Psychometric evaluation and measurement equivalence of the Adolescents’ Societal Belongingness Scale (ASBS)

ABSTRACT

Societal belongingness – feelings of being a connected, an affiliated, and a respected member of the larger society – may contribute to the understanding of adolescent development both as a person and a member of society in a multicultural context. The current study examined the psychometric properties and measurement invariance of the Adolescents’ Societal Belongingness Scale (ASBS) using data from 12- to 15-year-old youth (N = 1065, M_age = 13.12, SD = .42, 45% female). Multiple group CFA models revealed configural, scalar, and metric invariance of the societal belongingness scale across adolescents with immigrant, mixed, and Swedish backgrounds. The scale scores were positively correlated with class belonging and social trust and negatively correlated with depressive symptoms and delinquency. Between-group differences across immigrant, mixed-heritage, and native youth provided further evidence for the sensitivity of the measure. Overall, the current findings suggest that the societal belongingness scale could be used for studying the sense of belonging towards the larger society across adolescents of immigrant, mixed, and majority backgrounds.

KEYWORDS:

• Societal belongingness
• the need to belong
• sense of belonging
• social belonging
• adolescence

Developing a sense of connection and belonging to the surrounding social groups is an important part of adolescents’ identity formation process (Erikson, 1968) and a key predictor of mental health outcomes (Hagerty et al., 1992). Social connection and belonging in adolescence, however, has been primarily examined with reference to national, ethnic, or cultural groups or to school and classroom contexts. There has been very limited focus on understanding how young people see themselves in relation to the larger society. Due to globalization and migration, adolescents develop and prepare for adulthood in increasingly diverse social environments. Thus, studying how adolescents define their self in the larger societal context that is characterized by ethnic and cultural heterogeneity is at least as equally important as studying their sense of belonging to their own cultural, ethnic, or national group, all of which refer to a homogeneous group. Hence, a strong ingroup identification may facilitate distancing from (Allen, Gray, et al., 2021) or negative treatment of the members of the outgroups (Perreault & Bourhis, 1999). On the other hand, feelings of being a connected, affiliated, and accepted member of the larger society – a strong sense of societal belonging – may facilitate open and harmonious relationships with others in diverse societies. Nevertheless, to date, there has been limited research on adolescents’ feelings of belonging and connectedness to the larger social context. The current study, to facilitate scientific inquiry in this field, reported on the psychometric qualities and validity of a measure developed to assess adolescents’ feelings of societal belongingness.

The correlates, predictors, and outcomes of belonging have been widely studied across different disciplines and age groups; however, there has not been a consensus on measurement of this construct (Allen, Kern, et al., 2021; Mahar et al., 2014). A review of commonly used measures identified both common characteristics and major differences across the sense-of-belonging scales (Mahar et al., 2014). First, most scales focus on individuals’ perceptions of their connection to others or specific groups. Relationships, friendships, and bonding are the frequently used terms in the items. Several studies use single-item measures to assess sense of belonging to school (Napoli et al., 2003) or to the country (Hou et al., 2018) or include items referring to the perceived reaction from others, such as ‘people here notice when I am good at something’ (Goodenow, 1993). In sum, there is variation in the available approaches to the measuring of sense of belonging.

The current study proposes a measure to assess adolescents’ sense of belongingness to a larger society. Baumeister and Leary have recently acknowledged the importance of focusing on ‘belongingness to groups rather than emphasizing pair-bond relationships’ to advance the field of research (Allen, Gray, et al., 2021, p. 7). However, belonging to a group may also encourage detachment from other groups (Allen, Gray, et al., 2021). For example, a strong sense of belonging to their own ethnic group may lead adolescents to distance themselves from other ethnic groups, which may pose a threat to an individual’s social and emotional adjustment as well as inter-ethnic relationships and societal cohesion in today’s increasingly diverse societies.

