Constrained from leaving or comfortable at home? Young people’s explanations for delayed home-leaving in 28 European countries

Home-leaving contexts

Young people experience very different levels of choice and constraint in their transitions from the parental home across Europe (Pohl and Walther, 2007; Van de Velde, 2008; Walther, 2006). These different home-leaving contexts likely shape their judgments about what is delaying their generation, particularly the salience of different factors like the adequacy of the housing market, youth labor market, and social policies, as well as the importance of family expectations and young people’s own inclinations. Home-leaving contexts in Europe are frequently clustered into three groups or household formation models (hereafter HHFMs) (Bendit et al., 2006) that largely overlap with the classic three-way welfare regime typology (Esping-Andersen, 1990). First, the Northern model combines relatively earlier home-leaving (early to mid-20s) with delayed family formation and is illustrated in the cases of the United Kingdom, France, Belgium, Germany, and Austria. Rental housing stock is more abundant compared to Southern Europe and is related to the younger age at which youth leave home (Iacovou, 2002: 66). School-to-work transitions occur quicker and are more flexible (in the United Kingdom) or more secure (in the continental conservative regime) compared to the extended “waithood” that exists in Southern European labor markets (Walther, 2006). Parents expect young people to leave home before marriage to learn self-sufficiency and responsibility (Holdsworth, 2004). For young people in the Northern model, domestic transitions are characterized by relative autonomy, meaning that young people are expected to leave home as soon as possible, but delays in the process are accepted as increasingly normal.

Second, the Nordic model represents an “extreme” version of the Northern model that promotes “earliest-early” (age 18 or earlier) home-leaving (Billari, 2004) and is best captured by Finland, Denmark, Sweden, and the Netherlands. These countries follow an enabling youth citizenship strategy (Chevalier, 2016) marked by high levels of social investment in youth. This cultural logic manifests in young people’s entitlement to generous public subsidies for education and training that allow for movement in and out of both spheres over the life course (Pohl and Walther, 2007). In addition, emancipation policies for youth are among the most substantial in Europe, leading some scholars to refer to Nordic home-leaving patterns as “state-subsidized experiments in independence” (Newman, 2008: 647). Youth transitions are characterized by more choice than constraint. Cohabitation levels are among the highest in Europe, and young people often move in together, as friends or as romantic partners, and explore employment and education opportunities in an environment with limited parental oversight (Billari, 2004).

Third, the Southern model (represented by Italy, Spain, Portugal, Greece, and, following Iacovou (2010), Ireland) is characterized by unappealing structural conditions for residential independence. Labor markets dominated by older generations of working men effectively prevent youth from full-time work (Chauvel and Schröder, 2014), while low levels of rental housing stock (Poggio, 2008) and strong family-based norms (Luetzelberger, 2014) help prevent home-leaving before marriage. Tight credit conditions additionally make housing markets inhospitable for new entrants (Mínguez, 2016). These countries have the highest average age of home-leaving (often late 20s or early 30s), and typical departures are illustrated by direct (or “traditional”) transitions from the family home into a new household consequent with marriage or full-time employment. Parents and youth value their relationships and time together, with home-leaving often causing significant emotional loss on both sides (Moreno, 2012). Young people’s transitions are marked more by constraint than by choice. The dominant aim for youth is to “install themselves” and “settle down” as adults in a society marked by high labor market stratification, weak public policies, and strong family ties (Van de Velde, 2008).

Given the lack of comparable data and the scarcity of analytical discussion upon which to build (Mandic, 2008; Roberts, 2003), the post-socialist countries of Eastern Europe have only recently been included in comparative home-leaving studies, but initial findings report that youth in these countries are faced with “outstandingly unfavorable” conditions in all aspects of the welfare mix, suggesting high constraint. Home-leaving occurs late, but not extremely late as in Southern Europe. Housing shortages are severe due to an overall decline in the social rented housing sector after the fall of communism and an underdeveloped private rented sector (Mandic and Cirman, 2012). Along with cultural values that promote early parenthood and extended generations living together, this exacerbates overcrowding. Furthermore, high levels of unemployment and low wages contribute to housing cost overburden (Mandic, 2008: 632). Young people’s transitions from home in these countries, as well as in Southern Europe, are characterized by extended dependence (Bendit et al., 2006).

