Socialization and generational political trajectories: an age, period and cohort analysis of political participation in Britain

ABSTRACT The role of political socialization in explaining disengagement from specific modes of activism beyond voting remains largely unexplored, limited to date by available data and methods. While most previous studies have tended to propose explanations for disengagement linked to specific repertoires of political action, we propose a unified theory based on the different socialization experiences of subsequent generations. We test this theory using a new dataset of collated waves of the British Social Attitudes Survey and by applying age–period–cohort models for repeated cross-sectional data and generalized additive models to identify generational effects. We show that generational effects underlie the participatory decline across repertoires. Consistent with our expectations, the results reveal that the generation of “Thatcher’s Children” are much less likely to engage in a range of repertoires of political action than “Wilson/Callaghan’s Children”, who came of age in the more politicized 1960s and 1970s. Significantly, and in line with our theoretical expectations, the “Blair’s Babies” generation is the least politically engaged of all. We reflect on these findings and highlight the concerning implications of falling levels of activism for advanced democracies.


Introduction
The role of political socialization in explaining disengagement from specific modes of activism beyond voting remains largely unexplored, limited to date by available data and methods. While most previous studies have tended to propose explanations for disengagement linked to specific repertoires of political action (e.g. Franklin 2004), we propose a unified theory based on the contexts of socialization of subsequent generations. We test this theory using a new dataset of collated waves of the British Social Attitudes Survey (BSAS 1 ) and by applying age-period-cohort (APC) models for repeated cross-sectional data with a categorized generation variable as well as generalized additive models (GAMs) with year of birth as a continuous variable to identify generational effects through two different methods and add robustness to our results. We show that generational effects underlie the participatory decline across repertoires (Grasso 2016). Consistent with our theoretical expectations, the results show that the generation that came of political age during the 1980s -"Thatcher's Children"is less politically involved than the generation that came of age in the rather more politicized 1960s-1970s -"Wilson/Callaghan's Children"and this finding holds across different repertoires of political action. Moreover, the Millennial generation -"Blair's Babies"emerges as the least engaged of all. We reflect on these results in the conclusion and highlight the concerning implications of falling levels of activism for advanced democracies.

Participation and democracy
Citizen participation lies at the heart of a healthy democracy (Barber 1984) and yet there have been growing concerns about political disengagement in advanced democracies (Gray and Caul 2000;Pharr and Putnam 2000;Pattie and Johnston 2001;Hay 2007;Van Biezen, Mair, and Poguntke 2012;Grasso 2016). This decline in participation is said to leave democracy "hollowed out" (Mair 2006). Falling turnout levels are particularly alarming since they call into question the representative claims of democratic government (Pattie, Seyd, and Whiteley 2003;Pattie, Seyd, and Whiteley 2004).
Contrary to this negative picture, scholars studying unconventional or protest participation tend to present a more positive picture (Grasso 2018;Grasso and Guigni 2018 Forthcoming). These scholars argue that since the post-war period there has been a growth in engagement in public demonstrations and other modes of protest activism, such as signing petitions, boycotting products or companies, joining new social movement organizations, etc. (Inglehart 1977;Inglehart 1990;Norris 1999;Norris 2002;Inglehart and Welzel 2005;Marsh, O'Toole, and Jones 2007;Mayer 2010;Guigni and Grasso 2015). In the words of key authors, advanced democracies have now become "social movement societies" (Meyer and Tarrow 1998) marked by the "normalization of protest" (Van Aelst and Walgrave 2001).
Given these considerations, any study of political engagement in advanced democracies must analyse participation in both conventional and unconventional political actions (Giugni and Grasso 2017;. By shedding light on the variation that underlies these long-term trends in political participation, it is possible to assess the state of democracy in advanced societies (Stoker 2006;Hay 2007). In order to examine the question of the extent of inequalities in political participation and underlying trends of political disengagement, we must compare different political generations and use methods that aim to distinguish those differences due to age, time period and generational cohorts (Grasso 2016).

