerational transmission is occurring, because unstable attitudes are less likely to be transmitted (Jennings andNiemi, 1968, 1981).
We infer intergenerational transmission based on congruence between the attitudes of respondents in the 2016 referendum and those of their parents in the 2006 survey (when controlling for other factors).Specifically, respondents who supported the United Kingdom leaving the EU in 2016, and whose parents were extremely critical of the EU and supported leaving it in 2006, are taken to indicate the intergenerational transmission of hard Euroscepticism.This does not capture, therefore, all of the possible expressions of Euroscepticism that could reflect parental influence; rather, it captures the most 'extreme' manifestation of that influence in the form of replication of the parents' attitudes, and only with regard to support for Brexit.The constraints of the data available in the UKHLS, therefore, mean that the analysis is highly likely to produce a conservative estimate of the extent of intergenerational transmission.In addition, this method does not account for reverse causation i.e. the view of respondents' influencing the Euroscepticism of their parents in 2006.Instances of reverse causation are unlikely, however, because (by virtue of their age) the respondents were unlikely to have particularly developed views of EU membership compared to their parents, who are more likely to have more established and stable views that are less receptive to external influence (Dinas, 2013).Moreover, this analysis focusses on convergence of hard Euroscepticism between parents and children as evidence of intergenerational transmission; previous research has shown that the majority of external or socialization influences that may affect the attitudes of the children, such as completing higher education, are if anything likely to make them more supportive of the EU.By focussing on a rare form of transmission (i.e.exact convergence of attitudes), and one which has to overcome the opposing influence of secondary socialization agents and experiences, any reverse causation (which cannot be entirely ruled out) is highly unlikely to undermine the validity of this conservative analysis.Finally, this approach does not look to identify the potential causes of the parents' Euroscepticism that they may share with their children (such as financial precarity); rather, it focusses on establishing evidence that growing up in a Eurosceptic household increases the likelihood of the child being Eurosceptic in later life after controlling for such potential causes of Euroscepticism.

Sample
The sample is limited to respondents who: (a) answered the UKHLS question about the EU referendum in 2016; (b) were aged under thirty in 2006 (indicating that they were still in their 'politically formative' years); and (c) for whom there is data on either their mother's or father's Euroscepticism in 2006.The influence of parents' attitudes are analyzed separately (though the potential for interactions between them are explored) as limiting the sample to those for whom data from both their mothers and fathers is available in 2006 would render it too small to sustain reliable analyses.This results in an effective sample of 1,179 respondents.
The definition of the formative years of political socialization (during which respondents should be most receptive to the attitudes of their parents) of under the age of 30 is different from the more commonly used definition of between the ages of 15 and 25 in political science; a definition we argue is problematic.It is based on an assumption that political awareness and interest do not begin to form until age 15-a claim that is undermined by studies of children showing political awareness as early as the pre-teen years (Jennings and Niemi, 1968;Van Deth, Abendschon and Vollmar, 2011).Indeed, Bartels and Jackman (2014) found evidence of political learning in children as young as age 7. Sociological conceptualizations of familial socialization and habitus development (Bourdieu, 1977(Bourdieu, , 1986) also imply a process beginning much earlier than adolescence.Similarly, stipulating that formative political socialization stops at age 25 is equally problematic in light of growing evidence of the protraction of the 'youth' stage of the political lifecycle (Flanagan et al., 2012;Smets, 2016).Economic and social changes mean that it takes longer for young people to achieve the markers of 'adulthood' associated with becoming independent citizens (i.e. economic independence, completing full-time education, entering the labour market, buying a home and starting a family).Many of the experiences associated with such status do not occur until a later age, meaning the period during which young people seek guidance on how to respond to political stimuli from socializing agents-including parents-is likely to be longer.In accordance with this literature, a broader age range is used to define the formative years that takes account of both the earlier age at Downloaded from https://academic.oup.com/esr/advance-article-abstract/doi/10.1093/esr/jcz005/5359475 by Cardiff University user on 22 March 2019 which political learning begins, and the later age at which it becomes more limited.

