Social origins of support for democracy: a study of intergenerational mobility

ABSTRACT

Recent evidence suggests that popular disaffection with liberal-democratic norms and institutions has been growing in different regions of the world, but studying the social origins of democratic versus authoritarian political preferences are especially relevant in countries with immature democratic practices. The main concern of this article is the association between intergenerational social mobility and support for democracy in post-socialist societies. I present a theoretical framework in which individuals’ political attitudes are affected by their intergenerational social mobility experiences. I model this theoretical argument using two complementary data-sets and various multivariable and multilevel statistical techniques. The results indicate that intergenerational social mobility, particularly its subjective perception, has statistically significant links with attitudes towards democracy and that this association is moderated by the attained level of democracy in the country where an individual resides. This may suggest that studying social origins of support for democracy by means of intergenerational social mobility can be an important tool to understand the conundrum of democratisation and democratic backsliding in post-socialist societies.

KEYWORDS:

• Social mobility
• democracy
• post-socialism
• public opinion
• multilevel analysis

Introduction

Intergenerational social mobility is the process where individual socio-economic standings in adult life represents a movement away from their parents’ socio-economic position. Some of the most influential social scientists suggested that intergenerational social mobility can facilitate social stability and integration into the political system, whereas downward mobility – or simply stability of marginalised groups – is expected to foster radicalism and political alienation among individuals. De Tocqueville (1835) in his ‘Democracy in America’ argued that in traditional societies all citizens occupy social positions that are fixed across generations, while in democratic nations some individuals and their families are rising up and others falling down the social ladder. Sorokin (1956), compiling data on upward mobility of the monarchs and presidents of different countries, stated that democratic societies are often characterised with more intensive upward mobility compared with non-democratic ones. More recently, Houle (2019), using data for more than 100 countries worldwide, empirically demonstrated that social mobility levels are positively associated with political stability.

If intergenerational social mobility affects attitudes towards the political system then existing regimes might, at least partially, derive their legitimacy by providing opportunities for intergenerational social mobility to their citizens through policies improving individuals’ life chances (Acemoglu et al., 2018; Leventoglu, 2005, 2014). Nonetheless, it is still unclear what individuals who experience upward or downward social mobility think of the political regime in which they live. In this article, I enquire into the association of intergenerational social mobility and support for democracy, which, at minimum, can be understood as individuals’ support for a system of government where the citizens exercise power by voting in free and fair elections. Although recent evidence suggests that popular disaffection with liberal-democratic norms and institutions has been growing in different regions of the world (Foa & Mounk, 2017; Széll, 2018), studying the implications of social mobility for democratic versus authoritarian political preferences are especially relevant in post-socialist societies. These countries experienced radical change and inherited specific social mobility patterns vis-à-vis the previous authoritarian regimes with their forceful de-stratifying policies (Gugushvili, 2017b; Jackson & Evans, 2017). However, we do not know much about the concrete implications of upward or downward mobility for individuals in these societies. Studying the links between social mobility and freely expressed political attitudes was clearly problematic in the socialist context, because political, economic and civil society activities were severely restricted, if not completely blocked.

There is an extensive literature on the process of democratisation, which has implications for our understanding of post-socialist political transitions. The most influential theoretical perspectives on democratisation focus on macro-level determinants such as economic development (Acemoglu & Robinson, 2000) and the distribution of resources (Boix, 2003), but they also imply that mass support for a democratic political regime underlies the process of democratisation. Welzel and Inglehart (2008), for instance, argue that in one of the most important waves of democratisation since the end of the 1980s, struggle for democracy, in socieites as different as South Korea, Indonesia, and East Germany, was not primarily about economic welfare and income inequality but rather about political equality. Support for democracy in a society is considered as a vital component of democratic consolidation because it prevents authoritarian political forces to seize power through electoral cycles and various types of revolutions (Hegre et al., 2001; Welzel & Inglehart, 2005).

There is a large scholarship on democratic perceptions in post-socialist societies and various aspects attached to democracy such as human rights, civil society and authoritarian leadership (e.g. Colton & Mcfaul, 2002; Evans & Whitefield, 1995; Gugushvili & Kabachnik, 2019; Hale, 2011; McAllister & White, 2017; Mishler & Rose, 1997; Gugushvili et al., 2016). Describing post-socialist societies, Lipset (1994), one of the leading political sociologists of the twentieth century, argued that ‘the success of democracy in these countries depends in large part on their populations’ ability to adapt to freedom, to break away from their former views on the role of the state’ (p. 13). From an individual-level perspective studies generally find that democracy in post-socialist countries is negatively perceived by unemployed, objectively and subjectively deprived, retired and less-educated respondents, while from a macro-level point of view, conditions such as low GDP per capita and income inequality levels affect support for newly established democratic institutions (Andersen, 2012; Gibson, 1996; Hayo, 2004; Kluegel & Mason, 2004; Linos & West, 2003; Pacek, 1994). To my knowledge, the role of intergenerational social mobility in attitudes towards an emerging post-socialist political system has not been systematically and comparatively explored.

In this study I intend to answer two related research questions: Are there statistically significant differences in attitudes towards the political system between intergenerationally mobile and non-mobile individuals? And, how does the attained level of democracy comparatively moderate the role of social mobility in attitudes towards the political system? The empirical evidence on the effect of intergenerational social mobility experience on various political attitudes is scarce, dated, and inconclusive (Blau, 1956; Daenekindt et al., 2018; Nieuwbeerta et al., 2000; Sorokin, 1956; Weakliem, 1992). In this study, however, I will argue that, unlike the previous research which mostly deals with voting behaviour and left-right cleavages within the established democratic environment, the consequences of intergenerational social mobility on political attitudes can be more relevant to support for democracy in post-socialist societies.

