Anti-democratic youth? The influence of youth cohort size and quality of democracy on young people’s support for democracy

ABSTRACT This paper aims to investigate whether a country’s youth cohort size and quality of democracy, independently and jointly predict young people’s propensity to support democracy as a political system. We use pooled data from World Values Survey Waves 5–7, comprising 81 country-waves with 25,125 observations from 39 established and new democracies, in multilevel binary logistic regression analyses. The paper finds evidence that firstly, against conventional expectations, a large youth cohort exerts a positive influence on young people’s support for democracy as a political system. Secondly, the effect of youth cohort size depends on the quality of democracy of countries: young people growing as part of the youth cohorts in established democracies show stronger propensities to support democracy than their peers in new democracies. This has implications for both theory and empirical research.


Introduction
Young people's attitudes toward democracy remain one of the most contested issues within the scholarship on political attitudes.According to one strand of the literature, the posture of young people towards democratic politics within the established democracies of the West has been one of apathy, disinterest, indifference (Putnam, 2000; See similar discussions in Quintelier, 2007, p. 165) anti-democratic and even illiberal (Foa & Mounk, 2016;Mounk, 2018).They have even been infamously labelled as the 'harbingers of an incipient crisis of democracy' (Farthing, 2010, p. 181).If democracy would deconsolidate, its main causal agents, it is argued, would certainly include young anti-democrats (Foa & Mounk, 2016).In contrast, a second strand of the literature argues that contemporary young people are neither anti-democratic nor indifferent.Instead, they represent a new generation of citizens who are expanding the repertoire of political expressions, through their critical and authority-challenging attitudes, which in themselves, represent a deepening support for the democratic culture, rather than a threat to its consolidation (Dalton, 2009;Dalton & Welzel, 2014;Norris, 2002Norris, , 2011)).
In this paper, we explore an important but neglected theoretical/empirical space within the ongoing debate on young people's attitudes toward democracy.The study responds to the questions: (1) Does youth cohort size (YCS) affect young people's propensity to support democracy as a political system? (2) Is the effect of YCS on young people's support for democracy conditional on the quality of democracy (QoD) of the democratic countries in which they live?We explore these two relationshipsmain and moderated effectsacross both established and new democracies.Our interest in these two relationships is informed by three main blind spots within existing literature.
Firstly, a strand of the youth bulge literature suggests that the growing cohort size of young people within the adult population of democratic countries, poses a threat to the stability of democracy as a political system across the democratic world (Cincotta, 2009;Cincotta et al., 2011;Urdal, 2006;Weber, 2013).Despite this concern, traditional explanations proffered for young people's supposed dwindling and/or growing support for the democratic culture, have mainly focused on factors such as decline in social capital (Putnam, 2000), young people's inexperience with non-democratic systems (Mounk, 2018) and cultural and value change (Dalton & Welzel, 2014;Inglehart, 1990Inglehart, , 2016Inglehart, [1977]]).The potential effect of young people's demographic size on their commitment to the democratic culture, however, remains unexplored in the literature.Secondly, the current debate on the state of citizens' support for democracy as a political system has been mainly focused on established democracies of Europe and North America, and generally lacks a comparative focus with countries outside the western hemisphere (Claassen & Magalhães, 2022;Klingemann, 2014Klingemann, , 2018)).Exceptions to this geographical focus are usually single case studies outside the West (For example, Asomah & Dim, 2021;Zhai, 2022).Thirdly, comparative research focusing on the extent of support for democracy as a political system across both established and new democracies, specifically focused on the youth sub-population within the general polity, remains frequently overlooked.This study is an attempt to address these blind spots in the comparative politics literature on youth political attitudes.
Our analyses of pooled data for the youth sub-population, gathered from 39 democratic countries selected across both established and new democracies, using multilevel binary logistic regression reveal three main findings.First, the YCS of a country associates positively with the propensity of its youth to support democracy as a political system.Second, young people living in countries with high QoD show stronger propensity of individual support for democracy as a political system, than their peers living in new democracies.Third, the effect of YCS on young people's propensity to support democracy differs by the QoD of the countries in which they live.Accordingly, young people growing as part of youth cohorts in established democracies show much stronger propensities to support democracy as a political system, than their peers in new democracies.We discuss the implications of these results for both theory and empirical research in the latter sections of the paper.

Definition of key terms
Our understanding of support for democracy builds from the hierarchical model proposed by Dieter Fuchs and Edeltraud Roller, which was inspired by David Easton's ideas about democratic legitimacy (Easton, 1965(Easton, , 1975)).They distinguish between three ordered levels of support for democracy (Fuchs, 2007;Fuchs & Roller, 2019).The highest level of support for democracy is seen in citizens' commitment to democratic values (also called the democratic culture), expressed in terms of their loyalty to the concept of democracy (rule by the people), and associated values such as liberty and equality.This level of support is argued by Fuchs and Roller as essential to the consolidation of democracy as a political system.The second level of support relates to support for the democratic regime of one's own country.Here, support is based on an evaluation of the congruence between democratic culture/values, and the performance of the democratic regime of one's country.The more a regime adheres to democratic values/culture, the higher the support of citizens for that democratic regime.The lowest level of support for democracy within the model is seen in support for political actors within a democratic regime.This is expressed in terms of satisfaction with the performance of political actors.
In this paper, we define support for democracy at the highest level, as commitment to the democratic culture/values or the democratic political system.We define youth as persons aged between 15 and 29 years and use the term interchangeably with 'young people'.We also use the term youth cohort size to refer to the proportion of young people within the adult population of a country.We similarly use the term youth bulge to refer to a disproportionately large number of young people within the adult population of a country (Flückiger & Ludwig, 2018).The term quality of democracy is used in the paper to refer to the objective evaluation of the level of democratisation of a country.