Bearing these profiles in mind, both the normativity and meaning of “late” home-leaving vary by country. For instance, “latest-late” leaving (Billari, 2004) has been part of the dominant transition pattern in Southern Europe for generations and is generally expected by youth and their families (e.g. Mínguez’s (2016) “in-house adulthood”; Sgritta’s (2001) emancipation within the family). In contrast, late-leaving is a departure from the norm for youth in Scandinavia and to a lesser extent for youth in continental Europe and the United Kingdom (Bendit et al., 2006). While parents in Southern Europe understand the riskiness of home-leaving in current conditions and to some degree encourage longer co-residence (Holdsworth, 2005), parents in Northern and Nordic countries interpret later leaving as undue dependence on them and worry that longer stays in their home will diminish their youth’s desire to become residentially independent (Luetzelberger, 2014). The attitudes of parents, then, and their receptivity to having young adults in the house longer likely play a role in the degree to which young people see their generation as willingly staying home longer. Similarly, a probable age for “late” leaving for Nordic youth, say age 22, is still quite “early” for youth in the Southern model or even “on-time” for youth in the Northern model. This inconsistency in the semantic meaning of the phrase across Europe does not pose a problem for the current analysis, since I seek only to understand how explanations for late-leaving, whatever that may mean in a given country, can be categorized into a useful typology.

Explaining delays in home-leaving

The mixture of causes for delayed home-leaving varies considerably across the countries illustrated in the above profiles. Scholars often distinguish between structural and cultural factors (Bendit et al., 2006: 11; Billari and Liefbroer, 2007: 181) as the two main camps of determinants for the timing of home-leaving. Structural factors indicate external conditions that youth and family confront that are largely beyond their control. These factors circumscribe home-leaving opportunities available to youth and their families. Common structural factors delaying home-leaving include a lack of housing (either rental or owner-occupied), housing cost overburden, rigid mortgage markets and young people’s lack of access to credit to enter these, segmented labor markets that prevent youth from stable employment, low wages for those who are employed, and minimal or no state support for youth emancipation (Golsch, 2003; Luetzelberger, 2014; Mandic, 2008; Moreno, 2012; Newman, 2008). Studies emphasizing such factors tend to note the cumulative effect of each of these on the overall age at leaving home. As more of these conditions converge, the less feasible earlier home-leaving seems.

Cultural factors involved with the timing of home-leaving include historically rooted norms surrounding the maintenance of certain living standards, the perceived quality of the parental home, and the warmth and openness of the parent–child relationship (Giuliano, 2007; Seiffge-Krenke, 2010). Particularly in Southern Europe, such norms can contribute to household income’s positive relation to the age at leaving home (Iacovou, 2002). If parents are willing to let their children remain at home longer, the lower the incentive for youth to move out to a new household that is likely less comfortable (Aassve et al., 2001: 8). Non-transferable material resources, like adequate space in the parental household and the preparation of meals, are associated with delays in leaving home (de Jong Gierveld et al., 1991). Parents who create a “feathered nest” consisting of high material and consumption standards can undermine youth aspirations for independence, especially when these conditions are combined with parents’ desires to keep youth nearby or conforming to their expectations (the “gilded cage”). Indeed, depending on their priorities, youth may elect to stay in the parental home indefinitely provided that material well-being remains high and their parents allow them a measure of increased autonomy (Avery et al., 1992; Giuliano, 2007). Cultural norms also include notions of religiosity, family responsibility, and obligation that encourage mutual support between kin and intergenerational solidarity that involves providing welfare for young, elderly, ill, or otherwise vulnerable family members (Luetzelberger, 2014; Moreno, 2012).