The argument

Self-interest, middle class, and support for democracy

There are several reasons why the experience of intergenerational upward or downward mobility might be relevant to support for democracy. All post-socialist countries began formal transition from an authoritarian to a democratic political order between 1989 and 1991. Unlike situations where links between mobility and attitudes are researched in relation to specific social and economic policies, as well as concrete political parties, attitudinal preferences among the socially mobile during and after transition are more likely to be related to the broad macro-structural changes. In comparison to non-mobile social groups, upwardly and downwardly mobile individuals have more reason to support or oppose the political system as it affects their current socio-economic status. Studies of post-socialist societies, for instance, show that attitudes towards the political system are based on perceptions that the system is open and fair and provides opportunities to everyone, while the perception of social inequality is positively associated with support for ‘strong-hand’ authoritarian government (Whitefield & Loveless, 2013).

The role of self-interest in shaping attitudes of political nature has been extensively researched (e.g. Barnes, 2015; Linos & West, 2003). However, it is still not clear what type of political system is most desirable for socially mobile individuals. In some post-socialist countries, small groups of people managed to succeed in transitional economic reforms (Titma & Roots, 2006). The systems within which these individuals operated typically were not full-fledged democracies. A significant share of their populations experienced downward mobility (Gerber & Hout, 2004; Gugushvili, 2017a). Downwardly and upwardly mobile individuals are likely to, at least partially, blame or support emerging systems, democratic or authoritarian, for personal failures and successes. The self-interest thesis implies that support for either democracy or autocracy depends on which regime is likely to deliver the policies which serve the utility of upwardly or downwardly mobile individuals. A baseline economic self-interest view suggests that, given their interest in preserving attained gains, the upwardly mobile are primarily interested in the regime that best delivers ‘political order.’ Since political order may be a characteristic of either democracies or autocracies, the ‘economic view’ is alone insufficient to explain support for either democracy or authoritarianism.

Among various alternative perspectives on socio-economic foundations of democracy, the democratic transition literature in the political economy tradition probably most appropriately explains political preferences of intergenerationally mobile individuals in post-socialist societies. This approach implies that democratic transition and its sustainability require the existence of a strong ‘middle class’ which demands increased political participation (Przeworski et al., 2000). Assuming that intergenerational upward mobility is a pre-requisite of middle-class expansion, we can expect that, on average, upwardly mobile individuals would prefer a democratic political order. This thinking is also corroborated by insights from social psychology suggesting that people are more likely to attribute failure to factors that are beyond their control and more likely to explain successes by pointing to their own merits, abilities and effort (Miller & Ross, 1975). If we assume that individuals’ attitudes towards the political system in post-socialist societies are affected by life-long experiences, then their initial set of attitudes are amended over the years based on their experience of intergenerational social mobility and an associated perception of the role ascribed and attained factors play in determining downward or upward mobility (Gugushvili, 2016a, 2016b). Although democracies do not necessarily guarantee good governance and high human well-being, the nature of democracy implies that democratic societies are more open and people have more opportunities to shape their lives than countries with authoritarian political regimes (Rothstein, 2018).

Type of regime and support for democracy in comparative perspective

Building on the framework presented above, we still cannot expect that upward social mobility has an outright positive effect on attitudes about a new and emerging political system, and there are reasons to believe that the strength of these links varies in different countries. It has been argued that high levels of social mobility may in fact facilitate the stability of a regime via middle-class or lower-class behaviour, but stagnant upward mobility or the threat of downward mobility among the middle class creates incentives that facilitate democratic transition (Leventoglu, 2005). In the latter scenario, the existing regime could try to sustain the autocracy by providing the lower classes with more opportunities for social mobility, which in turn makes it less likely that they will support a democratic transition. The opposite applies to democratic environments in which increased opportunities for upward social mobility among the lower classes create weaker incentives to support autocratic tendencies, while a democratic transition accompanied by prolonged high levels of intergenerational downward mobility and too much erosion of such a ‘middle class’ will likely lead to dissatisfaction with democracy (Leventoglu, 2014). In this scenario, individuals would prefer the old regime, or some variant of authoritarianism, to restore their former status.

Based on the described framework, people who had opportunities of moving up in the social hierarchy of post-socialist authoritarianism may exhibit strong attachment to pre-existing social and political orders and tend to reject support for democracy. It seems reasonable to expect the emergence of a ‘cult of gratitude’ among the upwardly mobile individuals who will be more likely to support the existing political system that allowed them to succeed by providing them with increased life chances (Jackman, 1972). Therefore, I expect that attitudes towards the political system are affected by the existing level of democracy within a given society. To reiterate, the argument here is that people are more likely to associate their success not only with equality of opportunities, as such, but with the existing system – no matter how democratic. In other words, when people rise to the top of the social hierarchy and maintain this position, the experience likely creates strong attachment to the society that provided them with increased life chances, not to abstract political notions or concepts (Kelley, 1992). In this vein, support for democracy among upwardly mobile individuals will be stronger in more democratically advanced post-socialist countries, while authoritarian rule might be preferable among the upwardly mobile in less democratic post-socialist countries.

Patterns of democratisation in post-socialist societies

Post-socialist societies are significantly different from each other by the levels and types of democracy they have achieved and this provides an opportunity to explore the main research question of this article in comparative perspective. Transitional societies in post-socialist Europe can be divided in several territorial, cultural and geopolitical sub-groups, which also reflect the democratisation trends in these countries. The most democratically advanced post-socialist societies are Central and Eastern European new EU member countries. Existing evidence suggests that, among other factors, the process of EU accession itself contributed to democratic advances particularly in the least developed countries such as Bulgaria and Romania (Spendzharova & Vachudova, 2012). Three Baltic States – Estonia, Lithuania and Latvia – have diverged with their democratisation paths from other former Soviet republics since gaining independence in 1991. Although they have experienced significant advances towards full democratisation, Latvia and Estonia due to their multi-ethnic population compositions struggled to balance ethnic pluralism, nation building, and democracy (Duvold & Berglund, 2014).