Theoretical arguments
2.2.1.Main effect: youth cohort size and youth support for democracy Conventional explanations for the variations in citizens' support for democracy as a political system have been rooted in factors such as the political legacies of countries (Klingemann, 2014), economic (Magalhães, 2014;Pennings, 2017), social and cultural changes (Norris, 1999(Norris, , 2011)).The impact of the demographic structure of democratic societies has, however, rarely received mention in such discussions.Notwithstanding, young people's cohort size in particular, may be of special interest in explaining their attitudes towards democracy.For example, the relationship between their disproportionately large numbers, and protests, political violence, terrorism, and democratic stability remains a contested issue within the political violence and conflict studies literature.While some studies find evidence of a link between youth bulge and political/social disorder and various forms of violence (Cincotta, 2009;Flückiger & Ludwig, 2018;Korotayev et al., 2022;Sawyer et al., 2022;Urdal, 2006;Weber, 2013), others find only limited evidence of this association (Campante & Chor, 2014;Yair & Miodownik, 2016).Some studies also do not find any significant association between youth bulge and political/civil disorder (Ang et al., 2014;Collier & Hoeffler, 2004;Costello et al., 2015;Urdal & Hoelscher, 2012).
The arguments in support of youth bulge as a reliable predictor of political instability, protests and various forms of violence, suggest that young people are the main protagonists of many protests and politically destabilising acts (Huntington, 2011(Huntington, [1996, p. 117], p. 117]; Moller, 1968), and are often overrepresented by numbers in such acts (Cincotta et al., 2011;Goldstone, 2002).Decades of persistently high fertility rates or stalled fertility decline, due among others, to the influence of religious and cultural values which promote large family sizes, are argued to create the challenge of large, educated youth cohorts with limited economic opportunities in countries which also tend to be relatively fragile politically (Goldstone, 2002;Goldstone et al., 2014;Weber, 2019).Countries with YCS of 35% or more are, for instance, three times more likely to experience conflicts, than others with demographic structures similar to developed societies, and are also less successful at achieving democratic consolidation (Urdal, 2006).
Much of the grievances and opportunities which motivate young people's increased propensity to engage in acts of political and civil disorder (Collier & Hoeffler, 2004), are related to the consequences of being a member of a large cohort within the population.For any group of people, their social and economic fortunes are evidenced to decrease with their growing cohort size (Easterlin, 1987;Macunovich & Easterlin, 2008).For instance, large YCS has been demonstrated to reduce labour market opportunities for young people due to competition within the group for the same limited employment prospects (Apolte & Gerling, 2018;Bricker & Foley, 2013;Flückiger & Ludwig, 2018;Korenman & Neumark, 2000).The same phenomenon reduces wages among even the fortunate to be employed young people (Brunello, 2010).The consequences of being a member of a youth bulge on the social and economic stability of young people, therefore, becomes an existential challenge to their smooth transition into social adulthood, marked by indicators such as completion of school, employment, personal accommodation, marriage, and financial support for significant others (Eguavoen, 2010;Ozerim, 2019).This can evidently be a potential source of grievance which can lead to their involvement in politically destabilising acts.
Also important to the present discussion are the socio-biological experiences of youthfulness.Young adults (mostly men) in particular show greater approval for antidemocratic, aggressive and extremist attitudes (Weber, 2013).This is in part, attributable to the socio-biological experience of adolescence during which, as part of transition into adulthood, young people detach themselves from the cultural values and belief systems of their parents, and by extension the older generation, associate more with peers, and become more vulnerable to antidemocratic and extremist ideas (Weber, 2019).Importantly, these are general socio-biological tendencies of young people, regardless of the countries in which they live.Notwithstanding, if their cohort size is large, it increases the propensity to harbour antidemocratic attitudes for both the young men themselves, and also the rest of the population (Weber, 2013).This is because when young people live in communities with youth bulge, they interact more, and are influenced more by their peers, than when they live in communities with a disproportionately large adult population (Hart et al., 2004).
The above skewness in social influence in favour of the younger generation due to their large cohort size implies that they are more likely to be politicised and influenced in their attitudes and beliefs, through a large pool of peer agents within the population, than through the more conventional structures of political socialisation, such as the family and the school system.Past studies on the effect of peers on the political socialisation of young people show this position to be strongly tenable (Dostie-Goulet, 2009;Nkansah & Papp, 2022;Pilkington & Pollock, 2015;Quintelier, 2015).
The foregoing arguments suggest that; firstly, the phenomenon of large YCS holds the potential of reducing the social and economic fortunes of members of the cohort.This can in turn, predispose them to grievances towards the state/government.Secondly, large YCS increases the sheer numbers of peers with similar socio-biological tendencies to detach from familial and traditional values and be attached to extremist and antidemocratic tendencies.Lastly, large YCS shifts the balance of social influence and social learning from the traditional 'adult-youth' to a more 'youth-youth' experience, although the youth are conventionally known to be less knowledgeable politically and civically.In theory, therefore, even if a country has a strong democratic culture, the above arguments suggests that the combination of these three conditionssocioeconomic, socio-biological, and social influencewithin a growing youth population can be a source of concern for the democratic system.Such challenges can cause young people to have doubts about the ability of the democratic system to address their needs as a cohort.We therefore hypothesise based on the above argument that: H1: The larger the youth cohort size of a country, the less supportive the youth are of democracy as a political system.