Few studies examine young people’s beliefs about how these two sets of factors matter in determining home-leaving. Iacovou (2002) and Seiffge-Krenke (2010) both find evidence that in Southern European countries, cultural factors based on family responsibility and proximity are perceived to be more important in determining the timing of departure than structural issues associated with the housing, education, or labor market. Examining Italian youth, Sgritta (2001: 76) finds that age is strongly associated with offering subjective preferences as the main reasons for living at home. As young people age, their explanations for remaining at home shift from emphasizing structural obstacles to more “convenience-based” and discretionary reasons. Thus, some initial evidence suggests that cultural or subjective factors might be interpreted in Southern Europe (where home-leaving tends to be late) as having increased importance in delaying leaving.

Sgritta (2001) also examines cross-national Eurobarometer data from 1997 on the issue and offers a descriptive analysis, concluding that “the impression one receives from the replies is that, both from a material point of view and from a sentimental one, there are no clear-cut differences between the Southern European countries and the rest (p. 79).” Billari (2004) uses some of the current data to briefly examine 15 EU countries’ percent of agreement with three different reasons for why youth are living longer at their parents’ home: youth cannot afford to move out, not enough suitable housing, and parents are less strict nowadays than in the past. His conclusion is the opposite: young people’s reasoning about why they take longer to leave home accords with the opposite patterns of “latest-late” and “earliest-early” leaving and the cultural and structural factors that drive them. Affordability problems were frequent in Sweden, where youth are relatively more subsidized in their push for independence. Housing problems were mentioned most frequently in Denmark and the Netherlands, where the state provides relatively more housing support. More relaxed parenting was cited somewhat equally over the continent.

I build on these prior works by utilizing a more rigorous statistical technique that approximates the most common response patterns and therefore the different “types of thinking” about this issue. Moreover, I examine how the prevalence of these views is related to certain institutional aspects of each country, with the hypothesis that key aspects of the education, labor, or housing markets might help determine the likelihood on the national level of “cultural” or “structural” explanations. The variation in young people’s types of explanations for delayed leaving also might resemble the cross-national distinctions already found in adult samples for explanations of social problems, in terms of whether blame is assigned to larger structural forces circumscribing the individual or the individual’s own cultural choices.

How people understand social problems

Two bodies of literature that can potentially help explain differences in the interpretation of late home-leaving (or of delayed transitions in general) are the “attitudinal worlds of welfare” (AWW) theory developed by European scholars to explain attitudes about social problems and the “modes of explanation” concept developed by American scholars to explain different racial groups’ explanations for social inequality. Distinct AWW stem from different welfare state arrangements and the underlying logic of the principles that uphold these arrangements (Arts and Gelissen, 2001). A small subset of this literature examines how people interpret responsibility for social problems, specifically whether they hold the state and structural flaws at fault or whether they ascribe responsibility to individuals and their choices (Örkény and Székelyi, 2000). These rationales for “negative” outcomes, in particular their ascriptions of the causal reasons behind them, are linked to larger, more complex sets of interlocked attitudes that stem from two competing, core value frames at the national level: those emphasizing market-based values and those emphasizing state-based values (Sabbagh and Vanhuysse, 2006).

Countries in the liberal welfare regime, such as the United States, Canada, Australia, and New Zealand, typically employ a market-based value frame to explain poverty, unemployment, and other social problems. Market-based frames direct attention away from collective responsibilities for justice and toward ill-advised individual choices, justifying minimal state involvement in rectifying social ills. People attribute social problems more frequently to personal issues, like the failure to take advantage of opportunities, laziness, lack of talent, or loose morals (Hasenfeld and Rafferty, 1989). This is referred to as an internal attribution. Conversely, the class coalitions and long-term power structures of countries such as France, Germany, Sweden, Finland, and the Netherlands have developed state-based frames to explain social inequalities and justify state interventions. In these nations, social and economic problems are more likely to be seen as a consequence of larger social forces like prejudice, limited opportunities, bad schooling, and exploitation (Svallfors, 1993); this is known as an external attribution.