The group of Balkan countries generally made progress towards democratisation since the 1990s, but this group also consists of severely war-torn societies characterised by considerable variance of democratic practices (Grimm & Mathis, 2018). The group of non-Baltic former Soviet Union republics consists of countries in Central Asia and South Caucasus as well as Belarus, Moldova, Russia and Ukraine. These countries are the least democratic among other post-socialist societies. The so called ‘coloured revolutions’ in Georgia, Ukraine, and Kyrgyzstan in the early 2000s generated hopes for an accelerated process of democratisation throughout the former Soviet Union, but these revolutions in fact provoked autocrats in Azerbaijan, Belarus, Kazakhstan, Russia, and other post-Soviet countries to strengthen domestic efforts to pre-empt challenges related to democratisation processes (Silitski, 2010). The recent comparative analysis of the Regime Legitimation Expert Survey suggests that one of the mechanisms through which the authoritarian rulers in these countries maintain power is their legitimating claim that they represent the guardians of citizens’ socio-economic well-being (Soest & Grauvogel, 2016). The latter relates to the main thesis of this study that support for a political regime by socially mobile individuals is conditioned by the level of attained democracy in post-socialist countries.

In sum, based on the proposed theoretical framework, in the empirical part of this article I test the following hypothesis: Intergenerational upward (downward) mobility is positively (negatively) linked with support for democracy but this association is primarily manifest in countries with high levels of democratic attainment.

Research design

Data-sets

European Values Studies (EVS) and Life in Transition Survey (LITS), respectively conducted in 2008 and 2010, provide an opportunity to study the link between intergenerational social mobility and support for democracy in post-socialist societies. EVS is employed to look at links between intergenerational mobility in occupational status and support for democracy; LITS in turn provides information on subjective perception of intergenerational mobility and allows investigating if the latter is associated with support for democracy. Both in EVS and LITS face-to-face interviews were conducted using representative, multi-stage (or stratified) random samples of the adult populations in the following 21 countries: Albania, Armenia, Azerbaijan, Belarus, Bulgaria, Croatia, Czech Republic, Estonia, Georgia, Hungary, Latvia, Lithuania, North Macedonia, Moldova, Poland, Romania, Russia, Serbia, Slovak Republic, Slovenia, and Ukraine.¹ After list-wise deletion of missing data and censoring individuals below age of 25, the sample sizes for multivariable analysis are 14,657 and 19,073 individuals in EVS and LITS, respectively.²

Dependent variables

In EVS, the dependent variable on attitudes towards the political system comes from the question that asks respondents whether they ‘disagree strongly’ = 1, ‘disagree’ = 2, ‘agree’ = 3 or ‘agree strongly’ = 4 that ‘democracy may have problems but it is better than any other form of government.’ Descriptive statistics for the pooled sample indicate that support for the idea that democracy is better than any other form of government is strong with a mean value of 3.14 (SD 0.68). The categorical distribution of responses is presented in Figure 1. It is clear that only a minority of respective populations disagree that democracy is better than any other form of government, but differences within the countries are also obvious (Table s1 in online supplementary materials). The new EU member countries, both Central and Eastern European and Baltic societies, expressed the lowest support for democracy among other post-socialist regions. In sensitivity tests of the main results, I also analyse alternative dependent variables on democracy which are available in the employed data-set.

Figure 1. Descriptive statistics of dependent and independent variables, distribution of valid answers, %. Source: Author’s calculations based on data from EVS (2010) and EBRD (2010).

For the dependent variable in LITS, respondents are asked to choose one of the three following options: (1) ‘democracy is preferable to any other form of political system,’ (2) ‘for people like me, it does not matter whether a government is democratic or authoritarian,’ (3) ‘under some circumstances, an authoritarian government may be preferable to a democratic one.’ The distribution of answers for the pooled sample in Figure 1 indicates that half of the respondents agreed that democracy is preferable to any other form of government, while only one fifth of respondents agreed that an authoritarian form of government may be preferable under some circumstances. Support for democracy is highest in the Former Soviet Union and lowest in the Baltic States.³

Independent variables

The main explanatory variable employed in this article is intergenerational social mobility experienced and reported by respondents. EVS gives information about the Standard International Socio-economic Index of Occupational Status (ISEI) (Ganzeboom & Treiman, 1996) which varies from 16 to 90. To operationalise social mobility in occupational status, I divide both respondents and their parents into the bottom, middle, and top tertiles of occupational status attainment (Gugushvili et al., 2017). Table 1 shows all possible mobility trajectories and suggests that most individuals do not experience intergenerational upgrades in occupational status. I coded those respondents who moved from the middle to the bottom and from the top to the middle tertiles as downwardly mobile, while those respondents who moved from the top to the bottom tertile are coded as strongly downwardly mobile. I apply the reverse order for the upper end of intergenerational mobility. It is noticeable that only a minority of respondents experienced strong upward and strong downward mobility in occupational status.

Table 1. Operationalisation of intergenerational mobility in occupational status.

		Respondents
Parents	Bottom tertile	Middle tertile	Top tertile
Bottom tertile	1→1 Non-mobile: 17.7%	1→2: Upward: 12.4%	1→3: Strongly upward: 8.2%
Middle tertile	2→1 Downward: 10.6%	2→2: Non-mobile: 11.7%	2→3: Upward: 9.4%
Top tertile	3→1 Strongly downward: 5.7%	3→2: Downward: 9.3%	3→3: Non-mobile: 15.0%

Source: Author's calculations based on data from EVS (2010).