Moderation argument: youth cohort size, quality of democracy and youth support for democracy
Dalton and Welzel identify two main types of citizen attitudes towards democracy and its institutions: allegiant and assertive dispositions (Dalton & Welzel, 2014).The allegiant position demands persistent loyalty and support for the democratic political system from the citizenry, as the desired disposition of the citizenry, needed to preserve the democratic culture (Almond & Verba, 2015[1963]).Against this normativity, the attitude of young people towards the democratic political system has been found to be especially problematic.For instance, the much-cited empirical studies of Roberto Stefan Foa and Yascha Mounk show that the younger generation of established democracies of Western Europe and North America, hold a growing disaffection towards democracy as a political system (Foa & Mounk, 2016;Mounk, 2018).In the words of Yascha Mounk, 'One possible explanation for why a lot of young people have grown disenchanted with democracy is that they have little conception of what it would mean to live in a different political system' (Mounk, 2018, p. 122).Mounk suggests that young people within established democracies appreciate democracy less, because they are oblivious of the ills of life under undemocratic regimes.An important implication of this argument is that compared to their peers in new democracies, which have relatively recent transitions from nondemocratic cultures, young people in established democracies are likely to show less support for democracy as a political system.
The alternative evaluation of young people's attitudes toward the democratic culture is offered within the assertive citizens framework (Dalton & Welzel, 2014).This line of literature rejects the notion of growing opposition to democracy among young people in established democracies.It contends that young people's perceived antidemocratic attitudes, rather represent a shift from attitudes that gave open support to democracy as a patriotic duty, to alternative sets of critical norms and values, which demand greater accountability from politicians and the democratic institutions they represent (Dalton, 2009;Norris, 1999;Stoker et al., 2017).Any perceived critical disposition towards democracy is consequently a manifestation of the growing public expectation with democracy, not a rejection of the system (Norris, 2011).The assertive tradition sees the contemporary younger generation as a type of critical citizens (Dalton, 2009;Norris, 1999Norris, , 2011)), or dissatisfied democrats (Dahlberg et al., 2015;Klingemann, 2014): they indeed believe in and support democracy as a political system.They are, however, dissatisfied with the empirical outcomes of the regime in terms of performance.Put differently, critical citizens/dissatisfied democrats successfully decouple their faith in democracy as a political system, from their disappointments with systems outputs such as, for example, economic crisis, and do not support democracy as a political system less, due to such system performance challenges (Claassen & Magalhães, 2022;Huang et al., 2008;Klingemann, 2018).
Also importantly, the literature suggests that these patterns of citizen behaviour (critical citizens/dissatisfied democrats) are more pronounced in established democracies than new democracies.For instance, Klingemann's empirical analysis of the relationship between level of democracy and support for democracy showed that respondents in countries with higher levels of democracy or higher quality of democracy exhibited the strongest support for democracy as a political system (Klingemann, 2014).He reasoned, however, that causality could be implied in both directions: 'Countries might have a high level of democracy because support for democracy is widespread since long.Alternatively, support for democracy might have become an integral part of a country's political culture because it is democratic since long ' (p. 123).A recent multilevel study of European democracies also suggests that young people in older and established democracies are more likely to engage politically (and by implication actively support the democratic systems) than peers in new democracies (Kitanova, 2020).Importantly, this strong belief of citizens in the democratic culture is argued to be cultivated over time through political socialisation and social learning (Bratton et al., 2004;Fuchs & Roller, 2019;Klingemann, 1999).
In contrast to the well-developed institutions and structures of established democracies for political socialisation and social learning, new democracies have much shorter democratic legacies, are still reconstructing their political identities, and hence, struggle with the transition from undemocratic norms and orientations developed over years, into the new democratic political culture (Dahlberg et al., 2015;Nový & Katrňák, 2015).Comparatively then, we argue that young people in new democracies will show lower levels of support for democracy than their peers in established democracies.
Additionally, the phenomenon of large youth cohort which confronts new democracies implies that, the challenges associated with youth bulge discussed earlier, and the relatively low quality of democracy of new democracies, can combine to create a stronger disaffection with democracy as a political system among the youth in new democracies.By contrast, the strong institutions of political socialisation and social learning of democratic values and culture within established democracies should significantly mitigate the negative effects of a youth bulge on young people's propensity to support democracy as a political system in established democracies.We accordingly expect the effect of YCS on young people's support for democracy as a political system, to depend on the QoD of the countries in which they live.We expect that members of youth cohorts in new democracies will show relatively weaker propensities to support democracy, compared to their peers in established democracies.H2: The negative effect of youth cohort size on young people's support for democracy is stronger in the case of individuals living in countries with low quality of democracy.