Accordingly, we propose that a focus on societal belongingness – feelings of being a connected, an affiliated, and a respected member of the larger society – may contribute to our understanding of adolescent development both as a person and a member of society in a multicultural context (Özdemir & Bayram Özdemir, 2021). The sense of belonging implied in societal belongingness is concerned with perceived connection, affiliation, and acceptance with reference to the superordinate level of society, which might be composed of people of diverse ethnic, cultural, and religious backgrounds, and does not undermine either the existence or importance of individuals’ feelings of belonging to other social groups. In addition, Yoon and colleagues (Yoon et al., 2012) showed that social connectedness to mainstream American society – a similar construct to societal belongingness – had significantly positive but low correlation with connectedness to ethnic community among both Mexican American and Asian international college students (r = .14 and .28, respectively). These findings suggest that feelings of belonging to the larger society and own ethnic community, at least in cultural groups, can co-exist, but they seem to be different constructs. Similarly, societal belonging is distinct from the sense of relational belonging. Specifically, Valcke and her colleagues (Valcke et al., 2020) showed that societal belonging positively predicted social trust and social acceptance among ethnic minority adults. On the other hand, sense of relational belonging was not related to neither of these two outcomes. Overall, existing evidence suggests that sense of societal belonging is distinct from belonging to an ethnic or cultural group and is a better predictor of trust in and perceived acceptance by the members of the larger society at least in ethnic minority adult samples. Yet, there is a dearth of evidence regarding the adolescents’ sense of societal belongingness.

Societal belonging differs from similar concepts like national belonging that often refer to a majority group or a country. Measures of national belonging make explicit reference to the majority ethnic group and often include egocentric references (e.g., when Israelis are criticized, I take it personally) (Tartakovsky, 2009) and patriotic views (e.g., I am proud of the Netherlands) (van Vemde et al., 2021). In fact, a strong Dutch identification among 12–21-year-old youth was found to be positively correlated with perceived threat from others, ingroup superiority, distance to other people and positive attitudes towards right-wing violence, and own violent intentions among native Dutch youth (Doosje et al., 2012). On the other hand, feelings of being a connected, an affiliated, and a respected member of the larger society without any reference to the social, ethnic, cultural, and religious groups may contribute to more positive attitudes towards other people.

In the current study, we evaluated the psychometric properties of the societal belongingness scale. We tested the factor structure of the scale and explored measurement invariance across youth with immigrant, mixed-heritage, and majority background. In addition, we examined concurrent validity of the scale by inspecting its associations with class belonging, social trust, depressive symptoms, and delinquent behaviours in line with the previous evidence (Anderman, 2002; Valcke et al., 2020) and the arguments suggesting that feeling connected to mainstream society would be positively related to general feelings of social connectedness (e.g., Yoon et al., 2012). We expected positive associations between societal belongingness, class belonging and trust, whereas we expected negative associations between societal belongingness and depressive symptoms and delinquency.

Methods

Participants

The data were drawn from an ongoing three-year longitudinal study, Youth & Diversity Study, which collected data from seventh-grade students in 55 classrooms in 16 different schools, which were strategically selected to represent the local population characteristics, in four medium-sized Swedish cities. The target sample included 1286 students, and 83% participated in the study. Seventeen per cent were either missing on the day of data collection or their parents did not give consent or they themselves declined participation in the study. The analytic sample was composed of 1065 adolescents (M = 13.12, SD = .42, 45% female). Overall, 27.7% of the adolescents had both parents born abroad (immigrant origin), 12.8% had one parent born abroad (mixed heritage), and 59.4% had both parents born in Sweden (Swedish). Immigrant parents originated from 60 different countries. The largest three groups were from Syria (16% of mothers and fathers), Somalia (13% of mothers and fathers) and Iraq (10% of mothers, 12% of fathers). Fifty-eight per cent of the adolescents of immigrant background were born in Sweden and 42% were born abroad. Majority of the parents (83% fathers; 72% mothers) were employed. Seventy-one per cent of the adolescents lived in intact families. The research protocol was approved by the Regional Research Ethics Committee in Uppsala.

Measures

Adolescents’ Societal Belongingness Scale (ASBS)

The societal belongingness scale was developed as part of the Youth & Diversity Study. The scale items were adapted from the school belonging measure for adolescents (Cemalcilar, 2010; Neel & Fuligni, 2013). One of the items (i.e., I am glad to be a student in my school) was dropped as it does not refer to sense of belonging but refers to perceived satisfaction. The original items were revised so that the respondents could rate their sense of belonging to the larger society. Because the data was collected in Sweden, the five scale items referred to how much youth feel connected to, appreciated by, and being part of the Swedish society, which refers to the diverse society that comprises the population in Sweden (see Table 1). The items were rated on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). Cronbach’s alpha for the scale was .85.

Table 1. Summary descriptive statistics, factor loadings, and ICCs for the items of the Adolescents’ Societal Belongingness Scale (ASBS).

