Introduction

Subjective well-being (SWB, ^{Diener & Biswas-Diener, 2000}) is a person’s assessment of their own life, which involves an emotional component, encompassed by the construct of happiness (^{Arita, 2005}), and a cognitive component, satisfaction with life (^{Diener et al., 1985}). Different cultures hold different meanings regarding what encompasses well-being and a good life. The Satisfaction with Life Scale (SWLS; ^{Diener et al., 1985}) is one of the most widely used scales to globally measure the cognitive component of SWB (^{Atienza et al., 2016}; ^{Dimitrova & Dominguez, 2015}; ^{Oishi, 2006}; ^{Sovet et al., 2016}; ^{Tomás et al., 2015}; ^{Whisman & Judd, 2016}). Given its extensive use across different cultures and populations, researchers using the SWLS need to ensure that the psychometric equivalence of this construct holds up across groups or across time (^{Huang et al., 2011}). The assessment of this equivalence is termed measurement invariance, a key analysis in both psychological and developmental research because it is a prerequisite for comparing group means (^{Huang et al., 2011}; ^{Jang et al., 2017}; ^{Putnick & Bornstein, 2016}). Cultural differences in interpreting satisfaction with life can result in nonequivalence of measures, or measurement noninvariance (^{Jang et al., 2017}).

Research shows that individuals typically compare their actual life circumstances with their ideal life circumstances when evaluating satisfaction with life (^{Zanon et al., 2014}). These “ideal life circumstances” and perceptions of well-being may differ among nations due to country characteristics, political climates, and societal values (^{Ponizovsky et al., 2013}). ^{Diener et al. (2013)}, and ^{Jang et al. (2017)} suggesting that there are cultural differences in values and practices which affect how people understand and report life satisfaction, as well as the weight they give to different variables that influence life satisfaction judgments. These judgments may differ based on whether individuals belong to individualistic or collectivistic cultures, religious or non-religious societies, or in developed or developing countries (^{Diener et al., 2013}). Judgements may also differ among members in the same overall cultural context but with multilanguage acculturation (^{Salazar-Fernández et al., 2020}).

Researchers have broached the cultural sensibility of the SWLS (^{Atienza et al., 2016}; ^{Schnettler et al., 2017}), although results have been inconclusive. Differences in the SWLS scores among countries may be ascribed to a measurement error due to cultural differences in interpreting the given scale items (^{Zanon et al., 2014}), plus the same mean scores from different countries may not reflect the same actual levels of satisfaction with life (^{Jang et al., 2017}). ^{Tucker et al. (2006)} reported metric invariance when comparing US and Russian university student and community samples, and lack of invariance when only community samples were included. ^{Swami and ChamorroPremuzic (2009)} reported that the SWLS remained invariant across genders and ethnic groups when comparing samples of Malay and Chinese adults from Malaysia. ^{Dimitrova and Dominguez (2015)} found partial scalar invariance (i.e., the intercept of each item for all groups) in emerging adults and adults from Argentina, Mexico, and Nicaragua. ^{Casas et al. (2012)} found metric invariance in four out of the five scale items among adolescent samples from Brazil, Chile, and Spain. ^{Ponizovsky et al. (2013)} demonstrated SWLS measurement invariance among immigrants of various age groups. ^{Díaz et al. (2015)} found that the SWLS and the Psychological Well-being Scales load on the same factor, showing an excellent data fit, with factor loadings and error variance being equivalent among adult samples from Spain and Colombia. ^{Whisman and Judd (2016)} confirmed partial scalar invariance using a sample of 50-to-79-year-old adults in the US, England, and Japan, as did ^{Atienza et al. (2016)} in a sample of Spanish and Portuguese adolescents. ^{Esnaola et al. (2017)} found strict invariance (i.e., residual invariance for each item is the same throughout groups) regarding country and gender, and scalar invariance for age, in a sample of Mexican and Spanish adolescents. ^{Schnettler et al. (2017)} reported partial measurement invariance among Chilean and Ecuadorian older adult samples. In Chile, ^{Salazar-Fernández et al. (2020)} reported strict invariance between acculturated and bicultural, and between acculturated men and women from the Mapuche indigenous group. Lastly, ^{Jang et al. (2017)} conducted a measurement invariance study in 26 countries finding that configural (i.e., items load onto the same latent factor throughout groups) and metric (i.e., factor loading of each item onto latent factor is the same for all groups) invariances held for all countries, but scalar invariance (i.e., intercept of each item is the same among groups) did not.

Differences in how individuals interpret and conceptualise the SWLS items may contribute to the failure to meet the criteria for different invariance levels (^{Oishi, 2006}; ^{Tucker et al., 2006}, ^{Zanon et al., 2014}). This occurrence may be due to cultural differences (^{Atienza et al., 2016}; ^{Oishi, 2006}, ^{Salazar-Fernández et al., 2020}; ^{Schnettler et al., 2017}; ^{Zanon et al., 2014}) or age (^{Casas et al., 2012}; ^{Esnaola et al., 2017}; ^{Tucker et al., 2006}, ^{Whisman & Judd, 2016}, ^{Wu et al., 2009}). ^{Jang et al. (2017)} have shown that a dissimilar mean age throughout countries significantly affected the intercept of Item 4, which states “So far I have gotten the important things I want in life”. The literature indicates that this item is endorsed more by elder than by younger respondents, thus elder samples may have a significantly higher intercept than younger samples. This suggestion is consistent with cross-sectional and longitudinal studies which point out that satisfaction with life is age-sensitive (^{Clench-Aas et al., 2011}; ^{Wu et al., 2009}).