Inter-generational change and changing patterns of participation
Social change occurs through generational replacement and therefore the analysis of generational patterns of participation can allow us to establish the sources of any long-term trends (Grasso 2016). This is because as older generations die out, they are replaced by new generations exhibiting new patterns of behaviour (Franklin 2004). Moreover, the years of youth are "impressionable" so that the characteristics of an epoch tend to mould their imprint on those "coming of age" in that period (Mannheim 1928). While this idea was first famously articulated by Mannheim (1928), research on APC effects has since repeatedly shown that generations can differ markedly in their values and political behaviours (see for e.g. Tilley 2001;Tilley 2002;Grasso 2014;Neundorf and Niemi 2014;Tilley and Evans 2014;Grasso 2016;. Franklin (2004) notably examined the role of cohort effects in explaining turnout decline in Western Europe. This research showed that turnout decline results from younger generations participating at lower rates than older generations. However, the explanations offered by Franklin focused on factors strictly related to elections and turnout such as the closeness of the race. As such, there was no discussion in this seminal study of how political socialization might also explain disengagement from unconventional alongside conventional activism. To address this important gap in the literature, in this paper, we show that similar generational effects underlie participatory decline in both repertoires. While previous studies have tended to propose explanations for (dis-)engagement linked to specific repertoires of political action, we propose a unified theory which pertains to both conventional and unconventional domains based on the different political contexts of socialization.
We argue that politicization and the contestation of ideas are key elements determining the extent to which citizens are motivated to engage in politics and to participate politically (Hay 2004). The political contexts in which different political generations have come of age in Britain are markedly different in terms of the extent of political contestation of key ideas Temple et al. 2017). The context of socialization of "Wilson/Callaghan's Children"those socialized during the 1960s and 1970swas one of much greater politicization relative to those of the Second World War and immediate post-war generations. The Cold War was still raging and militant groups, particularly on the Left, multiplied across the Western world. Events such as the world-wide revolts of 1968 symbolized the progressive potential of the era and the generation coming of age in this period came to be known as "the protest generation" (Jennings 1987). In contrast, the late 1970s brought with them economic crisis and the rise of New Right politics. The 1980s were marked by a political shift towards neoliberal market economies in many Western democracies. The rise of the New Right most clearly signalled a rightward shift in opinion in the United States, United Kingdom and other Anglo-American democracies in the 1980s (Himmelstein 1990;Le Grande and Bartlett 1993). "Thatcher's Children"socialized during the 1980s and early 1990sthus came of age in an era which emphasized individualism and self-reliance over collective pursuits such as political participation for social change which had been so central previously (Clarke et al. 2004). This political epoch was marked by a shift away from collective wage bargaining and corporatist governance towards private ownership and control of many previously publicly owned enterprises and the decline in union membership (Gamble 1988). Individuals, rather than collectives, came to be seen as the key units of society (Gamble 1988).
Soon, other advanced democracies also followed with neoliberal reforms as social democratic parties distanced themselves from socialist ideals and moved towards the centre of the ideological spectrum across Europe (Mair 2006). While ideological contestation still existed in this phase of "normative neoliberalism" (Hay 2004) where Thatcher's Children were coming of age ), a consensus was increasingly emerging over the contemporary model of market-based economics married to limited social protections. This "normalised neoliberalism" phase (Hay 2004) emerged in the aftermath of the Cold War, as even leftist political opponents from social democratic parties previously favouring state ownership and nationalization of major industries as well as much wider universalized social welfare systems had come to internalize its market precepts as "the rules of the game" . As such, "Blair's Babies"the generation socialized in the years of New Labourgrew up in an epoch marked by the depoliticization of public debate in the wake of "TINA" ("there is no alternative"to the market) and the acceptance of neoliberal ideology by the Labour Party. Moreover, this process had led to the rise of technocratic imperatives and the suggestion that questions of politics have now become simply questions of good governmental management, devoid of normative meaning (Mair 2006). In this context, politicians tend increasingly to portray their decisions as above contestation in an effort to evoke consensus (Hay 2007). In the process, they avoid political arguments in favour of technocratic reason, often appealing to external economic constraints that they are powerless to shape in the process. This avoids the disagreement that has traditionally been at the heart of politics and presents elite decision-making as void of ideology (Mair 2006). Thus, depoliticization can be seen as a major factor alienating the public, and particularly younger generations, from politics in presenting the impression that there is no normative underpinning to elite decisionmaking and that all the political questions have already been solved (Hay 2007).