Measures and Hypothesis
The central hypothesis is that respondents (Generation 2; 'G2') whose parents (Generation 1; 'G1') held strongly Eurosceptic attitudes during their formative years would be more likely to share that Euroscepticism in later life.As outlined above, however, this should be moderated by the salience of politics and EU membership to the parents during those formative years: the more politically interested the mother or father, the more salient politics and issues such as EU membership were likely to be during the formative years of the child, and so the greater the likelihood of intergenerational transmission.While the analysis allows for different effects from G1 mothers and fathers, no specific mother or father effects are hypothesized because (as shown above) the existing literature is divided as to what expectations of those effects might be.
The hypothesis was tested using logistic regression analysis.The dependent variable was derived from the question 'Should the UK remain a member of the European Union?' in the 2016 data.G2 respondents who supported 'leave' received a score of '1' and those who responded 'remain' or 'don't know' received a '0'.Three independent variables were used to operationalize social learning, based on measures of the parents' Euroscepticism and interest in politics.In the 2006 survey, G1 respondents were asked three questions about the United Kingdom's membership of the EU: (1) whether they thought EU membership was 'good', 'bad', or 'neither good nor bad'; (2) whether the United Kingdom had benefitted from membership; and (3) what the United Kingdom's long term policy towards membership should be.Latent structure analysis (see Supplementary Appendix) confirmed that all three variables were indicative of a common latent trait, and so responses to these questions were summarized so that parents who gave 'hard' Eurosceptic responses to each were identified as being 'Eurosceptic' and given a score of '1' (i.e. they said that EU membership was bad for the United Kingdom, that the United Kingdom had not benefitted, and that the United Kingdom should reduce the EU's power or leave altogether).Parents who did not give all 'hard' Eurosceptic responses were given a score of '0'.The salience of politics (and by association Euroscepticism) to the parents was measured by their interest in politics in 2006: those who were 'not at all interested' or 'not very interested' in politics were identified as having low engagement (and given a score of '0'), and those who were 'fairly' or 'very' interested were identified as having higher engagement (with a score of '1'). 2 An interaction variable between each parents' Euroscepticism and political interest was also created, to reflect the expectation that the transmission of Euroscepticism should be dependent upon the latter.A further interaction between mothers' and fathers' Euroscepticism was also included, to see if the effect of one parents' attitudes was affected in some way by the attitudes of the other.
Finally, control variables representing respondents' traits related to Euroscepticism and support for Brexit in the literature were also included, to isolate the effect of the parents' attitudes as far as possible, including: age, marital status and gender; education; interest in politics; occupational social class; religious affiliation; trade union membership; political party support; region; and whether or not respondents reported that they were struggling financially (Gabel, 1998;Nelsen, Guth and Fraser, 2001;Hooghe and Marks, 2005;Down and Wilson, 2013;Clarke, Goodwin and Whiteley, 2017;Curtice 2017;Fox and Pearce, 2018). 3

Results
While the majority of G2 respondents (64 per cent) supported 'remain' in the referendum (unsurprisingly as they were all under 40, the age group most likely to oppose Brexit), the descriptive statistics support the hypothesis that those with hard Eurosceptic parents were more likely to support leaving the EU.Of those whose mothers were Eurosceptic during their formative years, 45 per cent supported remaining in the EU while 49 per cent wanted to leave; of those whose mothers were not Eurosceptic, the figures were 64 per cent and 33 per cent respectively.Similarly, of those who had hard Eurosceptic fathers, 53 per cent supported remaining in the EU and 43 per cent supported leaving.For those whose fathers were not Eurosceptic, 62 per cent supported remain and 35 per cent supported Brexit.
There was also evidence that this relationship varied depending on the parents' political engagement, and that it differed for mothers and fathers.Figure 1 shows the proportion of respondents who supported 'remain' or 'leave' depending on the Euroscepticism and degree of political engagement of their mothers and fathers in 2006.It also shows the proportion of respondents in each category of mothers/fathers Euroscepticism by political engagement.Of those who had hard Eurosceptic mothers who were highly engaged with politics (8 per cent of our sample), just over half supported leaving the EU, compared with 47 per cent of those with hard Eurosceptic but not highly engaged mothers.The inverse of this group-those with not hard Eurosceptic but also highly engaged mothers (whom we would expect to be disproportionately likely to support 'remain')-also responded as expected, with 28 per cent supporting leaving the EU and 71 per cent supporting remaining.The likelihood of mothers' attitudes being transmitted to their children seems higher, therefore, if she was interested in politics.