These opposing attributions also appear in studies of Americans’ “modes of explanation” for disparities between social groups, like the Black–White income gap (Apostle et al., 1983; Hunt, 2007; Kluegel, 1990). Authors operationalize these modes or types of explanations based on the joint configuration of endorsements of different causes for Blacks’ economic situation compared to Whites. In explaining Black–White inequality, Whites historically tended to favor reasons like motivation or innate ability, while Blacks tend to cite causes like discrimination and unequal chances (Kluegel, 1990). Those who are part of the group affected by the problem are more likely to offer structural explanations for why it occurs, shifting attention away from possible individual faults. Conversely, those who are unaffected by the problem tend not to see these structural causes and are more likely to attribute its occurrence to individual choices. In the case of late home-leaving, similar dynamics (degree of affectedness) might indeed be at work in shaping explanations. For instance, those who have left already and at a normative age might be more likely to offer individualist or cultural reasons for why others cannot do the same, while those who are still home despite wanting to move out might be more likely to offer the structural explanation of poor opportunity. In this study, since my data do not allow me to control for people who have left home already, I examine this potential mechanism at the country level by seeing whether explanations of late-leaving are associated with the nation’s relation to late-leaving, operationalized as the proportion of 18–34 year olds currently living at home.

Hunt (2007) notes that most research on explanations for inequality tends to make a general distinction between beliefs that locate reasons for this in either the personal attributes of individuals or features of the broader social structure. However, he also notes that dichotomies of these explanatory strategies are not always appropriate, since seemingly inconsistent or contradictory beliefs can be combined into “compromise” explanations (Lee et al., 1990). Therefore, analyses of explanations for negative outcomes should allow for “mixed mode” response patterns (Thompson and Bobo, 2011) that do not assume unicausality in one’s beliefs about the root causes for a given phenomenon.

Current study

As noted, few studies examine how youth perceive their transitions out of the parental home to be hindered for certain reasons. Yet, there are important reasons to expect national-level differences in causal explanations for the relatively increased time to leave home. First, wide variation exists across Europe in the structure of the educational system, the labor market, the linkages between the two, and perhaps most important, the housing market and the type of preferred tenure (renter or homeowner) (Kemeny, 1981). For instance, it would not be surprising to see clear differences between countries in external attributions of housing deficits, since rental housing is abundant in some countries and very sparse in others. Second, as noted, in some countries, the population is socialized to interpret social problems (of which late home-leaving might be a particularly youth-relevant example) as due to personal attributes of people, whereas in other societies, these systemic problems are seen as due to larger structural forces constraining the individual. This article adds to existing studies comparing adults’ modes of explanation for social problems by examining how these modes appear to youth. Evidence suggests a degree of crystallization on basic social values by the time adolescents reach young adulthood as a result of various forms of political socialization (Sabbagh and Vanhuysse, 2010).

I address two primary research questions in this article. First, what are the different modes of explanation young people employ for late home-leaving, and can these be sorted into individualist (internal attribution) and structuralist (external attribution) groups similar to explanations for other social problems? Second, what are the institutional aspects of countries associated with either structuralist or individualist explanations? Correlations at the national-level will help detect whether there are certain institutional and demographic “components” (House, 1981) linked with the prevalence of these views.

Data

I use data from two Eurobarometer surveys that were administered 2 years apart (2001 and 2003). Both surveys gauge youth (age of 15–24 years) opinions on reasons for young adults living at home with their parents. The substantive questions asked were identical across the two surveys, which led to the creation of a combined dataset that matched the equivalent variables. The first survey, conducted in 2001, took place during the fieldwork for Eurobarometer 55.1: Globalization and Humanitarian Aid, which targeted adults only (April to May 2001). An oversample of young respondents aged 15–24 years (Eurobarometer 55.1 OVR: Young Europeans; Christensen, 2010) was added to the basic sample in order to obtain a total of at least 600 young respondents per country. The usual 15 Eurobarometer countries were surveyed: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom. After 2 years in March 2003, another Eurobarometer survey, “Candidate Countries Eurobarometer 2003.1 YouthOVR: Youth in New Europe,” queried youth in the 13 candidate countries that had applied for membership in the EU: Bulgaria, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Romania, Slovakia, Slovenia, and Turkey (Papacostas, 2010). I loosely include Malta, Cyprus, and Turkey as part of the Southern HHFM in analyses, since these countries do not have a history of communist rule.