In this paper, we build on these insights and analyse whether those that came of age in distinct political eras have different proclivities for engagement as would be deduced from the period of their socialization. What are the differences in political engagement between generations that came of age during the highly politicized 1960s and 1970s (Wilson/Callaghan's Children), under the "normative neoliberalism" of the Thatcher and Major governments (Thatcher's Children), and during the "normalized neoliberalism" of New Labour (Blair's Babies)? We theorize that while less active than the "protest generation" that came of age during the Wilson and Callaghan years, that Thatcher's Children (also sometimes referred to as "Generation X" in popular culture) who came of age during the 1980s may still be more participatory than Blair's Babies (also sometimes referred to as the "Millennials"), as depoliticization peaked during the latter generation's formative years.
Scholars have tended to argue for a decline of conventional participation. On the other hand, they have argued for a rise in protest activism such as demonstrating, petition-signing and boycotting, seen also as replacing conventional activism Giugni 2016a, 2016b). We hypothesize that we should see similar generational patterns across modes of political action. We thus challenge the conventional substitution thesis, since we expect the same underlying processes of depoliticization to be at play across all activities. While Grasso (2016) provided some evidence using European survey data, here we focus on the specifics of the British context and draw on data that covers a greater number of important control variables from the BSAS as newly collated in a harmonized repeated cross-sectional dataset (Jennings et al. 2015), to add empirical breadth to this theorizing. While previous studies hypothesized that younger generations coming of age since the 1960-1970s will have a greater propensity to engage in unconventional activities, following the political generations account we hypothesize that: H 0 : Generational patterns for conventional and unconventional forms of political participation will be similar (contrary to the substitution thesis) H 1 : Thatcher's Children are less likely than previous generations to engage in all political activities H 2 : Blair's Babies are less likely than Thatcher's Children to engage in all political activities In the remainder of the paper, we first outline influential theories of political engagement. After this theoretical section, we discuss the data and methods employed, followed by the presentation of results. We conclude with a discussion of the implications of our findings. Brady (1999, 739) notes how participation has four key characteristics: action, citizens, politics and influence. One set of activities focuses around voting, elections and parties and is normally understood as "conventional"; this has been traditionally distinguished from "unconventional" acts or protest activism such as demonstrating, signing a petition, joining a boycott, etc. (Inglehart 1977; Barnes and Kaase 1979;Inglehart 1990;Inglehart and Welzel 2005;Dalton 2008). In addition to voting and campaign activities, conventional participation includes different types of community-based actions, contacting activities (elected legislators, officials, the media, etc.) and working with others to address issues/solve problems (Verba and Nie 1972). Unconventional activities encompass more "low risk" forms of protest such as signing a petition or joining a boycott or attending demonstrations and more costly or "high risk" forms such as occupations, illegal protests and even violence.

Determinants of political engagement
Several key theories from political science, political sociology and social movement studies allow us to draw out the key individual-level determinants of participation which might help to explain generational inequalities in political action. In other words, while "net" generational differences show differences in socialization, there are different potential narratives as to what lies behind them. In this way, political socialization is an additional factor that needs to be taken into account in analyses of political participation and not one that is contained in any of the below categories. Political generations are distinguished through their socialization experiences in particular political and historical contexts as famously articulated by Mannheim (1928). Much research has shown that generations coming of age in specific periods tend to exhibit characteristics that differentiate them from each other in both values and political behaviours (Tilley 2001;Tilley 2002;Grasso 2011;Bartels and Jackman 2014;Grasso 2014;Neundorf and Niemi 2014;Tilley and Evans 2014). In Britain, "Thatcher's Children" (Russell, Johnston, and Pattie 1992;Heath and Park 1997), the generation that came of age during Thatcher's and Major's time in office (between 1979 and 1997), has been the object of a number of studies.