Research design
The study utilised a large -N design with 39 democratic countries, purposively sampled from WVS Waves 5-7 (i.e. 2005Waves 5-7 (i.e. -2009Waves 5-7 (i.e. , 2010Waves 5-7 (i.e. -2014Waves 5-7 (i.e. , 2017Waves 5-7 (i.e. -2020)). 1 These countries together generated a total of 81 country-waves pooled across all three WVS waves (Wave 1 = 36 country-waves; Wave 2 = 27 country-waves; Wave 3 = 18 country-waves).Countries covered in the respective WVS Waves were categorised as democratic or otherwise based on the Economist Intelligence Unit (EIU)'s 0-10 democratic index score scale 2 .Our list included countries with a minimum mean EIU score of 6.01 per each WVS Wave.These corresponded to EIU's categorisation of flawed democracies (6.01-8.0)and full democracies (8.01-10).We excluded hybrid regimes (4.01-6.0)due to their peculiar profile of being mixture regimes.An equally viable measure of democracy we could have used was the V-DEM liberal democracy index, which is argued to be an improvement over alternatives such as the Polity IV and the Freedom House indices (Boese, 2019;Elff & Ziaja, 2018).Our preliminary selection of cases based on the EIU democracy index and V-DEM liberal democracy index, however, generated important differences in the mix of countries which qualified as democracies.The V-DEM index, for instance, excluded countries such as India, Pakistan, Malaysia, Indonesia, Mexico, and a few others from some of the WVS Waves as not being liberal democracies, as they scored less than 0.6 on the V-DEM liberal democracy scale.
The exclusion of the above-named countries reduced the sample size considerably as the eventual set of countries were overwhelmingly established democracies of the West.It also importantly reduced the much-needed variation in YCS among countries required for our study design, given that the countries listed above are among those with the largest YCS within the WVS database.At the same minimum threshold of 0.6, which equates to 6.01 on the EIU democracy index, however, the EIU index qualified the same as democratic countries.This afforded the study two main benefits.Firstly, consistent with our research ambition, which seeks to know whether young people's support for democracy as a political system varies across different democratic societies based on the YCS and QoD of countries, the EIU democracy index generated a wider pool (a relatively larger sample of countries) of both old and new democratic societies.Secondly, the EIU's set of countries offered considerable variations in YCS by the inclusion of many new democracies with large youth cohorts.Given the nature of the research design, the above variations offered by the EIU democracy index in the choice of countries represented a major improvement in our design.The final list included 25,125 young respondents aged between 15 and 29 years sampled across 81-country-waves from 39 countries.The Online Appendix 1 shows the list of democratic countries selected based on the above index.

Data collection
We collected data for our analyses from four main sources.The WVS provided us with pooled individual level data on political attitudes and sociodemographic and socioeconomic statuses of young people aged 15-29 years, across Waves 5-7.Our preference for the WVS was due to its national representativeness, geographic reach and longitudinal nature (Inglehart et al., 2014), all of which made it the best suited for our kind of study.We also retrieved data from the World Population Database of the UN Population Division 3 for the YCS of countries, measured at 5-year interval for 2005-2020.The World Bank Database 4 provided data on youth literacy rate, and GDP per capita of all countries included in the analyses, whiles the EIU democracy index scores provided data on the QoD of countries.

Dependent variable
Our dependent variable (DV) for the study was Political System: having a democratic system.The WVS questionnaire asked respondents whether in their opinion, democracy as a political system was (1) 'very good, (2) fairly good, (3) fairly bad or (4) very bad way of governing this country?'. 5 We chose this variable as a suitable proxy for estimating young people's support for democracy as a political system.Although many studies combine different questions to generate a composite index to measure support for democracy (Claassen & Magalhães, 2022;Magalhães, 2014;Mattes & Bratton, 2007), this particular variable is argued to be an adequate measure for a broader set of commitments to democratic values, and a good proxy for assessing the highest level of citizen support for democracy (Fuchs & Roller, 2019, p. 230;Klingemann, 2018, pp. 205, 209).
We followed the cited scholars in treating the variable as a binary in analyses, despite its ordered nature.Thus, we also grouped the fairly bad and very bad responses as one category of responses.We treated this group of responses as our reference category named 'bad' and recoded as zero (0), and the fairly good and very good responses as the target category named 'good' and recoded as one (1).A total of 86.87 per cent of all respondents across the selected countries thought democracy was good (i.e.fairly good or very good), while the remaining 13.13 per cent thought it was bad (i.e.very bad or fairly bad).We explain the rationale for our choice of binary categorisation of the DV in the endnotes 6 and elaborate in the Online Appendix 10 of the paper.Table 1 shows the proportion of youth who thought democracy was a bad or good political system in the 39 countries in our sample.