In LITS, respondents are asked to compare their parents’ position to their own by agreeing or disagreeing with the following statement: ‘I have done better in life than my parents.’ I transform the answers from this question into five categorical variables: ‘strongly disagree’ = strongly downwardly mobile, ‘disagree’ = downwardly mobile, ‘neither disagree nor agree’ = non-mobile, ‘agree’ = upwardly mobile and ‘strongly agree’ = strongly upwardly mobile. The main characteristic of this indicator of intergenerational mobility in post-socialist context is that it does not necessarily imply social mobility in terms of educational, occupational, or income mobility. It can also reflect the structural upgrade or downgrade of the economy, e.g. collective mobility (Gugushvili, in press). But when the answers on subjective intergenerational mobility are contrasted with individuals’ perceptions on how well they have done in life when compared with their high school classmates and colleagues (measuring collective mobility), subjective intergenerational mobility still maintains unique variations and therefore can serve as an alternative indicator of objective intergenerational mobility (Zaidi et al., 2009).

Figure 1 shows the distribution of occupational and perceived intergenerational mobility in the pooled sample of post-socialist societies. The main difference between the depicted frequencies is that a much higher share of individuals are intergenerationally non-mobile in terms of occupational mobility, while significantly more individuals declare that they have experienced upward mobility than is observed for mobility in occupational status. The frequencies for separate countries (shown in Table s2 in online supplementary materials) also suggest that cross-national differences are more vividly pronounced in subjective rather than objective occupational mobility.

Control variables

I employ an array of socio-demographic variables that are expected to associate with the dependent variables: gender; marriage, age and age of respondents squared and divided by 100; the highest level of completed education is based on the 1997 version of International Standard Classification of Education (ISCED), which varies from ISCED 0 (pre-primary education) to ISCED 6 (second stage of tertiary education); I created dummy variables for the following types of labour market status: employed, unemployed, students, retired, and an ‘other’ category which includes homemakers, disabled people, those who do not want to work, have no need to work, or cannot find suitable jobs.

In addition to the described variables which are identical in EVS and LITS, I also account for survey specific comparable controls. In EVS, type of settlement is accounted for with the size of the population in a respondents’ location, while in LITS type of settlement is given with urban vs. rural residency. For respondents’ socio-economic status, in EVS, I control for monthly household income, while in LITS, respondents were asked to place their households on a ten-step hierarchical ladder. In EVS, I account for the individuals’ tertiles of occupational status, while in LITS, respondents’ occupational attainment is classified with the 1958 version of the International Standard Classification of Occupations (ISCO). I operationalise the following major categories: white collar occupations, unskilled service occupations, agricultural workers and minors, and manual workers. Lastly, in both data-sets I account for parental education. In EVS, the mean of parental education is given as an ISCED variable, while in LITS parental education is years of educational attainment which I divide in three groups: 0–5 years, 6–11, and 12 years or more (for descriptive statistics refer to Table s3 in online supplementary materials).

Macro-level variables

I employ macro-level data for 2008 and 2010 for EVS and LITS, respectively. The measure of democracy used in this article is the democracy index of the Economist Intelligence Unit (EIU) (EIU, 2010). The EIU index of democracy values vary from 0 = no democracy to 10 = absolute democracy. In the pooled sample of post-socialist societies, the mean value of this index slightly decreased from 6.4–6.1 in 2008–2010 (Kekic, 2007). Non-Baltic former Soviet Union republics clearly lag behind by the level of democratic reforms (4.7), while all EU member countries on average score much higher (7.3).

The level of economic development is measured by GDP per capita based on purchasing power parity (PPP). Data are in constant 2005 international dollars and are derived from the World Bank’s (2017) World Development Indicators (WDI) database. As the consequence of the economic crisis in the end of the 2000s, the mean value GDP ppp per capita decreased from 15,191 in 2008 to 14,828 in 2010. There are also clear differences in economic development between the former Soviet Union (USD 9,701) and other EU member countries (USD 20,491). I use standardised contextual variables in regression models (actual values for separate countries are shown in Table s4 in online supplementary materials).