However, other than socialization, there are other potentially confounding influences distinguishing generations that could explain variations in differing rates of participation. A number of factors have been identified as important in studies of political participation. A first group of factors explaining political participation with respect to various socio-demographic characteristics have been highlighted primarily in the biographical availability model (McAdam 1986). It is important to take socio-demographic factors into account when testing for socialization effects since younger individuals tend to be more likely to be in education, unmarried and free from obligations of careers and families and thus more "biographically available" to participate. Moreover, young people will tend to be less invested in their careers and thus have more time and the mental energy to get involved in and commit to politics; they are less likely to have entered marriage 2 and to have children: both things which decrease free time and the willingness to take risks (Wiltfang and McAdam 1991). This literature also suggests that people in full-time employment will be less likely to participate (McCarthy and Zald 1973;McAdam 1986). However, McAdam (1986) and Nepstad and Smith (1999) demonstrated that people in full-time (inflexible) employment were more likely to participate; this has also been found to vary by commitment (Saunders et al. 2012).
The civic voluntarism model also highlights socio-demographic variables, and particularly education and class, as very important given the argument that people who are better off and more resource-rich, more middle class, better educated, with higher incomes, are more likely to participate (Parry, Moyser, and Day 1992;Brady, Verba, and Scholzman 1995;Verba, Schlozman, and Brady 1995). However, the civic voluntarism model moves beyond this and conceptualizes resources widely, including time (Verba, Schlozman, and Brady 1995). It also links resources to political engagement at the psychological level (Schussman and Soule 2005). Feelings of involvement with the political system and of mobilization, normally through organizational membership, are emphasized since individuals must be elicited in their participation (Verba, Schlozman, and Brady 1995). And indeed, research has shown the importance of campaigning by political parties for mobilization (Seyd and Whiteley 1992;Whiteley, Seyd, and Richardson 1994;Pattie, Johnston, and Fieldhouse 1995) or asking people to protest (Schussman and Soule 2005). Other than political interest, consumption of political information such as newspaper readership matters (Verba, Schlozman, and Brady 1995). Ideology is also seen as important as more liberal individuals are normally seen as more likely to engage in political activities aimed at challenging the status quo (Hirsch 1990;Verba, Burns, and Schlozman 1997;Schussman and Soule 2005;Dalton 2008).
Organizational membership and embeddedness in social networks facilitating the recruitment to political action (McAdam 1986;Passy 2001;Passy and Giugni 2001) are also emphasized by the structural availability model. Since organizations forge ties between people, they multiply the possibilities that individuals will be mobilized to actionthrough the flow of information, through persuasion, etc. (Oberschall 1973;McAdam 1986;Paulsen 1991;McAdam and Paulsen 1993). Another way in which organizations can promote participation is by developing civic skills (Brady, Verba, and Scholzman 1995;Verba, Schlozman, and Brady 1995). Building on insights from the original rational choice model, Whiteley and Seyd's general incentives model (Seyd and Whiteley 1992;Whiteley, Seyd, and Richardson 1994;Whiteley and Seyd 1996;Whiteley and Seyd 1998) combines rational choice and social psychological models. They emphasize the role of expressive incentives (e.g. group attachment) as well as system benefits (e.g. a sense of duty) and collective benefits for participation. The social capital model also sees organizational memberships as important and emphasizes trust (Putnam 1993); the trust that is developed through voluntary activities spills over into other spheres of social, political and economic life, leading to more well-governed, affluent and politically engaged citizenries (Putnam 1993;Hall 1999;Whiteley 2000).
The cognitive mobilization model in particular emphasizes the role of dissatisfaction and grievances, in line with the social psychological model (Van Stekelenburg and Klandermans 2013), and suggests that individuals with higher levels of education, political interest and political information will be more likely to become dissatisfied with political outcomes and inclined to protest (Skocpol 1979;Norris 1999;Dalton 2008). As such grievances could also play an important role in explaining participation. Indeed, one of the oldest models of participation, developed with reference to revolutions, is the perceived equity-fairness or relative deprivation model employed by social historians (Hobsbawm 1962;Skocpol 1979) and social scientists in various forms such as the J-form and frustration-aggression hypotheses and strain or breakdown theories (Davies 1962;Runciman 1966;Gurr 1970). Here it is particularly emphasized that what drives individuals to engage in political activism is a sense of injustice, particularly in relation to some reference group.