These are the most recent and comprehensive data on youth interpretations of late home-leaving. Despite potential issues regarding the time lag between the early 2000s and the present day, recent research on trends in home-leaving shows that countries’ home-leaving contexts are largely path-dependent, meaning that the current data, albeit aged, are likely still valid and can potentially speak to the current variation in interpretations nowadays. “[T]he presence of multiple equilibria and path dependence imply a much stronger stability of long-term differences (i.e. based on long-term family models or on institutional settings), independently of the convergence in terms of other factors” (Billari, 2004: 29). Because home-leaving “may be rooted in systems that are consistent and particularly resisting to changes and common trends” (Billari, 2004: 32), the data should be useful in understanding where and for whom the delay in home-leaving is attributed to structural faults as compared to individual preferences.

Eurobarometer polls employ multi-stage random probability sampling designs that attempt to represent the whole territory of the respective country while gathering a representative picture of the opinions of that country (Inglehart and Reif (2016); see Christensen (2010) and Papacostas (2010) for more information on Eurobarometer survey design and sampling). The oversample collected in each nation covers the population of the respective nationality aged 15–24 years. In the process of collecting the oversample, survey administrators imposed overriding controls to verify the sample based on sex and age. All interviews were conducted face-to-face in the home and in the appropriate national language. The final sample for analysis consisted of respondents who had valid answers to the item serving as the dependent variable (n = 18,440). I combine within-nation weights from both surveys using a “weight the weights” procedure (Korn and Graubard, 1999) and utilize these in the analysis.

I take advantage of a relatively young sample, most of whom have not left home already and are in varying stages of planning their departure. Thus, the sample provides a depiction of how especially young Europeans, such as students in secondary school, perceive the delay in home-leaving. Youth plan their futures around the act of leaving home and likely have some impression formed already of when and how they would like to leave and how hospitable the external conditions are for potentially earlier leaving. Based off their social network and exposure to older family members or peers, they might have some degree of knowledge about how the moving out process occurred for some of them, including when it occurred and for what reasons. Young people still living in the parental home may or may not be knowledgeable of the actual extent of delayed leaving in their region, but the primary point of the article is not to see how “accurate” youth interpretations are. Rather, the objective is to assess what extent European youth assign the same causes to the phenomenon of “late” home-leaving, whether the dominant modes of explanation for this can be classified into structuralist and individualist explanations, and if institutional and demographic conditions are related to the dominance of some types of views over others.

Variables

The dependent variable in this study is a question asking about perceived reasons why youth are staying at home longer. Respondents were asked, “Some people say that nowadays, adolescents and young adults tend to live in their parents’ homes longer than they used to. What do you think are the three main reasons for this?” The respondent could pick up to three from eight possible answers:

This question was asked only to the oversample of youth aged 15–24 years. Respondent selections of these answer choices were not weighted or ranked, only mentioned or not mentioned. In addition, the phrasing of the question supports the interpretation that respondents are reporting a generalized sense of why the age at leaving home has increased and not about their own personal situation, although there may be some overlap between the two. These responses range from financial concerns to changing cultural norms to housing deficits and cover a wide spectrum of possible reasons for late home-leaving.