Methods

I start multivariable analysis by fitting generalised ordered and multinomial logistic regressions. The ordinal form of the dependent variable in EVS permits the estimation of simple ordered logistic regression, but the conducted Brant Test indicated that in the parallel-lines model assumption was violated for a number of independent variables. Instead, the maximum-likelihood generalised ordered logit model was used which relaxes the proportional odds assumption and allows the effects of the independent variables to vary across ordered categories of the dependent variable (Williams, 2006). On the other hand, the nominal nature of the outcome variable in LITS allows fitting multinomial logistic regression. For both EVS and LITS, the following equation is the general regression specification applied for estimating the role of intergenerational social mobility in attitudes towards democracy where i indicates individual respondents, $\sum \delta X_{ik}$ represents control variables and country fixed-effects, while ε depicts the error term:(1) $\mathrm{S}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{d}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{c}\mathrm{r}\mathrm{a}\mathrm{c}{\mathrm{y}}_i={\gamma }_0+{\gamma }_1\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{y}\mathrm{u}\mathrm{p}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_i+{\gamma }_2\mathrm{U}\mathrm{p}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_i+{\gamma }_3\mathrm{D}\mathrm{o}\mathrm{w}\mathrm{n}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_i+{\gamma }_4\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{y}\mathrm{d}\mathrm{o}\mathrm{w}\mathrm{n}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_i+\sum \delta X_{ik}+{\epsilon }_i$(1) To estimate how macro contextual factors moderate the association between intergenerational social mobility and support for democracy, I employ a multilevel regression technique (Hox, 2002). It consists of level one individual analysis, level two country contextual characteristics of democracy and economic development, and the cross-level interactions of level two variables with intergenerational mobility in occupational status and subjective perception of mobility. These interaction terms indicate how contextual variables moderate the relationship between social mobility and support for democracy. Equation 2(2) $\begin{array}{rl}\mathrm{S}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{d}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{c}\mathrm{r}\mathrm{a}\mathrm{c}{\mathrm{y}}_{ij}=& {\gamma }_0+{\gamma }_1\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{y}\mathrm{u}\mathrm{p}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}+{\gamma }_2\mathrm{U}\mathrm{p}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}+{\gamma }_3\mathrm{D}\mathrm{o}\mathrm{w}\mathrm{n}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}\\ & +{\gamma }_4\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{y}\mathrm{d}\mathrm{o}\mathrm{w}\mathrm{n}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}+\sum \delta X_{ik}+{\epsilon }_{ij}+{\phi }_1\mathrm{D}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{c}\mathrm{r}\mathrm{a}\mathrm{c}{\mathrm{y}}_j\\ & +{\phi }_2\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{i}\mathrm{c}\mathrm{d}\mathrm{e}\mathrm{v}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{m}\mathrm{e}\mathrm{n}{\mathrm{t}}_j+\sum _{\left(l=1\right)}^2{\eta }_1({\gamma }_{1-4}\mathrm{M}\mathrm{o}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}{\mathrm{y}}_{ij}\\ & \times ({\phi }_1\mathrm{D}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{c}\mathrm{r}\mathrm{a}\mathrm{c}{\mathrm{y}}_j+{\phi }_2\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{i}\mathrm{c}\mathrm{d}\mathrm{e}\mathrm{v}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{m}\mathrm{e}\mathrm{n}{\mathrm{t}}_j))+U_{0j}.\end{array}$(2) formally outlines the applied analytical strategy.(2) $\begin{array}{rl}\mathrm{S}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{d}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{c}\mathrm{r}\mathrm{a}\mathrm{c}{\mathrm{y}}_{ij}=& {\gamma }_0+{\gamma }_1\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{y}\mathrm{u}\mathrm{p}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}+{\gamma }_2\mathrm{U}\mathrm{p}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}+{\gamma }_3\mathrm{D}\mathrm{o}\mathrm{w}\mathrm{n}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}\\ & +{\gamma }_4\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{y}\mathrm{d}\mathrm{o}\mathrm{w}\mathrm{n}\mathrm{w}\mathrm{a}\mathrm{r}{\mathrm{d}}_{ij}+\sum \delta X_{ik}+{\epsilon }_{ij}+{\phi }_1\mathrm{D}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{c}\mathrm{r}\mathrm{a}\mathrm{c}{\mathrm{y}}_j\\ & +{\phi }_2\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{i}\mathrm{c}\mathrm{d}\mathrm{e}\mathrm{v}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{m}\mathrm{e}\mathrm{n}{\mathrm{t}}_j+\sum _{\left(l=1\right)}^2{\eta }_1({\gamma }_{1-4}\mathrm{M}\mathrm{o}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}{\mathrm{y}}_{ij}\\ & \times ({\phi }_1\mathrm{D}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{c}\mathrm{r}\mathrm{a}\mathrm{c}{\mathrm{y}}_j+{\phi }_2\mathrm{E}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{m}\mathrm{i}\mathrm{c}\mathrm{d}\mathrm{e}\mathrm{v}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{m}\mathrm{e}\mathrm{n}{\mathrm{t}}_j))+U_{0j}.\end{array}$(2)

Where U_0j represents the random component for the intercept (γ₀). To simplify analysis and the interpretation of results in multilevel models, I transform both dependent variables in binary form (in EVS ‘agree strongly’ that ‘democracy is better than any other form of government’ = 1 and in LITS ‘democracy is preferable’ = 1) and run multilevel mixed-effects logistic regressions with maximum likelihood estimators using ‘xtmelogit’ command in Stata 15 statistical package.

Results

Bivariate associations and multivariable analysis

I first present bivariate associations between intergenerational social mobility and support for democracy. The results in Figure 2 indicate that if there is any relationship between objective social mobility and democratic preferences then it is weak. Those who experience strong upward mobility have marginally stronger preferences for democracy than non-mobile individuals. It is important to remember that the bivariate associations do not account for country differences or various socio-demographic variables that are expected to affect support for democracy. Figure 2 also illustrates the links between subjective perception of intergenerational mobility and political system preferences. It is evident that fewer individuals who have experienced downward mobility believe that democracy is preferable to any other form of political system. On the other hand, the majority of upwardly mobile people agree with the latter assertion.

Figure 2. Occupational/perception of intergenerational social mobility and attitudes towards democratic political system, distribution of valid answers, %. Source: Author’s calculations based on data from EVS (2010) and EBRD (2010).

In Table 2, I run generalised ordered logistic regression for intergenerational mobility in occupational status and multinomial logistic regression for subjective perception of intergenerational mobility (see Table s5 in online supplementary materials for full results with control variables). Before describing the main results, it is important to explain how to interpret relative risk ratios (RRRs) in generalised ordered logistic regression. The first panel in Model 1 contrasts the answer option ‘disagree strongly’ which has a value of 1 with answer options (2) ‘disagree’, (3) ‘agree’, and (4) ‘agree strongly’. The last panel in turn contrasts the first three answer options ‘disagree strongly’, ‘disagree’, and ‘agree’ with the last answer option ‘agree strongly’. The same logic applies to Panel 2 which is located in between.