Data and methods
In order to analyse APC effects on political engagement and explain generational differences, this paper uses a publicly available collation of the BSAS datasets (see Jennings et al. 2015). The repeated cross-sectional data are necessary to disentangle APC effects and as such we could not utilize data spanning shorter periods that include a larger battery of activities such as the Citizen's Audit (Pattie, Seyd, and Whiteley 2003;Pattie, Seyd, and Whiteley 2004). The analysis of political activism is based on eight indicators included in the survey between 1983 and 2012, covering three of the dimensions identified by seminal studies (Parry, Moyser, and Day 1992;Pattie, Johnston, and Fieldhouse 1995;Pattie, Seyd, and Whiteley 2003;Pattie, Seyd, and Whiteley 2004): contacting activities, group activities and protest activities. These variables will measure both traditional and newer (e.g. electronic or internet-based) forms of political activity. 3 Generational theories tend to argue that the context of one's socialization is the most important factor for understanding differences in values. To capture generational differences, we need to develop methods that allow us to address the potentially confounding influences of age and period effects. Age effects suggest that participation trajectories change with social ageing and indeed research has found that older people are more likely than younger people to be involved in conventional activities whereas younger people are more likely to engage in protest activism (Dalton 2008). Moreover, certain historical momentsor periodsare understood to stimulate more general participation. To identify generation effects, we need to control for both age and period, or year of survey, in our models. However, given the APC "identification problem", 4 we need to apply certain restrictions. A rich statistical literature has emerged over the years to "solve" the "identification problem". While there are strictly no statistical solutions to this problem, the use of theoretically informed choices can allow simplifying assumptions to estimate the effects of interest (Tilley 2002).
In this paper, we categorize generations based on the historical period of their formative years. Moreover, we identify the models differently through GAMs where we use year of birth as a continuous variable and use these to plot the identified, smoothed cohort effect. In this way, the GAMs add robustness to the results presented in the paper since they are identified differently from the other models. Here, we use year of birth estimated as smoothly changing whereas in the APC models, we use the theoretically informed categorization of generations based on the distinct historical periods of their formative years. Moreover, we use the results from the APC models to conduct post-estimation Wald tests for inter-generational differences.
Given our interest in generational effects, year of birth is the main independent variable. This ranges from 1910 to 1990 in our survey data. The key period of socialization will largely depend on the mechanism hypothesized (Inglehart 1990(Inglehart , 1977Bartels and Jackman 2014). Given our hypotheses rely on political awareness, we would expect that socialization should occur during the mid-teens to the mid-to late twenties (Jennings and Niemi 1974). We assign individuals to different political generations based on the historical phase in which they spent the majority of their formative years (see . As such, we define Thatcher's Children as those born between 1959 and 1976 and coming of age in the protracted period of Conservative rule between 1979 and 1996. This generation and the others analysed in this study are presented in Table 1. This method of categorizing generations has the advantage that it places emphasis on the historical period of a generation's socialization. The years of birth of the political generations are then derived from this information. We include the categorized political generations variable in the APC models 5 in order to (1) cross-check the robustness of the results from the GAMs and (2) use Wald tests to test for cohort differences. In the GAMs, we use the continuous year of birth variable to derive the smoothed cohort effects which means the models in the GAMs are identified differently to those from the regression models. Other than year of birth/cohort, we also control for age and period to identify the APC models. Age is a three-level factor coded: (1) 34 years and younger; (2) 35-59 years; (3) over 60 years. 6 We also control for year of survey as a linear term since the literature on participation extensively argues that we are witnessing a decline in conventional political participation on the one hand and a rise in unconventional participation on the other (see Norris 2002;Inglehart and Welzel 2005;Hay 2007).