This article is concerned with the direction of “national opinion” regarding late home-leaving. As such, although differences in interpretations of this phenomenon undoubtedly vary significantly by individual-level factors like age, gender, education, income, and having left home already, I refrain from any individual-level analysis. In the current data, respondents make an overall statement about others in general without specifying their own situation. Thus, an individual’s answers do not necessarily hold for the respondent themselves. Similarly, theories about home-leaving or interpretation of social problems do not offer any clues as to why individuals who are married, highly educated or female should interpret home-leaving behavior any differently than their referent groups. To avoid concluding from personal circumstances to the behavior of others, I maintain a country-level focus of analysis.

The independent variables in this study are nation-level indicators for important aspects of the home-leaving context that may help explain why some countries have substantially different frequencies of structuralist or individualist interpretations. To best re-construct what the actual conditions would have been at that time, these values are predominantly drawn from Eurostat for the years 2001 or 2003 (depending on the country’s survey year), or as close to these years as possible (see Appendices 1 and 2 for specifics on all data used in the analysis). For two indicators, I consulted the 1999 wave of the European Values Survey (EVS, 2011). Indicators were selected according to two overall criteria. First, they had to be recognized in the comparative literature on home-leaving as having an impact on the timing of leaving. Second, they had to be available for both the original member states as well as the candidate countries. This latter specification significantly narrows the choice of variables since many new member states lack reliable data prior to 2004. Following Mandic (2008: 625), I organized the most relevant structural factors into five domains: the demographic/historical factor, the robustness of the labor market, robustness of the housing market, the extent of welfare state involvement, and the extent of family in welfare production. In addition, I used an indicator for the extent of tertiary educational opportunities, since routes out of the parental home nowadays are often linked to attending higher education. I also incorporated a proxy measure of the extent to which a country used a state-based or market-based frame. I looked for one or two reliable indicators that would approximate these concepts in all countries considered in the analysis. Still, some variables were not available for all states. I note the slightly limited number of countries in some of the correlation analyses where appropriate.

For the demographic/historical factor, I used mean age of women at childbirth. This is a commonly used proxy for family formation dynamics. For the country’s extent of late home-leaving, I used the percent of young people aged 18–34 years living with parents. For the health of the labor market, I used both youth unemployment rates for the population aged 15–24 years and overall 3-year averaged unemployment rates for the population aged 15–74 years. In anticipation that young people’s responses might be especially related to the housing market, I used three measures for this domain. First, to gauge which countries were homeowner societies as opposed to renting societies (Ronald, 2008), I used the percent of the population at least 18 years old living in rented housing. Second, I used the overcrowding rate for youth aged 15–24 years and the housing cost overburden rate, which refers to the percent of the population spending 40 percent or more of their disposable household income on housing expenses. This approximates the relative availability and affordability of housing in a nation. For the extent of state involvement in welfare production, I used expenditures on social protection as a percent of gross domestic product (GDP) and housing benefits as a percent of GDP. For the importance of family in welfare production, I consulted a measure from the EVS asking respondents how willing they would be to do something to improve the conditions of their immediate family; I use the proportion of respondents in a nation saying they would “absolutely” act in order to help. In addition, I included Eurostat’s indicator on the percent of households reporting contact with family or relatives every day. This serves as a general proxy for the relative importance of family in everyday life. For the robustness of tertiary education, I used the total public expenditure on tertiary education as a percent of GDP and the tertiary education participation rate. For measuring whether national populations used market-based or state-based frames, I examined a measure from the EVS asking respondents to rate whether the individual should take more responsibility providing for themselves or whether the state should take more responsibility in ensuring that everyone is provided for. This variable ranges from 1 (individuals) to 10 (the state); I computed the mean value for each country. Finally, I included some typical “control” variables when comparing youth populations across countries: the size of the youth population and GDP per capita.