Explanatory variables
Our two main explanatory variables were the youth cohort size and the quality of democracy of countries.We operationalised the YCS of countries as the proportion of persons aged 15-29 years within the adult population of a country (15 years and above).The 15-29 years range adequately captured our group of both theoretical and empirical interest, compared with other estimates which have much lower upper cut-off points, such as the 15-24 years (See Korotayev et al., 2022;Nkansah, 2022;Weber, 2019).Since the World Population Report database estimates within 5-year intervals, we calculated and applied the YCS estimates for countries to the respective WVS Waves which fell within the population years.Thus, for the Wave 5, covering the period 2005-2009, we calculated the YCS for 2005-2010 and applied to each country.The mean YCS across countries was 31.06, with a standard deviation of 7.67.Japan recorded the smallest YCS of 16.12, while Ghana had the largest YCS of 45.4 over the period.We operationalised quality of democracy as the mean EIU score for each WVS Wave for each country.The overall mean EIU score was 7.36 with a standard deviation of 0.92.Across all three Waves, we observed the lowest EIU score of 6.14 and the highest of 9.88, for Ghana and Sweden respectively.Figures 1 and 2 show the distribution of YCS and EIU scores across the sample.
Given that one of our hypotheses focuses on estimating the interaction effect of youth cohort size and quality of democracy on our DV, we also show the visual relationship between the two variables across the three Waves of the WVS.Overall, the scatterplot shows that the two demonstrate an inverse relationship.Countries with relatively high QoD tend to have relatively small YCS.On the other hand, countries with relatively low QoD tend to have relatively large youth cohorts.

Control variables
We limited our choice of control variables to those which had been reported in extant literature as having the strongest effects on support for democracy.In terms of age, support   for democracy is reported to be weakest among the youngest polity, due among others, to their lack of experience with alternatives to democracy, and limited cognitive competences to assimilate political information (Bratton et al., 2004;Foa & Mounk, 2016;Fuchs & Roller, 2019;Mounk, 2018).Also, while the conventional position in the literature shows females as less supportive of democracy compared to males, particularly in places with high gender inequality and discrimination (Maseland & van Hoorn, 2011;Walker & Kehoe, 2013), recent evidence suggests that gender difference in support for democracy becomes insignificant when the socio-political arrangements are favourable and supportive of women (Konte & Klasen, 2016).
The dissatisfied democrats literature also positively associates support for democracy with rising levels of education and economic resources of the public (Dalton, 2004;Norris, 2011).Counter evidence from African democracies, however, shows that there is no significant difference in citizen's levels of support for democracy based on economic status, with poorer people, being even more likely to be dissatisfied democrats (Doorenspleet, 2012).Education, notwithstanding, associates positively with support for democracy (Asomah & Dim, 2021;Bratton et al., 2004;Evans & Rose, 2012).We chose the variable employment status in the WVS database as proxy for individual employment, and recoded it into three clear categories: unemployed, student and employed, from the original eight categories.Closely linked to the individual economic situation, we also controlled for household economic situation with the variable, satisfaction with financial situation of household.Our education variable was categorised as low, middle and upper.These correspond to null/primary, secondary and tertiary levels of educational attainment respectively.
Also, building from the strand of the literature which associates group membership with the development of strong democratic and civic values (Putnam, 2000;Verba et al., 1995), and young people's current inclination towards joining groups more cause-oriented and micropolitical, as part of their political expressions (Benedicto, 2013;Soler-i-Martí, 2015), we controlled for young people's membership in environmental and humanitarian groups.We created a combined variable out of two variables within the WVS database which measure commitments towards environmental and humanitarian values: membership of environmental group and membership of humanitarian group, to try and estimate the strongest possible relationship with support for democracy.
At the contextual level, we accounted for the level of economic development and educational attainment of countries.Both factors are long argued in extant literature as strong predictors of support for democracy (Dalton & Welzel, 2014;Lipset, 1959;Norris, 1999).We accordingly controlled for GDP per capita at Purchasing Power Parity, log-transformed to reduce residual variance.We also controlled for literacy rate, youth total (per cent of people ages 15-24), as proxy for educational attainment of our target group.As with all of our contextual variables, we calculated their mean values per WVS Wave, and assigned them to each wave period.Details of the nature, distribution and coding of all variables are contained in the Online Appendix to the study.