Table 2. Occupational/perception of intergenerational mobility and attitudes toward democratic political system, RRRs from generalised ordered logistic (M1) and multinomial logistic (M2) regressions

	M1: Objective mobility			M2: Subjective mobility
	Democracy is better than any other form of government			Democracy vs. authoritarianism
	(1) Disagree strongly vs. disagree, agree, agree strongly	(2) Disagree strongly, disagree vs. agree, agree strongly	(3) Disagree strongly, disagree, agree vs. agree strongly	(1) Democracy is preferable	(2) Authoritarian. may be preferable
Intercept	66.2 (56.7)***	1.87 (0.61)	0.23 (0.06)***	0.37 (0.06)***	0.03 (0.01)***
Mobility
Strongly downward	1.36 (0.44)	1.11 (0.14)	1.00 (0.10)	0.89 (0.06)	0.86 (0.07)
Downward	0.86 (0.14)	0.96 (0.07)	1.04 (0.06)	1.01 (0.06)	1.05 (0.07)
Upward	1.04 (0.19)	0.88 (0.06)	0.98 (0.05)	1.33 (0.07)***	0.98 (0.06)
Strongly upward	1.30 (0.40)	0.80 (0.09)	0.91 (0.07)	1.34 (0.09)***	0.95 (0.08)
Statistics
Observations	14,480			17,822
Pseudo R-squared	0.04			0.06

Notes: ***, **, and * denote statistical significance at the 0.001, 0.01, and 0.05 levels. Robust standard errors are in parentheses. Models account for control variables and country fixed-effects. Source: Author's calculations based on data from EVS (2010) and LITS (EBRD, 2010).

The output of the generalised ordered logistic regression indicates that objective intergenerational mobility does not have a statistically significant association with support for democracy. The systemic and statistically significant associations are revealed, however, between subjective perception of intergenerational mobility and the dependent variable. Both the upward and strongly upward mobile individuals tend to agree with the idea that democracy is preferable to any other form of political system. These two groups of individuals are 1.33 (SE 0.07) and 1.34 (SE 0.09) times more likely to support democracy than non-mobile individuals. Thus, in the multivariable analysis, I find no association between objective intergenerational mobility and attitudes toward democracy, though perceived social mobility in the upward direction is significantly and positively associated with preferences for democracy.

Multilevel analysis

In Table 3, I fit the multilevel mixed-effects logistic regression models with contextual variables and their cross-level interaction terms in order to observe if the level of democracy moderates the association between intergenerational mobility and the dependent variables. Regarding intergenerational mobility in occupational status, the interaction term of the EIU democracy index and strong downward mobility maintains statistical significance both in Model 1 with odds ratio of 0.80 (SE 0.07) and in Model 3 with odds ratio of 0.77 (SE 0.08). The results are in line with my expectation that downwardly mobile individuals in more democratic societies are less enthusiastic about the democratic political system. The interaction term between the level of democracy and strong upward mobility is also statistically significant for subjective mobility with odds ratio of 1.26 (SE 0.09) in Model 3. At the same time, however, when economic development and its interaction term with intergenerational mobility are accounted for, the interaction effects between democracy and downward mobility are also positive and significant. Nonetheless, controlling for the attained level of democracy, support for democracy among downwardly mobile individuals is weaker in more economically developed societies. The latter suggests that downwardly mobile individuals are less likely to prefer the democratic political system when their countries achieve the comparatively high level of economic prosperity.

Table 3. Occupational/perception of intergenerational mobility, contextual variables and support for democracy, ORs from multilevel mixed-effects logistic regression models.

	I. Objective intergenerational mobility			II. Subjective intergenerational mobility
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Mobility
Strongly downward	0.87 (0.08)	0.98 (0.12)	0.81 (0.11)	0.94 (0.06)	0.94 (0.06)	0.91 (0.06)
Downward	0.95 (0.05)	0.95 (0.07)	0.89 (0.07)	0.98 (0.05)	0.99 (0.05)	0.96 (0.05)
Upward	1.02 (0.05)	1.02 (0.07)	0.96 (0.08)	1.34 (0.06)***	1.35 (0.06)***	1.34 (0.06)***
Strongly upward	1.05 (0.07)	1.01 (0.10)	1.03 (0.11)	1.31 (0.08)***	1.32 (0.08)***	1.30 (0.08)***
Macro-level variables
Stand. democracy index	0.85 (0.06)*	–	0.95 (0.08)	0.81(0.08)*	–	0.85 (0.10)
Stand. GDP PPP per capita	–	0.76 (0.07)**	0.79 (0.09)*	–	0.83 (0.08)*	0.90 (0.10)
Cross-level interactions
Strongly downward*democracy	0.80 (0.07)**	–	0.77 (0.08)**	1.09 (0.08)	–	1.26 (0.12)*
Downward*democracy	0.95 (0.05)	–	0.91 (0.06)	1.17 (0.06)**	–	1.31 (0.09)***
Upward*democracy	0.95 (0.05)	–	0.92 (0.05)	1.07 (0.05)	–	1.11 (0.06)
Strongly upward*democracy	1.04 (0.07)	–	1.03 (0.09)	1.28 (0.08)***	–	1.26 (0.09)**
Strongly downward*GDP	–	0.88 (0.10)	1.10 (0.16)	–	0.91 (0.06)	0.81 (0.07)*
Downward* GDP	–	1.01 (0.06)	1.09 (0.09)	–	0.97 (0.05)	0.84 (0.05)**
Upward* GDP	–	1.00 (0.06)	1.08 (0.09)	–	1.00 (0.04)	0.94 (0.05)
Strongly upward* GDP	–	1.05 (0.09)	1.02 (0.11)	–	1.18 (0.07)**	1.05 (0.07)
Random intercept	0.32 (0.05)	0.31 (0.05)	0.29 (0.05)	0.42 (0.07)	0.39 (0.06)	0.40 (0.06)
Statistics
AIC	16,997	17,003	17,002	23,354	23,357	23,345
BIC	17,201	17,208	17,245	23,603	23,606	23,634
Observations	14,480	14,480	14,480	17,822	17,822	17,822
Countries	21	21	21	21	21	21

Notes: ***, **, and * denote statistical significance at the 0.001, 0.01, and 0.05 levels. Robust standard errors are in parentheses.