Item	Standardized factor loading	ICCs	M	SD	Skewness	Kurtosis
1. I feel that I belong to Swedish society	.76	.033	4.20	0.97	−1.28	1.42
2. I feel that I am an outsider here in Sweden (Reversed)	.45	.026	4.21	1.05	−1.37	1.28
3. I think that I am part of Swedish society	.91	.029	4.17	0.96	−1.15	1.00
4. I feel that I am connected to Swedish society	.75	.050	3.93	1.09	−0.82	−0.07
5. I feel appreciated in Swedish society	.74	.037	3.97	1.04	−0.95	0.48

Class belonging

Class belonging was measured using 6 items (Cemalcilar, 2010; Neel & Fuligni, 2013). Adolescents were asked to rate each item on a 5-point Likert scale (1 = strongly agree, 5 = strongly disagree). Sample items included ‘I feel like I am part of my class’ and ‘I feel like I am a valued member of my class.’ Adolescents’ responses were reverse coded and averaged to display the scale score. Cronbach’s alpha for the scale was .88.

Social trust

Social trust was measured using the two-item measure of interpersonal trust (e.g., ‘Most people can be trusted,’ ‘Most people are fair and do not take advantage of you’ (Flanagan et al., 2007; Flanagan & Stout 2010). The items were rated on a 5-point Likert scale (1 = do not agree at all, 5 = agree very much). Cronbach’s alpha for the scale was .79.

Depressive symptoms

Depressive symptoms were measured using the 16-item version of the CES Depression scale (Falstich et al., 1986) that has shown good reliability and validity across studies (Bayram Özdemir & Stattin, 2014). Participants rated the items on a 4-point Likert scale (1 = not at all, 4 = very often) to indicate the frequency of the listed experiences during the past week (e.g., ‘I felt down and unhappy,’ ‘I was bothered by things that usually do not bother me’). The mean of responses was calculated to construct a scale score. Cronbach’s alpha for the measure was .95.

Delinquent behaviours

Adolescents’ involvement in delinquent behaviours during the past year was measured using a scale measure with documented validity when compared to official reports (Stattin et al., 2010). The scale items included 9-items (e.g., ‘Taken a bicycle without permission,’ ‘Participated in a fight at school or in town’) that were rated on a 5-point Likert scale (1 = No, it has not happened, 5 = More than 10 times). Cronbach’s alpha for the scale was .90.

Analytic strategy

First, a single-factor CFA model was fitted to test the factor structure of the scale. The data were collected from 55 classrooms. ICCs for the items on the societal belongingness scale ranged between .026 and .050 with an average of .035, and the design effect ranged between 1.44 and 1.86 (Hox, 2010). Thus, the maximum likelihood robust estimator and the Type=Complex option in Mplus 8.7 were used to account for nonnormality and stratification in the data (Muthén & Muthén, 1998–2017). We also tested if ASBS referred to a related, independent, or the same construct as a class belonging measure. Next, multiple-group CFA models were fitted to test configural (same factor pattern), metric (same factor loadings), and scalar invariance (same intercepts) (Putnick & Bornstein, 2016). We inspected multiple fit indices to evaluate model fit including model chi-square, comparative fit index (CFI ≥ .95), Tucker–Lewis index (TLI ≥0.95), root mean square error of approximation (RMSEA ≤ .06), and standardized root mean square residual (SRMR ≤ .08) (Kline, 2015). In evaluating the measurement invariance models, we first expected Δχ² being non-significant at p > .05. According to Chen (2007), a decrease of .010 in CFI (the main criteria) supplemented by a decrease of .015 in RMSEA or .030 in SRMR for metric invariance (.010 for intercept) suggest non-invariance in samples where N > 300. The data included missing information on some variables ranging between 1.8% and 7.2% (average missingness M = .04, SD =.13). All models were fitted to the full data using FIML approach. We estimated the correlations of the societal belongingness measure with the criterion variables (i.e., class belonging, social trust, depressive symptoms, and delinquent behaviours) to examine the concurrent validity of the proposed scale. The correlations estimated across groups were compared using Fisher’s z-test (Eid et al., 2017). We estimated eta-squared (η²) and Cohen’s d effect sizes for group differences and interpreted following Cohen’s (1988) guidelines (i.e., eta-squared .01, .06, .14 and Cohen’s d .20, .50, .80 for small, moderate, and large effect sizes).