Based on our earlier discussion of models of political participation, we control for a number of salient factors identified in the literature that might contribute to observed generational differences in rates of participation. By controlling for age and period as well as these effects, we can then analyse "net" generational differences in participationin other words, the residual, or "pure" generational effect. In order to do this, we therefore operationalize the insights from the theories of political engagement discussed above. We control for gender, employment status (being unemployed cf. Dunn et al. 2014), student status, marital status and having children. 7 We also include a control for ethnic group since minorities have also traditionally been a relatively deprived group in British society (Clarke  1910-1924 1925-1944 1945-1958 1959-1976 1977-1990 Total N  1986-2011. 5 Wilson/Callaghan's Children is used as the reference category in the APC models since we expect this to be the most participatory generation based on their period of socialization. 6 The first age group is used as the reference category in the APC models so other age groups are compared to the young. 7 The variable for children in the household was missing between 1993 and 2001 so instead we include a dummy for three or more individuals in the household. et al. 2004, 240). Moreover, we include income, social class (manual occupation) 8 and education level, private education and home ownership for resources. For organizational membership, we include an indicator for union/staff association membership. We also include party attachment (whether individuals identify with a party or not), newspaper readership and ideology in the form of an item measuring left-right values ("Government should redistribute from the better off to the less well off") and another for libertarian-authoritarian values ("Schools should teach children to obey authority"). Additionally, for grievances for participation, we use the item "Management will always try and get the better of employees" to capture this dimension. Moreover, we include institutional trust ("How much do you trust British governments of any party to put the needs of the country first."). We do not have an indicator of social trust available. Finally, we noted how the relative deprivation model emphasizes a sense of injustice, particularly in relation to some reference group, as spurring people to action and as such we include an item picking up on this dimension ("There is one law for the rich and one for the poor.").
Descriptive statistics for all the variables are presented in Table A1 in the Online Appendix.

Results
First, we describe levels of participation by political generation from Table 2. Note that in these descriptive statistics, generation effects are not isolated from the confounding influences of age and period and therefore differences in the fully identified APC models may turn out to be much largeror smaller. As shown in Table 2, the first and third most popular activities on the list are the two forms of protest activism: signing a petition (43%) and demonstrating (11%). While even the large proportion that have signed a petition still do not match up to the proportion of the population that turns out to vote, this is at least some evidence of the "normalization of protest" in Britain. It is particularly interesting that over 10% of the population on average have engaged in the relatively costly act of attending a protest. Overall, almost 50% of the British public have engaged in some form of protest activism (which includes signing a petition based on classic definitions in the literature). The overall second most popular activity was contacting an MP (17%). The other contacting activities (contacting a government department, the media, an influential person) are not as popular in comparison, with 22% of the British public having contacted for political reasons. Group activities (raising an issue with a group and forming a group) are by far the least popular, being only practised 8 The "Registrar-General's Social Classes" were introduced in 1913 and were renamed in 1990 as "Social Class based on Occupation". Note that a few individuals will change social class over time because of changes in the classification of occupations. In 1981, the 1980 Classification was used, in 1991, it was the 1990 Standard Occupational Classification (SOC90) and in 2001, this was replaced by SOC2000.
by 2-5% on average. This is hardly surprising since group activities are invariably the most costly and resource-intensive modes of political action. From the same table, we note that the Thatcher's Children and Blair's Babies stand out as being less participatory than older generations. With respect to contacting and group activities, Thatcher's Children are as disengaged as the earliest generation; Blair's Babies are even less participatory than this. Even with respect to protest which is supposed to be the preserve of more recent generations (Inglehart 1990), Thatcher's Children and Blair's Babies are less politically engaged than the older Wilson/Callaghan's Children in both conventional and unconventional activities, as suggested by H 0 . This particularly progressive "protest generation" coming of age in the 1960s and 1970s is the most participatory across the board, at least according to these initial descriptive trends. We now turn to see whether these initial descriptive results are confirmed in the APC models identified in two different ways: first through the GAMs with the continuous year of birth variable, and then in the APC models with the categorized generation variable.