Source: Author’s calculations based on data from EVS (2010) and EBRD (2010), EIU (2010), and World Bank’s (2017).

Finally, to reveal how specific values of EIU index of democracy are associated with support of democracy, Figure 3 depicts marginal effects from strong downward mobility in occupational status and the perception of strong upward mobility on support for democracy. The interpretation of the results is straightforward – the interaction terms are significant if the 95% confidence intervals do not cross the horizontal zero line. For occupational intergenerational mobility, the higher values of the index of democracy decrease the marginal effects of strong downward mobility on democratic preferences, but these associations become statistically significant only when EUI democracy index reaches more than 7.0 in countries such as Croatia, Bulgaria, Poland or Latvia. In these countries, strongly downwardly mobile individuals are about 10% less likely to agree that democracy is better than any other form of government. The effect of democracy on the links between intergenerational subjective mobility and support for democracy is more obvious for perceived mobility. In countries with the highest democracy scores such as Slovenia and Czech Republic strong upward mobility leads to about 15% stronger support for democracy, while in countries with a democracy score of less than 5.0 such as Armenia, Georgia or Russia upward social mobility is not associated with greater democratic preferences.

Figure 3. Occupational/perception of intergenerational social mobility, the EUI democracy index and support for democracy.

Notes: Dashed lines represent 95% confidence intervals. Source: Author’s calculations based on data from EVS (2010), EBRD (2010) and EIU (2010).

Robustness of findings

In online supplementary materials, I further check if the derived results hold when I fit statistical models with different specifications. First, in models on occupational and subjective intergenerational mobility I control for only those variables which are identical in EVS and LITS (Table s6). Second, to simultaneously account for the effects stemming from individuals’ origins and destinations, I analyse how far specific mobility trajectories are associated with support for democracy by calculating post-estimation predicted probabilities for individuals in all possible trajectories of occupational mobility (Figure s1). Third, I employ alternative dependent variables on democracy from EVS which enquires into respondents’ attitudes towards economic development and maintaining order in a democratic political system (Table s7). Finally, instead of fitting generalised ordered logistic regressions, I run alternative model specifications with separate linear probability models for each outcome in the dependent variable (Tables s8 and s9). These tests confirm that objectively mobile individuals are not statistically different in their support for democracy when compared with non-mobile individuals; subjectively upwardly mobile individuals still think that democracy is preferable to any other form of political system, while the level of democracy maintains its moderating effect.

Discussion and conclusions

In this article, I attempted to determine whether intergenerational social mobility has any links to public perceptions of the legitimacy of democracy. Although the research framework intended to contribute to the general understanding of the social bases of support for democracy, this question is especially relevant and under-explored in post-socialist transition societies. Earlier research, which mainly concentrated on the Western European countries, looked at the implications of social mobility on more specific components of political action such as individual voting patterns (e.g. Graaf et al., 1995; Paterson, 2008; Weakliem, 1992). In my analysis, I used the broader variables on attitudes towards democracy, which I hypothesised to have significant links with intergenerational social mobility.

In the theoretical framework, I argued that upwardly mobile individuals have positive perceptions of the emerging democratic system, while downwardly mobile individuals have negative perceptions of democracy; and both of these associations should be primarily observed in countries with already high levels of democratic development. This reasoning is mainly based on an assumption those individuals who experienced upward mobility, or simply maintained an advantageous social status during post-socialist transition would favour emerging institutions no matter how democratic or authoritarian these institutions were. In addition, I also assumed that, despite its many shortcomings, democracies are more open and people have more opportunities to shape their lives than they do in countries with authoritarian political regimes, and this is one of the reasons why upwardly mobile individuals would prefer democratic political order. Using both an objective occupational indicator and the subjective perception of intergenerational mobility from EVS and LITS and attitudes towards democracy as dependent variables, I found that when the level of democracy in countries of respondents’ residence is not accounted for, intergenerational mobility in occupational status is not significantly associated with support for democracy. Nonetheless, this association in terms of subjective perception of intergenerational social mobility is statistically significant.

Finding the significant association between perceived upward mobility and support for democracy regardless of the type of regime in which individuals reside suggests that intergenerational mobility is associated with democratic preferences even if upwardly mobile individuals did not live under democracy or directly benefited from it. As mentioned in the research framework section, some evidence exists that an upward mobility experience likely generates the perception that success in life is determined by hard work, effort, skills and intelligence, while the experience of downward mobility is likely to create an opposite worldview – that life chances are conditioned by social injustice in society (Schmidt, 2011). If the latter is the case, intergenerationally mobile persons’ perceptions might be that democracy is a political system in which individuals have a greater control over their lives. Democracies undoubtedly provide better possibilities for winning elections for any individuals who run for political office on local, regional, national levels than do strictly controlled or rigged ballots (Collins, 2004). In the social stratification literature it is well known that minority groups are often socially less mobile (Heath, 2007), while in democratic societies minority viewpoints are more likely to be respected, individuals enjoy more personal freedoms and people are less likely to be discriminated against based on their ethnicity. Further, gender equality, right to travel, choice of work and study are more likely to be the norm in democracies than in autocracies – all of which could have an effect on intergenerational social mobility.