Results

The fit indices for the single-factor CFA model suggested an acceptable fit, CFI =.97, TLI =.94, SRMR =.024, despite a significant chi-square test, χ²(5) = 36.54, p < .001, and a mediocre RMSEA value, RMSEA =.072, 90% CI: .056, .105. Modification indices suggested adding a correlated error between two of the items (i.e., ‘I feel that I am connected to Swedish society’ and ‘I feel appreciated in Swedish society’). It has been argued that feelings of belonging and connection to a group imply recognition and acceptance by other members of the group (Hagerty et al., 1992). Thus, a modification was plausible and it improved the model fit significantly, Δχ² (1) 37.83, p < .001, and yielded improved fit indices, χ² (4) 13.99, p = .007, CFI =.99, TLI = 0.98, RMSEA =.05, 90% CI: .023, .080, SRMR =.016 (Table 2). The standardized factor loadings ranged between .45 and .91 with an average of .72. Cronbach’s alpha and omega square reliability estimates were .85 for the overall sample.

Table 2. Results from CFA and Measurement Invariance (MI) models testing invariance of the Adolescents’ Societal Belongingness Scale (ASBS) across immigrant background.

	χ^2 (df)	CFI	TLI	RMSEA [90% CI]	SRMR
CFA in full sample	13.99* (4)	.990	0.980	.050 [.023, .080]	.016
MI models across immigrant background, mixed-heritage, Swedish background
	χ^2 (df)	CFI	TLI	RMSEA [90% CI]	SRMR	Δχ^2(df)	p	ΔCFI	ΔRMSEA	ΔSRMR
Configural invariance	32.27*** (12)	.977	0.943	.072 [.042, .102]	.022
Metric invariance	44.85*** (20)	.972	0.958	.062 [.037, .086]	.065	13.53 (8)	.095	.005	.010	.043
Scalar invariance	56.21*** (28)	.968	0.966	.055 [.034, .076]	.071	10.12 (8)	.267	.008	.007	.006

*p < .05, **p < .01, ***p < .001.

We fitted three models including the items of the ASBS and class belonging measure to test if societal belonging measure refers to a construct that is different from class belonging. The single-factor model, χ²(42) = 874.20, p < .001, and the independent factors model (zero expected correlation between latent factors), χ²(52) = 284.50, p < .001, did not fit the data. The related, two-factor model yielded acceptable fit indices, χ²(41) = 171.05, p < .001; CFI =.96, TLI =.95, RMSEA =.055, 90% CI: .047, .064, SRMR =.037. The latent factors for societal belonging and class belonging had a significant positive correlation, r = .45, p < .001, suggesting that these two constructs were related, but referred to as different constructs.

The results from the measurement invariance models across immigrant background are presented in Table 2. Relative fit indices suggested acceptable fit of the configural invariance model across immigrant, mixed-heritage, and Swedish adolescents, χ²(12) = 32.27, p < .001, CFI =.98, TLI = 0.94, RMSEA =.072, 90% CI: .042, .102, SRMR =.022 despite the mediocre RMSEA. When additional constraints were introduced to test the invariance of factor loading (metric invariance) and intercepts (scalar invariance) across the groups, chi-square difference tests suggested that the more restrictive models were equivalent to the less restrictive models, suggesting that criteria for configural and metric invariance were met. The changes in CFI, RMSEA, and SRMR (except for ΔSRMR for metric invariance) were also in line with criteria suggested by (Chen, 2007).

To evaluate whether the levels of societal belongingness differ between immigrant, mixed, and Swedish adolescents, we compared the mean levels across these three groups. The results revealed that all three groups significantly differed from each other, with moderate-to-large effect sizes, F(2, 970) = 82.21, p < .001, η² = .15 (immigrant background, M = 3.58, SD = 0.87; mixed-heritage, M = 4.07, SD = 0.69; Swedish background, M = 4.31, SD = 0.69). In addition, there was a statistically significant difference with a small-to-moderate effect size in the societal belongingness scores of immigrant adolescents who were born abroad (M = 3.42, SD =.83) and born in Sweden (M = 3.77, SD =.89), t(245) = 3.18, p = .002, Cohen’s d = .41 (Table 3).

Table 3. Differences in Adolescents’ Societal Belongingness Scale (ASBS) scores across subgroups.

	M (SD)	F/t-test^1 (df)	p	η^2/Cohen’s d
Background differences		82.21 (2, 970)	<.001	.15
Immigrant background	3.58^a (0.87)
Mixed heritage	4.07^b (0.69)
Swedish background	4.31^c (0.69)
Generation differences		3.18 (245)	.002	.41
Immigrants born abroad	3.42 (0.83)
Immigrants born in Sweden	3.77 (0.89)

^aDifferent superscripts refer to statistically significant differences between the groups based on Bonferroni post-hoc test.