For fathers, however, the effects of their attitudes towards the EU and interest in politics appear to work against each other.Of those with hard Eurosceptic and highly engaged fathers (10 per cent of the sample), 39 per cent supported Brexit, compared with 47 per cent of those with hard Eurosceptic but not highly engaged fathers.The pattern for fathers who were not hard Eurosceptics was essentially the same as for mothers.Previous research has shown that higher levels of political engagement are associated with greater support for EU membership (Gabel, 1998); this data suggests, therefore, that respondents' fathers were transmitting characteristics that made their offspring more likely to be pro-EU if they were highly engaged with politics, even if they themselves were actually strongly Eurosceptic.
Table 1 presents the results of eight logistic regression models predicting a 'leave' vote in the referendum.For the sake of brevity, only statistics relating to parents' Euroscepticism and model fit are reported-the full outputs are in the Supplementary Appendix.The first model included only control variables; the second included mothers' Euroscepticism and political interest without controls, while the third included the interaction between the two; the fourth and fifth models showed the same outputs for fathers; the sixth and seventh showed those outputs for both parents together; and the eighth model included the data for mothers and fathers (though the interaction for fathers was omitted as it was not significant) with all of the control variables.
Model II showed that both mothers' Euroscepticism and political interest had the anticipated effects: a respondent whose mother was a hard Eurosceptic during their formative years was roughly 17 percentage points more likely to support Brexit than one whose mother was more supportive of the EU.Respondents with a highly engaged mother were around 9 points less likely to support Brexit than those with a less engaged mother.The two also interacted: the positive, significant interaction coefficient in Model III shows that if a mother was highly engaged it increased the likelihood of her Euroscepticism being reproduced in her children.Respondents with a hard Eurosceptic, highly engaged mother had a 52 per cent probability of supporting 'leave', while those with a not-Eurosceptic, highly engaged mother had a 24 per cent probability of doing so.Model IV showed that having a hard Eurosceptic and/or highly engaged father had a similar effect: respondents with a hard Eurosceptic father were roughly 10 points more likely to support Brexit, and those with a highly engaged father were 12 points less likely to do so.Model V showed, however, that there was no interaction between these two traits as there was for mothers.
When the effects of mothers and fathers were considered together (Model VI), they remained virtually the same, suggesting that the impact of mothers' and fathers' attitudes were largely independent.This is supported by the inclusion of an interaction variable (not reported) between mothers' and fathers' Euroscepticism, which was far from significant and had a negligible effect.As shown in previous models, indications of hard Euroscepticism amongst mothers and fathers increased the chances of their offspring supporting leaving the EU, while high levels of political engagement amongst mothers and fathers increased the chances of their offspring opposing it.Model VI showed, however, that the effect of the mothers' Euroscepticism was roughly twice as strong as that of fathers', while the effect of fathers' political engagement was slightly stronger than that of mothers'.Respondents with a hard Eurosceptic mother, for example, were around 16 points more likely to support Brexit than those with a less Eurosceptic mother, while for fathers the difference was 7 per cent.Those with a highly engaged mother, on the other hand, were shown to be 7 points less likely to support Brexit, compared with a difference of 11 points for fathers.
Model VII included the interaction effects for both mothers and fathers, and found that the interaction for mothers was significant and implied the same relationship as shown in Model III.The interaction effect for fathers was non-significant.The final model (Model VIII), therefore, included only the interaction for the mothers, as well as all of the control variables.Including the controls reduced the magnitude of the coefficients for mothers and fathers somewhat, but they remained substantial and highly significant, showing that not only is the impact of mothers' and fathers' attitudes largely independent from one another, they are largely independent of other characteristics and processes affecting Euroscepticism as well.The only exception was mothers' Euroscepticism, which no longer had a significant direct effect; instead, the effect was entirely captured by the interaction between her Euroscepticism and political interest.