First, we examine the results from the GAMs. In particular, we examine the plots of the smoothed cohort effect, as presented in Figures 1-4 in the Online Appendix, from the full model with all variables included. The curvilinear pattern across forms of action is striking and consistent. This provides support for our unified theory across participation acts, which moves beyond activity-specific explanations for generational inequalities in political action. The same patterns are found across both more conventional and unconventional domains. Across all four indicators of participation, the highest point is around the years of birth of the Post-WWII (1925-1944) and 1960s and 1970s Wilson/Callaghan's (1945-1958 generations. The curve slopes down on the other hand over the years of birth of Thatcher's Children, or the 1980s Generation (1959)(1960)(1961)(1962)(1963)(1964)(1965)(1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976) and Blair's Babies, i.e. the Millennials (1977Millennials ( -1990. This evidence provides support for the theoretical expectations of both H 1 and H 2 . While the curve over the years of birth of Thatcher's Children does not always fall back down to the levels of the Post-WWII generation, the curves clearly keep slowing down here (for those born during the period [1959][1960][1961][1962][1963][1964][1965][1966][1967][1968][1969][1970][1971][1972][1973][1974][1975][1976]. With respect to the political generation born between 1977 and 1990, i.e. Blair's Babies, they are even less participatory than the Post-WWII generation with respect to contacting activities ( Figure 1). Moreover, it looks as though group activism ( Figure 2) also records low levels of involvement amongst the youngest generations. It is particularly telling that even with protest (Figure 3), which is widely argued to be the activity of choice for younger generations, we see a downward sloping trend. This is replicated for the overall activism measure (Figure 4). The terms for the generation effects in the APC logistic regression models in Table 3 confirm the results from the GAMs very clearly. Across models and across forms of engagement, Thatcher's Children generation and the Millennial Blair Babies are less participatory than 1960s and 1970s Wilson/Callaghan's generation even in those modes of participationsuch as protest activismthat are meant to be much more commonly practised amongst the younger generations (Inglehart 1990). Across the models, and for the various types of participation, cohort differences generally remain robust to the inclusion of the various control and explanatory variables in the modelswith the exception of protest activism, further discussed belowsuggesting that the political context at the time of socialization was a very important influence in itself, and cannot generally be reduced to mediating factors such as the development of strong feelings of party attachment or the greater tendency to join associations. In order to make sense of where the greatest reductions in magnitude come from, we ran a series of stepwise models between the APC model and the full model (not shown here but available from the authors). These models showed that for contacting, the greatest reductions in the cohort coefficients for Thatcher's Children and Blair's Babies were achieved in the civic voluntarism model, even more than for the full model. This suggests that lower levels of political commitment (ideological attachments, newspaper readership), as well in terms of as party attachment and associational memberships/mobilization, have an important role to play in explaining these generational differences. If Table 3. APC models.
(1) Contacting (2) Contacting  individuals are not ideologically committed nor mobilized to political action through parties or other associations, it is unlikely that they will engage in contacting acts. Moving on to the results for group activity, as with contacting, the civic voluntarism model did particularly well in terms of contributing to the explanation for why Thatcher's Children and Blair's Babies are less politically active than Wilson/Callaghan's Children. Once again, political engagement (ideological commitment, newspaper readership), as well as party attachment and mobilization variables, in particular, appear important for explaining generational differences: younger generations have weaker left-wing beliefs and are less likely to be mobilized . As such, it is not surprising that they are less likely to form formal and informal groups struggling for political causes relative to the more politicized Wilson/Callaghan's Children.
Moving onto the results for protest activism, those models that include associational memberships or party attachment fully explain the lower protest proclivities of Thatcher's Children relative to Wilson/Callaghan's Children, as the generational coefficient ceases to be statistically significant. The civic voluntarism model also stands out as accounting for the greatest reduction in the generational effect for Blair's Babies, suggesting once again that leftist political beliefs, mobilization and party attachment are the key variables linking socialization in Britain to decreased political activism amongst these younger generations. In addition, both the cognitive mobilization grievances and relative deprivation injustice variables fare better than they did for the previous two forms of participation.
For overall activism, we also found the civic voluntarism model variables tended to explain away a greater proportion of the generational differences. Once again, the political engagement (ideological commitment), party attachment and mobilization variables look like they hold the most promise for providing part of the explanation for why Thatcher's Children and Blair's Babies are less participatory than Wilson/Callaghan's Children. They are less progressive ideologically, and less likely to hold strong political beliefs, they are more atomized and isolated and less likely to feel attached to parties and join associations. As such, they are less likely to have both the mental and structural resources to get involved in political activism Finally, to conclude our analysis, Table A2 in the Online Appendix presents the results of Wald tests checking for coefficient differences between the dummy variables included in the models (whereas in the models we can only test for differences with respect to the reference category). Blair's Babies are even less participatory than Thatcher's Children across modes of participation with the exception of group activity. Results presented in Table A2 also show that the Post-WWII generation is more participatory than both Thatcher's Children and Blair's Babies in both contacting and group activities, and also overall in the case of the latter. These results clearly confirm both H 1 and H 2 . The youngest Blair's Babies are even less involved in contacting activities than the Pre-WWII generation.