Results
We utilised random-intercept multilevel logistic regression with country clustered standard errors to analyse the data.This was due to the binary nature of our DV and the hierarchical nature of the data, which had individuals nested within countries.Likelihood ratio tests between standard logistic regression model and the multilevel model showed our choice of multilevel modelling as better fit for the data.We also report an Intra Class Correlation of 0.081, or 8.1 per cent of variations at the country level.We note, therefore, as a limitation, that not much of the variation in the data is explained by the country-level variables.Support for democracy may thus be influenced more by individual level measures.We present the results of two models testing the main and interaction effects of our explanatory variables on the dependent variable in Table 2.We control for all the other predictors of support for democracy discussed in the methodological section of the paper.We also present the predicted probability plot for the interaction of the two explanatory variables on our DV in Figure 4.
Model I tested our first hypothesis (H1) of a direct negative relationship between YCS and young people's support for democracy.The evidence did not support our hypothesis.Against expectations, YCS was found to exert a significant positive effect on young people's individual propensities to support democracy as a political system.The effect of YCS was relatively marginal (β = 0.093) but not surprising, however, given an ICC of 8.1 per cent.The analysis also showed a significant relationship between QoD and support for democracy among young people.Higher levels of democratisation across countries corresponded with greater likelihood of support for democracy as a political system among the youth.The effect of QoD (β = 0.403) was stronger than that of YCS.Only higher education at the individual level showed stronger effect (β = 0.829) than a country's level of democratisation, on young people's propensity to support democracy as a political system in the Model.Quite surprisingly, our measures of contextual economic influence -GDP per capitaand social influenceyouth literacy ratewere both not significant in their effects on young people's likelihood of support for the democratic culture.As independent contextual factors, therefore, the Model shows that a country's QoD exerts the most pronounced effect on the opinions of its youth about democracy, followed by the demographic size of the youth.Other significant predictors included a student or employed status, both of which showed positive associations with support for democracy, and membership of causeoriented groups, which interestingly rendered young people less likely to show commitment to the democratic culture.
Since our interest was also in the moderating effect of QoD on YCS, we tested our proposition in H2, that the impact of YCS on young people's support for democracy will be conditional upon the level of democratisation of the countries in which the young people live.We supposed that the negative effect of YCS will be stronger for young people in countries with low QoD.We tested our assumption in Model II.Table 2 shows the outcome of the analysis.Here, the interaction of the two variables shows a significant positive effect on young people's likelihood to support democracy (β = 0.029).In other words, the positive effect of YCS on young people's propensity to support democracy differs by the quality of democracy of countries in which they live.The effect is stronger in the case of young people who live in established democracies, than their peers residing in new democracies.This result goes contrary to our expectation in H2, given that we had, based on the theorisation for H1, supposed a negative effect of YCS.The predicted probability plot for the interaction effects of the two variables is seen in Figure 4. Note: Entries are the logistic regression coefficients and standard errors clustered at country level (in parenthesis) for multilevel binary logistic Models.Models were estimated using individual level data on young people's support for democracy as a political system, drawn from WVS Wave 5 −7.Reference category: (1) = Male; (2) = lower level; (3) = Unemployed; (4) Not a member * * * p ≤ 0.001, * * p ≤ 0.01, * p ≤ 0.05 The plot shows that regardless of the YCS of a country, young people's probability to support democracy as a political system increases with increasing levels of democratisation within a country.For a given YCS, the probability of support for democracy is highest in established democracies (QoD ≥ 8).This corresponds to countries such as Sweden, Germany, Finland, and Norway, in our sample.Such countries consistently showed predicted probabilities of support for democracy ≥ 0.8, regardless of a growing YCS.As the quality of democratisation declines, however, the probability of support for democracy over increasing levels of YCS shows a decrease.Across countries, a onepoint decrease in QoD for any measure of YCS, decreases the probability of support for democracy considerably.For instance, for four democratic societies with equal YCS of 26.10, the plot shows predicted probabilities of support for democracy from the most democratic (QoD = 9.1) to the least democratic (QoD = 6.1) to be equal to 0.92, 0.875, 0.84 and 0.80 respectively.Put differently, young people in new democracies (QoD < 8.0) are more likely to show relatively lower overall positive levels of support for democracy, compared to their peers in established democracies with similar YCS.The plot also shows that the probability of support for democracy over varying YCS is lowest in the most fragile new democracies (QoD < 7.0).These in our sample correspond to countries such as Ghana, Argentina, The Philippines, Serbia, and Colombia.
Due to our initial observations in Figure 3, and the nature of the distribution of YCS across countries, we followed the recommendations of Hainmueller et al. (2019) and used an alternative semi-parametric modelling approach to further assess the marginal effect and the underlying distribution of the variables.We estimated a non-linear marginal effect using the kernel estimation option of the 'interflex' package in Stata.Figure 5 shows the scatter plot of the conditional effect of QoD on support for democracy by YCS, while Figure 6 reveals the uneven and non-linear effect of YCS over the different values of QoD.In other words, the conditional effect of YCS by QoD, although overall positive, may be more nuanced than our conventional margins plot in Figure 4 reveals.We  reflect on the implications of these results in our discussion.We also briefly discuss the limitations of the non-linear kernel estimation approach for our kind of data (i.e.binary dependent variables and hierarchical datasets) in the Online Appendix 11.