Figure 2 presents a more readily digestible summary of this final analysis, showing the predicted probability (with 95 per cent confidence intervals) of a respondent supporting Brexit depending on the Euroscepticism and political engagement of their parents, with controls accounted for.These data support the argument that parents' self-reported Euroscepticism and political engagement affected the likelihood of their children expressing Eurosceptic values when they were older.The data also confirm differences between mother-child and father-child relationships, with the effect of parental Euroscepticism being stronger for mothers than fathers.Respondents with a hard Eurosceptic and highly engaged father, for example, had around a 36 per cent probability of voting for Brexit; 6 percentage points higher than those with highly engaged but less Eurosceptic fathers.The difference for respondents with hard Eurosceptic, highly engaged mothers and not Eurosceptic, highly engaged mothers was almost five times larger (28 per cent).This indicates that for mothers, being highly engaged in politics increased the likelihood of her attitude towards the EU being reproduced in her children.Conversely, fathers' views of the EU (if they were transmitted at all) were likely to be transmitted independently of how interested in politics they  were: highly engaged fathers transmitted a tendency to be supportive of EU membership (even if they were Eurosceptics), while less engaged fathers transmitted a tendency to be Eurosceptic (even if they were pro-EU).

Discussion
Our analysis provides empirical support for the theory of intergenerational transmission of Euroscepticism between parents and children in the United Kingdom: parental views may have led to some respondents being more likely than their peers to either support or oppose Brexit in the 2016 referendum.Moreover, the effect of this transmission was largely independent of other demographic, socio-economic and political characteristics that lay at the heart of most explanations for Euroscepticism and the Brexit vote.The central hypothesis of this research and the social learning theory that underpins it-that having Eurosceptic parents during one's formative socialization would result in a greater propensity to share that Euroscepticism in later life, as long as that attitude was expressed consistently and clearly enough by parents-is supported.
Social learning theory would also expect, however, the likelihood of intergenerational transmission to be affected by the level of salience that European issues or values had for parents during the child's formative years; to the extent that the salience of EU membership is reflected in the salience of politics more broadly to the parents, this was found to be the case only for transmission from mothers.Fathers' political engagement had little impact on the chances of their Eurosceptic attitudes being transmitted; rather, it had an independent effect in which highly engaged fathers produced less Eurosceptic children regardless of their own Euroscepticism.As the regression models controlled for the respondents' level of political interest, this was not the result of the transmission of political engagement from father to child.Rather, it is likely to reflect the transmission of an additional trait related to political interest that is not captured in the model.Moreover, this means that the potential transmission of Euroscepticism from mothers-the likelihood of which could be increased by her political engagement-was far stronger (in terms of the likelihood of resulting in congruence) than that from fathers.This partially supports another expectation of social learning theory-that the likelihood of intergenerational transmission occurring would be different for mothers and fathers-though this did not extend to differences in the relationship between their Euroscepticism, interest in politics and the chances of transmission to their children.
The cause of such a considerable difference between transmissions from mothers and fathers is unclear.It does not reflect fathers typically being more politically engaged than mothers; while the data showed that fathers were typically more engaged, the analyses controlled for parents' level of political engagement.It is more likely to reflect either qualitative differences in the relationships between respondents and their mothers and fathers, therefore, or the different social, political, economic, and household roles typically played by mothers and fathers that lead children to experience their parents' political expressions in different ways.A further possibility is that our findings reflect different parenting roles, with respondents viewing a political issue (such as Euroscepticism) when discussed or expressed by their mother (typically adopting the more caring and home oriented role) differently from when it is discussed or expressed by their father (typically adopting the sterner and more socially or economically oriented role) (Berelson et al., 1954;Collins and Russell, 1991).This uncertainty cannot be resolved by the UKHLS survey, but poses an interesting question for future research, and suggests that the transmission of political and social characteristics from parents to children may be more heavily influenced by parent-child relationships, household roles or parenting roles than is often assumed.
While the sample in this research was not representative of the United Kingdom population, the distribution of mothers' and fathers' attitudes nonetheless allows for some reflection on the extent to which intergenerational transmission was likely to result in congruence of Eurosceptic attitudes between generations in the wider electorate.As reported in Figure 1, roughly one in five respondents had either a hard Eurosceptic mother or father, meaning that the vast majority of the sample were not exposed to Eurosceptic attitudes likely to make them more Eurosceptic during their formative years.Given that transmission from mothers is dependent on her also being highly politically engaged, this reduces the group further: only 8 per cent had hard Eurosceptic mothers who were also highly engaged.Overall, around a quarter of our sample had the potential to replicate the Eurosceptic attitudes of their parents as a result of intergenerational transmission, and of that quarter, 42 per cent supported leaving the EU in 2016.In total, around a tenth of the sample voted to leave the EU in 2016 and were significantly more likely to do so because of the attitudes of their parents during their formative years.