Conclusions
We argue that politicization and the contestation of ideas are key elements determining the extent to which citizens are motivated to engage in politics (Hay 2004). The political contexts in which the studied generations came of age are markedly different in terms of the political contestation of key ideas. The 1960s and 1970s Wilson/Callaghan generation came to be known as "the protest generation" (Jennings 1987) in many countries across the world, including in Europe. In contrast, while ideological contestation still existed in this phase of "normative neoliberalism", a consensus was increasingly emerging over the contemporary model of market-based economics married to limited social protections (Mair 2006). This "normalised neoliberalism" phase emerged in the aftermath of the Cold War, as even leftist political opponents came to accept it . On this basis, we theorized that while less active than the "protest generation" of Wilson/Callaghan's Children (coming of age during the 1960s and 1970s), the 1980s Thatcher's Children generation may still be more participatory than Blair's Babies or the Millennials, as depoliticization peaked during the formative years of the latter generation. We tested this theory by using a new dataset of collated waves of the BSAS and showed that the generation that came of political age during Thatcher's heyday in the 1980s was much less politically involved than prior generations across different repertoires; Blair's Babies generation emerged as the least engaged of all.
More broadly, our results show that we should be very concerned about declining levels of democratic participation in advanced democracies (Putnam 2000;Stoker 2006;Hay 2007;Grasso 2014;Giugni and Grasso 2018c Forthcoming). Even for protest activism, which is widely identified as the political repertoire of choice for younger generations in the theoretical literature, the increasing generational trend was halted and put into reverse. Against the expectations of prominent theories, inter-generational replacement is thus likely to lead to less, not more, participant publics in the future (Grasso 2016). The declining trend in civic engagement is not rescued through rising unconventional participation (Norris 2002;Inglehart and Welzel 2005). Based on these results, inter-generational replacement would lead to a deeper "Democratic Deficit" (Norris 2011) rather than a "Democratic Phoenix" (Norris 2002). While patterns of turnout among younger generations in the 2017 British General Election might have shown some signs of recovery, it is far too early to conclude whether this marks a lasting change or a temporary deviation from the long-term trajectory (Grasso 2013).
Overall, our results show that political activism is likely to continue to fall in the future unless a new politicized generation emerges. The generation coming of age now in this age of populism, political division, identity politics, nationalism and social media may well be distinct from the generations that came before it (Grasso 2014;Grasso and Bessant 2018). How these trends unfold will be enlightening, and only time will tell whether fallout from the economic crisis and the increasing fragmentation of party systems in advanced democracies will alter these trajectories Grasso 2018a, Lahusen andGrasso 2018). Social fragmentation and the decline of political loyalties may give rise to an era of concerted "identity" politics in which personal action may displace collective action (Bennett 2013).
Moreover, this research has shown that the civic voluntarism model is particularly useful for explaining generational differences in participation. It appears that ideological commitment, party attachment and mobilization, in addition to the socialization experiences in a given historical context, contribute to explaining the lower engagement levels of the youngest generations. Most importantly, this paper has shown that the "formative experiences" of new generations can affect the political behaviour trajectories and patterns of activism of entire nations. Our results, for both conventional and unconventional participation, raise fresh concerns about declining political involvement in advanced democracies. Globalization (Cambridge University Press, 2018, Forthcoming, with Marco Giugni)  Stephen Farrall is Professor of Criminology at the Law School at Sheffield University. Previously, he taught at the Universities of Oxford and Keele. He held visiting positions at ANU (Australia) and at the Dept. of Criminology at Keele University. His research has focused on the fear of crime (especially how best to measure it), why people stop offending, middle-class crimes, and crime histories. His other research interests include the long-term impact of Thatcherite social and economic policies on Crime, and cognitive interviewing to improve survey questions.
Emily Gray holds a B.A. in social policy from Goldsmiths College, University of London, and an MSc in criminology from Edinburgh University. She specialised in criminology for her PhD at Keele University. In 2013, she moved to the School of Law at the University of Sheffield to work on a major ESRC study exploring the legacy of Thatcherism and neo-conservative social policies on a range of attitudes and behaviours including with respect to crime.