Robustness checks
As a robustness check, we respecified our Models with country fixed effects for both our baseline Model and the Model with interaction, to see if the fixed properties of our cases would impact the net effect of both QoD and YCS on young people's support for democracy as a political system.We dropped all variables which were not significant in the main Model, to better estimate the net effect of our independent variables, controlling for fixed effects of the cases.We also replaced youth literacy rate with youth unemployment rate and dropped GDP, and significant variables such as membership of cause-oriented groups, and non-significant variables such as satisfaction with household income from the Models, to test the sensitivity of the effects of the two explanatory variables and their interactions on the support for democracy.In all respecified Models, both explanatory variables and their interactions showed significant positive effects on young people's support for democracy.Their effects were even stronger in some cases than in the main Models.We present the results in the Online Appendix 12.

Discussion/Conclusion
Are young people living in democratic countries with large YCS less likely to approve of democracy as a good political system for their nation, than their peers living in democratic societies with small YCS?Evidence from the youth bulge literature gives ample reasons to presume this to be the case, particularly as the presence of a youth bulge within a democratic country, has been consistently shown in extant studies to be a reliable predictor of democratic instability and deconsolidation (Cincotta, 2009;Cincotta et al., 2011;Urdal, 2006;Weber, 2013).Guided by the existing evidence, our study supposed that large YCS will be inimical to the faith of young people in the democratic culture.We assumed that the often-argued triune conditions created by a youth bulge: socioeconomic deprivation of the members of the cohort (Apolte & Gerling, 2018;Brunello, 2010;Korenman & Neumark, 2000), socio-biological experiences of adolescence and youthfulness which predispose young people to idealistic, antidemocratic and extremist ideas (Cincotta et al., 2011;Weber, 2013Weber, , 2019)), and the shift in social influence from the preferred 'adult-youth' social learning environment to the more peer dominant and driven 'youth-youth' socialisation experiences due to their sheer numbers (Hart et al., 2004;Nkansah & Papp, 2022), would combine to negatively influence the judgement of young people about democracy.
Quite surprisingly, the result of this study shows that contrary to conventional wisdom, large YCS exerts a significant positive influence over young people's likelihood to approve democracy as a political system (β = 0.093; OR = 1.098; 95% CI: 1.064 −1.132).We found this finding to be robust under different model specifications.What, however, can explain this generally unexpected attitude of young people growing as part of a youth bulge?We think that the conceptualisation of young people as critical citizens/dissatisfied democrats, as theorised earlier within this paper, provides useful insights in this regard (Dalton, 2009;Dalton & Welzel, 2014;Norris, 2011).We believe that we can look to the two main characteristics of such democrats for possible answers.The first is the fact that they are staunch adherents of the democratic culture, believe in and support democracy as their preferred political system, despite their critical attitudes towards the empirical outcomes of the regime.The second is the fact that the belief of dissatisfied democrats in the democratic culture is impervious to socioeconomic changes, such that they do not approve of democracy as a political system significantly less in the face of unfavourable socioeconomic challenges.
We think that as critical citizens/dissatisfied democrats, young people across both established and new democracies, are able to conceptually decouple their belief in the democratic system and values, from the empirical performance of their democratic governments.Therefore, despite the socioeconomic deprivation and associated challenges with social transition into adulthood, posed by their large cohorts size, due among others to poor macro socioeconomic outcomes, they tend to remain steadfast in their core beliefs in the idea of democracy as a good and preferred political system.The considerable pool of peer agents large YCS creates in this case, becomes a positive social force which reinforces their fundamental beliefs and convictions about democracy.Peer interactions have been reported in past studies to play positive roles in gathering and sharing political information, galvanising political interest and even recruiting peers into political activities (Bergh et al., 2021;Klofstad, 2011;Quintelier, 2015).We believe this to be also true in the case of young people's support for democracy as a political system.Large YCS seems to amplify this positive effect on young people's core commitments to the democratic culture, although they may in principle, hold strong reservations against the socioeconomic outcomes of democratic governance, as it relates to their peculiar situations as youth.
Importantly, nonetheless, the results of the study show that net effect of YCS on young people's support for democracy is dependent on the quality of democratisation of the countries in which they live.As a main effect of QoD (see Table 2), the analyses reveal that the higher the level of democratisation of a country, the greater the likelihood of support for the democratic culture among the youth (β = 0.403; OR = 1.497; 95% CI: 1.23-1.83).This finding for the youth sub-population of democratic societies is consistent with Klingemann's findings for the general population which showed that, respondents in countries with higher levels of democratisation demonstrate the strongest support for democracy (Klingemann, 2014).The net effect of QoD on youth support for democracy was only superseded by individual level educational attainment at the upper level (university education) in our Model.This was against the backdrop that our macroeconomic and social variables, GDP per capita and youth literacy rates, were both not significant in their effects on young people's belief in the democratic system.