Analysis technique

To assess what the different modes of explanation are for late home-leaving, I first construct a typology of beliefs about delayed leaving. Compared to a technique like factor analysis that is more concerned with the underlying relations between variables (i.e. correlations), a typology focuses on the structure of cases (clusters of subtypes) and the way they appear empirically. I employ latent class analysis (LCA) to model the most common response patterns observed in the data and observe whether these response patterns resemble structuralist, individualist, or mixed-mode explanations. LCA is an empirically based statistical method for explaining heterogeneity in response patterns in terms of unobserved, underlying classes that are distinct from one another, yet are internally homogeneous (Lanza and Rhoades, 2013). Second, to address how these modes relate to country-level conditions, I examine bivariate correlations at the national level between institutional factors associated with aspects of the home-leaving context and the prevalence of these explanatory modes.

Latent class models are a popular modern alternative to cluster analyses, attempting to illustrate the association between a given phenomenon (in this case, categorically different attitudes about why late home-leaving occurs) and its observed empirical indicators through a finite number of underlying patterns. As a data reduction technique, it identifies distinct subgroups of cases based on how they respond to multiple categorical indicators (Collins and Lanza, 2013). For a given case, LCA estimates two parameters: the overall probability of being in a given class (posterior class probability or latent class membership probability) and the probability of responses to a set of categorical indicators conditional on class membership (item-response probabilities). The item-response probability parameters capture the association between the observed items and the latent classes and are the basis for interpreting what the latent classes mean. Individuals are assigned to their most likely class based on the class for which they have the highest posterior probability of being in, which stems from their individual response pattern (Muthén, 2001).

I first determined the underlying structure of the data by estimating models with different numbers of classes. Each model was estimated with only the necessary information to construct the classes (the eight different binary reasons youth could have mentioned) and did not use covariates such as age, gender or education, since these are unrelated to the current research question and also can complicate model identification (Clark and Muthén, 2009). To make sure that each class solution was the maximum likelihood solution for that number of classes, 100 versions of each model were run using random sets of starting values. G² values (likelihood ratio chi-square statistics) were compared across the 100 models. The modal solution most frequently resulting in the same G² value was identified as the maximum likelihood solution.

In Table 1, I compare measures of model fit for latent class solutions for two through six classes based on different information criteria, including log-likelihood values, G², Akaike information criterion (AIC), CAIC (a consistent version of AIC), Bayesian information criterion (BIC), sample-size-adjusted BIC, and entropy R². Entropy R² values in all models are quite high, with over 95 percent of respondents being correctly classified into their most likely class, which supports using the latent class variable as an observed variable, since it is highly accurate. Although model selection based on BIC and G² values is typically recommended, I also considered pragmatism and past research in selecting the final model. In particular, a “good” model meant that the solution could be readily interpreted, the classes could be easily labeled, showed minimal overlap and redundancy, and agreed closely with prior findings (Lanza and Rhoades, 2013). I judged a three-class solution to be the most appropriate, striking a balance between parsimony and redundancy.

OPEN IN VIEWER

Table 1.

Latent class solutions.

N classes	LL	df	G 2	AIC	CAIC	BIC	Adjusted BIC	Entropy R2
2	−79,343.9	238	14,564.8	14,598.8	14,748.8	14,731.8	14,677.8	0.997
3	−78,483.3	229	12,843.6	12,895.6	13,125.0	13,099.0	13,016.3	0.950
4	−77,734.1	220	11,345.2	11,415.2	11,723.9	11,688.9	11,577.7	0.972
5	−77,157.6	211	10,192.1	10,280.1	10,668.3	10,624.3	10,484.5	0.979
6	−76,771.0	202	9418.9	9524.9	9992.5	9939.5	9771.1	0.976

LL: log-likelihood; AIC: Akaike information criterion; CAIC: consistent version of AIC; BIC: Bayesian information criterion.

Based on pseudo-likelihood since weights and clusters (nation-level) are used. All models have 18,440 cases.

In separate analyses available from author on request, I explored the effect of using a collapsed range of answer categories and running analyses within groups of countries and found that the same general typology emerged. This three-class solution is highly robust across a majority of countries. The structure of resulting typologies from these robustness checks was similar, but not as easily interpretable as the “full information/big N” model. All models and underlying data are available from author upon request.