One of the caveats of the finding that subjectively upwardly mobile individuals have higher support for democracy than non-mobile individuals, regardless of political context, is that there are many reasons why respondents may feel they are doing better than their parents; such as the general standards of living, family trajectories, and adverse life course outcomes which are not adequately accounted for by the employed control variables in my regression models. The very fact that most of the parents of survey respondents lived under socialism while the survey respondents did not may be a decisive factor in shaping whether respondents perceive they are better or worse off than their parents, and in a manner that could be endogenous to perceptions of the current political regime. Respondents who think that democracy fuels disorder and uncertainty may feel they are worse off than their parents because their parents lived in a socialist system, while those who associate democracy instead with freedom from censorship and oppression and more opportunity for individual expression and advancement may feel better off than their parents for the same reason.

Based on the theoretical framework, it is not surprising that upwardly mobile individuals prefer a democratic political order, but without knowing the context where individuals live in it is not completely clear what type of political system is in the best interest of socially mobile individuals. If people rise to the top of the social hierarchy or maintain their advantageous position within a particular type of society, the experience might build up strong attachment to the system that provided them with mobility opportunities, rather than a democratic political regime. So I expected that among upwardly mobile individuals, support for democracy would be stronger in more democratically developed societies, while authoritarian rule would be more strongly preferred among the upwardly mobile individuals in less democratic countries. My findings are broadly in line with this expectation. Using a multilevel research framework, I showed that, on the one hand, negative attitudes towards democracy among occupationally downwardly mobile individuals are stronger in more democratically advanced societies; and, on the other hand, positive attitudes towards democracy among subjectively upwardly mobile individuals are also stronger in more democratically advanced societies.

One of the salient issues in studying the links between support for democracy and objectively attained levels of democracy using cross-national survey data is whether or not survey respondents are consistently and meaningfully interpreting the concept of ‘democracy’ in different post-socialist countries (Ariely & Davidov, 2011). This might depend on the respondents’ self-assessment of the character of the regime they live in, and if there is discrepancy between these subjective views and the objective democracy measures such as the EIU index of democracy used in the current study. It is known that in authoritarian states individuals tend to overestimate how democratic their political systems are (Gerber & Chapman, 2018). On the other hand, respondents in more authoritarian societies with their strong support for democracy could be expressing a desire for political change rather than supporting their existing regime. In both instances, the issue of interpretation is not a major concern for the current study since I am interested in differences in support for democracy among various mobility groups rather than revealing if the subjective perception and preference for democracy has the same meaning in different political contexts.

Having in mind the limitations of this study, my findings have implications for understanding recent trends in democratisation in post-socialist countries. The observation that in authoritarian states upward intergenerational mobility does not lead to support for democracy is in line with recent claims that politically repressive countries may provide their citizens with greater social mobility opportunities (Leventoglu, 2014; Soest & Grauvogel, 2016). It should not be surprising then that socially mobile individuals in many post-socialist societies are less enthusiastic about democracy as the preferred form of political order. Interestingly, countries which selected their democratisation paths in the start of the post-socialist transition largely stayed on their paths as suggested by a strong association between democracy index between the first half of the 1990s and the first half of the 2010. In other words, there have not been the major shifts from democracy to authoritarianism, and vice versa. However, over the last few years even the most democratically advanced countries in this region experienced problems reverting to semi-authoritarian practices, what has been referred to as ‘backsliding’ of democracy (Greskovits, 2015).

These developments particularly intensified after the major economic crisis in Europe in the end of 2000s (Armingeon et al., 2016). In this period, steps towards authoritarianism have been primarily facilitated by the more radical political parties’ favourable electoral outcomes and the resultant policy changes. It is known that support for democracy, and consequently voting behaviour, can be affected not only by the general performance of the political system, but also by short-term economic shocks (Cordero & Simón, 2016). My findings imply that support for democracy can be affected if the major economic crisis, both in the beginning of the 1990s and in the end of the 2000s, also affected their perceptions how well they have done in life in comparison to their parents. In fact, the results of my analysis indicate that in most democratic post-socialist countries such as Poland, Hungary, and Czech Republic, both occupational downward mobility and subjective perception of upward mobility are important explanations of support for democracy. This may suggest that studying socio-economic origins of support for democracy by means of intergenerational social mobility can be an important tool to understand the conundrum of democratisation and democratic backsliding in post-socialist societies.

Acknowledgements

The author is thankful to Fabrizio Bernardi, Martin Kohli, Martin K. Whyte, Ellu Saar, and Nate Breznau, for valuable feedback on earlier versions of this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes on contributor

Alexi Gugushvili was Assistant Professor of Sociology at the Erasmus University Rotterdam before taking up the current position of Associate Professor of Sociology at the University of Oslo. He obtained a PhD in Political and Social Sciences (2014) from the European University Institute in Florence, Italy. His research interests include the trends, covariates and consequences of intergenerational social mobility, the process of democratisation, comparative social welfare, and health inequalities.

Additional information

Funding

This work was supported by The Italian Ministry of Foreign Affairs and International Cooperation.

Notes

1 Montenegro is excluded from analysis because it was not an independent state when LITS was initiated, while Bosnia and Herzegovina is excluded because of complicated socio-political arrangements. For a more detailed description of EVS and LITS consult reports from GESIS (2011) and EBRD (2010).

2 To deal with data missingness multiple imputation option is not employed mainly because only a few control variables are used in regression models.

3 The comparison of 2006 and 2010 waves of LITS suggests that support for democracy increased in the former Soviet Union countries by 5 percentage points on average, whereas in EU member societies, including Baltic countries, the share of people who agreed that democracy is preferable to other political systems declined by about 10 percentage points.