This figure does not account for the transmission of pro-EU attitudes from pro-EU parents to their children, and so the actual proportion of respondents exhibiting congruence with the attitudes of their parents will be higher.Nonetheless, this does show that intergenerational transmission leading to a convergence of attitudes between parents and children is only likely to occur for that minority of the electorate who have parents with clear and strongly held views on EU membership and/or whose mothers were sufficiently engaged with politics for her to 'transmit' them.The majority of European citizens are unlikely to have parents with such characteristics, and so intergenerational transmission leading to a convergence of attitudes is probably limited to a minority.
That said, there are other ways in which the attitudes and values of parents during their children's formative years can affect their political characteristics in later life, and which were not operationalized in this study.The constraints of UKHLS have limited this analysis to one particular form of intergenerational transmission, but other research has identified other manifestations of such transmission, such as the outright rejection of parents' views, or a partial or more limited transmission (Jennings, Stoker and Bowers, 2009;Rico and Jennings, 2012;Quintelier, Verhaegen and Hooghe, 2014;Quintelier, 2015).Our sample could have included respondents who, for example, were raised by extremely Eurosceptic parents and become very supportive of EU membership to rebel against them.It could also include respondents who were raised by Eurosceptic parents and were more Eurosceptic than their peers as a result, but who nonetheless decided to support remaining in the EU.Both such instances would not be detectable by the method employed here.While intergenerational transmission leading to convergence of Eurosceptic attitudes between parents and children is likely to be limited to a minority of citizens, therefore, it is likely that at least some of our sample were influenced by their parents' attitudes towards Europe in a manner not visible in this analysis.

Conclusion
Since the enlargements of the 2000s that led to substantial increases in migration between Member States, the global financial crisis of 2007 and subsequent Euro crisis, and the 'refugee crisis' of 2012, Euroscepticism has become increasingly important in shaping public opinion and politics throughout Europe.Not only are more citizens questioning whether the benefits of EU membership outweigh the costs, but the consequences of that membership are becoming more salient for other political concerns and values held by citizens, such as their conceptions of national identity and culture, views of state sovereignty, and beliefs about the role of the state.In order to understand and explain the origins of contemporary public opinion and political events (such as the victory of Lega Nord and the Five Star Movement in Italy, the improved electoral fortunes of France's National Front or the United Kingdom's Brexit vote), it is important to understand and explain the origins of contemporary expressions of Euroscepticism.
Our analysis is the first of its kind to use panel data to investigate the role of familial socialization and intergenerational transmission in the development of Euroscepticism-simultaneously accounting for potentially different effects from mothers and fathers.It provides compelling evidence that intergenerational transmission can lead to the reproduction of parents' attitudes amongst their children in later life.Fathers are the most likely to transmit attitudes to their children, because that transmission is not dependent upon his level of political engagement.Transmission from mothers, on the other hand, is rarer because it is shaped by how engaged she is with politics.Assuming that if a mother is highly engaged with politics and feels strongly Eurosceptic then she will demonstrate this clearly and consistently during her offspring's childhood, then there is a strong chance that her children will share her Eurosceptic attitude.This research suggests, therefore, that intergenerational transmission of some form is more likely between fathers and their children; but intergenerational transmission resulting in congruence is more likely between mothers and their children.
This article demonstrates the importance of family experiences in shaping the attitudes and values that frame an individual's support for European integration in later life.This suggests that the apparent rise of Euroscepticism in recent years, and the improved electoral fortunes of Eurosceptic political parties is, in part, a result of the socialization experiences of today's citizens: it is not solely the result of short-term influences and assessments about the consequences of EU membership.Whilst rising levels of educational attainment and the particular societal context experienced by younger generations predisposes them to be more pro-EU than their parents and grandparents (Down and Wilson, 2013;Fox and Pearce, 2018), this research suggests that some citizens have a greater propensity to oppose the EU, or to support Eurosceptic parties or candidates, than may otherwise be expected.The family can be positioned, therefore, not only as an important source of political values and attitudes that shape modern public opinion, but also as what Jaspers, Lubbers and De Vries (2008) describe as a 'buffer' against social change: the Downloaded from https://academic.oup.com/esr/advance-article-abstract/doi/10.1093/esr/jcz005/5359475 by Cardiff University user on 22 March 2019 experiences and values of parents, developed at an earlier stage of societal development, are shared with children and endow them with tendencies that are being challenged or eroded by social evolution.