^bF-test was used to compare means for groups with three levels and independent samples t-test was used to compare means with two levels.

To examine concurrent validity, we estimated the correlations of the societal belongingness scale with class belonging, social trust, depressive symptoms, and delinquent behaviours (Table 4). The results suggested that societal belonging was positively correlated with both class belonging and social trust. On the other hand, societal belonging was negatively associated with adolescents’ depressive symptoms, and engagement in delinquent behaviours. Fisher’s z-test to compare the correlations across groups yielded no statistically significant differences. That is, the findings held to the same degree among immigrant, mixed-heritage, and Swedish youth.

Table 4. Bivariate correlations of the societal belongingness scale scores with class belonging, social trust, depressive symptoms, and delinquency for the full sample and subgroups.

		Immigration background
	Full sample	Immigrant	Mixed-heritage	Swedish
Class belonging	.40***	.40***	.43***	.42***
Social Trust	.39***	.33***	.45***	.40***
Depressive symptoms	−.27***	−.21**	−.27***	−.34***
Delinquency	−.23***	−.15*	−.31***	−.22***

*p < .05, **p < .01, ***p < .001.

Note: None of the correlation coefficients across the groups were significantly different from each other in magnitude.

Discussion

Today, young people live in increasingly diverse environments where, by contrast with past societies, they were more homogeneous in terms of ethnic, cultural, and religious background. Developing a strong bond to one’s own group may distance youth from other groups (Allen, Arslan, et al., 2021), which may eventually harm social cohesion and harmony in diverse societies. On the other hand, a strong sense of societal belonging – feeling connected, affiliated, and accepted by the larger society – may facilitate harmonious interaction and respect among members of a society regardless of how diverse backgrounds they come from and their affiliations with other groups and communities. Nevertheless, the correlates, predictors, and outcomes of feelings of belonging to a larger society have not been studied among adolescents. One potential contributor to the lack of research into feelings of belongingness towards society among researchers is probably the lack of an established measure of societal belongingness.

The aim of the current study was to propose a new research tool to assess sense of societal belongingness and to evaluate psychometric properties on an appropriate scale. The results suggest that the societal belongingness scale demonstrates measurement invariance at the levels of factor structure, factor loadings, and item intercepts across adolescents of immigrant, mixed-heritage, and Swedish background, and across males and females. Thus, the proposed scale can be confidently used to assess societal belongingness and perform group comparisons across these groups (Putnick & Bornstein, 2016). In sum, the proposed scale may be helpful in identifying groups at potential risk of low levels of perceived belonging to the larger society.

The results of the comparisons across the adolescents of immigrant, mixed, and Swedish backgrounds showed that there were major differences in how much they felt they belonged to the Swedish society. Specifically, adolescents with immigrant background reported the lowest level of societal belongingness whereas those with Swedish background reported the highest level. In addition, immigrant adolescents who were born abroad reported lower levels of societal belongingness than those who were born in Sweden. Overall, these findings are consistent with studies that have reported lower feelings of national identification among immigrants and minorities than among members of the majority group (Fleischmann & Phalet, 2018).

A noteworthy finding was that the societal belongingness scale demonstrated evidence of concurrent validity across different subgroups. The correlations between societal belongingness scores and all the indicators selected to examine concurrent validity (i.e., class belonging, social trust, depressive symptoms, and delinquency) were in the expected directions, not only for the full sample but also for adolescents with immigrant, mixed, and Swedish backgrounds. Overall, the results lend support for the concurrent validity of the societal belongingness measure. However, the scope of the current evidence is limited to concurrent validity, and future studies are needed to examine the criterion and predictive validity of the measure. In addition, we treated students who originated from different countries as a single group due to small group sizes. Thus, there is a need to examine the validity of the measure in different cultural groups to better understand if the current findings are robust across different subgroups.

In sum, the proposed ASBS scale demonstrates evidence of measurement invariance and concurrent validity, which supports the use of the measure for studying societal belongingness across adolescents of immigrant, mixed, and majority background. The measure is also suited for large-scale surveys due to few response items and ease of responding. We suggest that interested users adapt the items by using a word or phrase to refer to a larger societal context rather than one, which may singularly refer to an ethnic or majority group. We hope that the scale will facilitate growth in knowledge of the correlates, precursors, and outcomes of societal belongingness among adolescents, which may eventually support the development of practices and policy recommendations that would reduce the tendency to seek affiliations in radical and marginal groups to meet the need for belonging.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by Vetenskaprådet (the Swedish Research Council) (2015-01057)