Of the four contextual variables initially envisaged to shape the individual attitudes of young people towards democracy as a political system, i.e. political (QoD), demographic (YCS), economic (GDP per capita) and social (education), the political context exerted the most pronounced impact on young people.This brings into focus the current debate on what drives support for democracy or otherwise among democratic citizens (Claassen & Magalhães, 2022;Klingemann, 1999Klingemann, , 2018;;Magalhães, 2014;Pennings, 2017).Our evidence suggests that support for the democratic culture may indeed be impervious to the economic context of countries, but much more dependent on the strength and depth of a country's political legacy.The findings of this paper show that young people's commitment to democratic values/culture tends to be stronger when they live in societies with strong democratic institutions which defend these values, and strong structures for political socialisation and social learning of such values.In fact, we find this assertion to be quite plausible when we consider the interaction of a country's QoD with its share of youth cohort, on young people's faith in democracy as a political system.
The predicted probability plot in Figure 4 shows that the positive effect of YCS on young people's support for democracy differs by the level of democratisation of the countries in which they live.For any given YCS beyond 16.1 percent of the adult population, the plot shows that young people within established democracies (QoD ≥ 8), show much higher probability to support democracy, than their peers in new democracies (QoD < 8) with equal YCS.In other words, while both the demographic size of the youth population and the strength of a nation's democracy jointly positively predict support for democracy, the effect is stronger in the case of individuals who live in countries with stronger institutions for socialising citizens into the democratic culture.A possible explanation for the observed interaction effect could be the phenomenon of young critical citizens/dissatisfied democrats that potentially becomes amplified when a large share of young people find themselves in highly democratised environments.In such contexts, their large numbers can combine effectively with strong institutions of political socialisation; this creates a condition where peer influence and peer learning become effective avenues for inculcating democratic values within the cohort.Admittedly, however, the lack of highly democratised countries (i.e.QoD ≥ 8) with YCS ≥30 in our sample, as seen in Figure 3, lends caution to such a conclusion.Most probably, if such cases were present, the observed interaction effects may not be as reported in Figure 4.The marginal effect (kernel estimation) plot in Figure 6, partly lends credence to this likelihood.As we can see, the conditional effect of YCS on support for democracy across the different levels of QoD may be more nuanced than what the conventional margins plot in Figure 4 reveals.Given that this is a data limitation, we are cautious not to overstate the interaction effect in this case, despite their overall joint positive effect on support for democracy.
As our concluding reflections, we note a few important theoretical and empirical implications of the foregoing.Firstly, the often-suggested threat of youth bulge to democratic stability presents an interesting case for re-examination in the light of the findings of this paper (For example, Cincotta, 2009;Cincotta et al., 2011;Weber, 2013).Our analyses show that even in the face of a youth bulge, young people in democratic societies at all levels of democratisationboth new and established democraciesremain more likely to support the idea of democracy as their preferred political system.Youth bulge related acts of democratic instability may, therefore, be motivated more by the poor empirical outcomes of a democratic government or political actors (dissatisfaction with democracy), than a disillusion with democracy itself as a political system.We think that it is important for this conceptual distinction to be made about young people's disposition towards democracy.Secondly, the present concerns about the commitment of young people in established democracies towards the democratic culture, and fears about a potential deconsolidation may need re-examination from a more causal and predictive analytical point of view, rather than the present descriptive statistical approaches to the analysis of the issue (See, Foa & Mounk, 2016;Fuchs & Roller, 2019;Mounk, 2018).Even for new democracies, the evidence in this paper shows that belief in democracy as a political system among the youth shows a strong positive trend.Also importantly, this study has demonstrated that in addition to the usual predictors of support for democracy, contextual factors such as YCS and QoD, and the interaction of both, increase young people's propensity to support democracy as a political system.This is a novel finding in the youth political attitudes literature and opens up new theoretical and empirical orifices of enquiry on young people's evaluation of democracy, which account for especially YCS; a factor previously unexplored in extant research.
We note, notwithstanding, that these findings are not without limitations.For instance, our choice of a single variable DV, which we deem adequate, over the composite index DV with multiple variables used in other studies to estimate support for democracy, may have impacted outcomes significantly.We also note that our choice of multilevel binary logistic models over the potentially better suited non-proportional odds ordered models might have been potentially limiting.Also, our use of the EIU democracy index over other indices such as the Freedom House index, Polity IV index and V-DEM may have also affected our selection of cases and eventual outcomes.Further studies can, therefore, address these limitations with the alternative methods and indices mentioned above.

Notes
Attila Bartha Associate Professor, Corvinus University of Budapest & Senior Research Fellow, Centre for Social Sciences.His research interests include economic policy analysis, governance, social policy, and public policy analysis.

Figure 1 .
Figure 1.Youth cohort size in the sample.

Figure 2 .
Figure 2. EIU democracy index score in the sample.

Figure 3 .
Figure 3. Scatter plot of QoD and YCS by Wave.

Figure 4 .
Figure 4. Interaction of QoD and YCS on young people's predicted probability of support for democracy as a political system.

Figure 5 .
Figure 5. Scatter plot of support for democracy on QoD, by YCS.

Figure 6 .
Figure 6.Marginal effect of YCS on support for democracy by QoD (kernel estimation).

Table 1 .
Percentage of young people who thought democracy is Bad vrs.Good across study countries.

Table 2 .
Full logit table with all independent variables for regression of support for democracy on quality of democracy and youth cohort size.