This study also sheds some light on the intergenerational conflict that is frequently associated with Euroscepticism.Younger generations are typically more socially, culturally and economically liberal than their elders, and more likely to support EU membership (Hobolt, 2016(Hobolt, , 2018;;Curtice, 2017;Fox and Pearce, 2018).This is often identified as a source of intergenerational conflict, with claims in the United Kingdom that older generations 'stole the futures' of their children common after the 2016 referendum (Abbasi, 2016;Kottasova, 2016;Cosslet, 2016;Shuster, 2016).Such narratives imply that views are formed in isolation and do not account for the role of parents in shaping the attitudes and values of their children.While there is undoubtedly a clear age divide in attitudes towards the EU-a potential source of intergenerational conflictthe relationship between generations in the development of those attitudes cannot be ignored.
Finally, the constraints and limitations of this study highlight several areas for further research.First, opportunities to explore other manifestations of intergenerational transmission should be explored in order to map processes of intergenerational transmission in shaping the citizenry's support for EU membership (and by association support for related political attitudes and campaigns, such as populist political parties).This field would also benefit from a more detailed study of how and why transmission varies between mothers and fathers.Future studies could consider whether different social and economic roles (as indicated by employment status, for example), or relationships between parents and children, can account for this variation.Questions also remain regarding the role of secondary socializing agents, such as peers or schooling-both known to have a considerable impact on the development of political attitudes and values in ways that might complement or challenge traits developed through parental socialization (Jennings and Niemi, 1968;Jennings, Stoker and Bowers, 2009).Understanding the interactions between primary socialization processes (in the family home) and secondary socialization processes (at school and amongst peers), in the development of pro-European or Eurosceptic views would be of value to policy makers in considering potential policy interventions (such as through education policy) that could affect the development of Eurosceptic (and related) attitudes amongst future generations.

Notes
1 Previous research has shown that different political generations have differing, lasting propensities to support or oppose EU membership that persist over decades, meaning that there is at least some stability to the trait (Down and Wilson, 2013;2017;Fox and Pearce, 2018).Lubbers and Jaspers (2010) also found such stability in panel data, finding that no more than one in five Dutch voters changed from being pro-to anti-EU between 1990 and 2008.This is also apparent in the UKHLS: respondents were asked identical questions in 1999, 2002, and 2006 about whether they thought EU membership was bad for the United Kingdom, whether the United Kingdom benefitted from EU membership, and what the country's longterm policy towards the EU should be.The correlation between respondents believing that EU membership was bad for the United Kingdom between 1999 and 2006 was 0.50; for believing that the United Kingdom did not benefit from EU membership was 0.52, and for believing that the United Kingdom's long term policy should be to take back power or leave the EU was 0.43.These figures are comparable to those for other political traits well established in the literature as being stable and relatively unchanging once an individual passes into adulthood, such as political interest (Plutzer, 2002), for which the correlation between 1999 and 2006 was 0.65, and party identification (Jennings, Stoker, and Bowers, 2009), for which it was 0.46. 2 Different ways of measuring parents' Euroscepticism and political engagement were tested.For the former, a distinction was drawn between parents with 'hard' Eurosceptic responses to all three questions, to one or two questions, and who gave no such responses.For political interest, a distinction was drawn between those who were 'not at all' and 'not very' interested in politics, and between those who were 'fairly' or 'very' interested.
The analyses revealed that the only substantial differences were between respondents' whose parents gave 'hard' Eurosceptic responses to all three questions and those who did not, and between those whose parents were 'fairly' or 'very' engaged with politics and those who were not.For the sake of model parsimony, therefore, the simpler, dichotomous measures were used.3 The literature also identifies national identity and ethnicity as having important influences on Euroscepticism (McLaren, 2002;Lubbers and Scheepers, 2007).These were not included because there are no measures of national identity in the relevant UKHLS waves, and while there are measures of ethnicity, there were very few respondents from nonwhite backgrounds who met the criteria for inclusion in the sample set out above.Neither is expected, however, to affect the findings because they are unlikely to be related to the processes of intergenerational transmission or familial socialization.

Table 1 .
Logistic regression results, likelihood of voting 'Leave' in 2016