To the best of our knowledge, only three studies have assessed the cross-cultural invariance of the scale exclusively using university student samples. Assessing satisfaction with life in these samples is attracting more interest in the literature, however (^{Tsitsas et al., 2019}) this interest entails attending the specific life conditions of emerging adults who are attending school, as they often face stress personally (^{Cadime et al., 2016}), socially, academically, economically, and in other areas of life (^{Çivitci, 2015}; ^{Hiçdurmaz et al., 2017}; ^{Lee et al., 2017}), regardless of the country or culture in which they live. Studies conducted with university student samples so far have included the measurement invariance of the SWLS across gender (^{Hultell & Gustavsson, 2008}; ^{Shevlin et al., 1998}; ^{Sovet et al., 2016}; ^{Tomás et al., 2015}), age (^{Hultell & Gustavsson, 2008}; ^{Tomás et al., 2015}), and time (^{Wu et al., 2009}). However, cross-cultural invariance for the SWLS in these populations has been less thoroughly explored.

Moreover, as with the findings in the samples of adolescents, adults and older adults, the findings with university student samples are inconclusive. Whereas ^{Tucker et al. (2006)} reported strict invariance when comparing US and Russian student samples, ^{Oishi (2006)} found that item 5 was not invariant among Chinese and US university students. Similarly, ^{Zanon et al. (2014)} found that items 4 and 5 were noninvariant among Brazilian and US university student samples. Therefore, further research is required to analyse the cross-cultural equivalence of the SWLS among university students from different countries and cultures. Cross-cultural equivalence among university students is especially relevant given student mobility between countries and continents for university studies. A measure that consistently captures the construct of satisfaction with life throughout cultures can aid in the development of institutional interventions and policies that improve the well-being of students from diverse backgrounds.

Therefore, the present study examines the measurement invariance of the SWLS among university student samples from Chile and Spain using multi-group confirmatory factor analysis. Although both Spanish-speaking countries share cultural characteristics due to their historical relationship (i.e., Chile was a Spanish colony until the early 1800s), each of these nations possesses significant cultural distinctions (^{Arenas-Gaitán et al., 2011}). For example, ^{Hofstede and Bond (1988)} identified differences in the cultural dimension of individualism vs collectivism in Chilean and Spanish citizens. Likewise, ^{Arenas-Gaitán et al. (2011)} found significant differences in the dimensions of power distance and masculinity vs. femininity among Chilean and Spanish undergraduate students. In ^{Jang et al.’s (2017)} cross-cultural study of the SWLS invariance, Chile was shown to have the highest factor mean of life satisfaction, whereas Spain presented the lowest. Taken together, these findings indicate the need for further exploration of the cross-cultural equivalence of satisfaction with life in young adults in Chile and Spain. Therefore, the purpose of this study was to evaluate the measurement invariance of the SWLS in university students in Chile and Spain. As research indicates that life satisfaction differs by gender in university students (^{Tomás et al., 2015}), the measurement invariance by gender was also assessed regardless of the country of origin.

Method

Data Analysis

The Statistical Package for Social Sciences (IBM SPSS) v. 23 was employed for descriptive analysis. The Cronbach 𝛼 coefficient was used to examine the reliability and internal consistency of the scale for each sample (> .7). Mplus v. 7.3 software was used to evaluate the one-dimensional factor structure of the SWLS in each subsample (Spanish, Chilean, male, and female; the latter two including participants from both countries) by means of confirmatory factor analysis (CFA), and to test the measurement invariance of the SWLS model between the Chilean and Spanish samples as well as between the male and female samples regardless of the country, using multi-group confirmatory factor analysis.

Considering the ordinal scale of the items, the polychoric correlation matrix was used to perform the CFA with Robust Unweighted Least Squares as the estimation method.

Different outcomes were considered for the analysis. To test the invariance of the factor model in the samples according to the country and gender, configural invariance, metric invariance and scalar invariance were verified (^{Dimitrov, 2010}). To determine model fit of the data of the one-dimensional factor structure of the SWLS, and to determine the achievement of configural invariance, model Chi-square (x²), Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), and Root Mean Square Error of Approximation (RMSEA) were used. Criteria for good model fit were nonsignificant model Chi-square, CFI > 0.96, TLI > 0.95, and RMSEA < 0.06 (^{Hu & Bentler, 1999}). Model fit was considered acceptable when CFI > 0.90, TLI > 0.90, and RMSEA < 0.08 (^{Marsh et al., 2004}). Lastly, to establish the achievement of metric and scalar invariance, a nonsignificant Delta of Satorra-Bentler Chi-square adjustment (D_SB x²) was required (^{Satorra & Bentler, 2001}).

An independent sample t-test was applied to compare mean levels of the SWLS scores of university students from Chile and Spain and between genders, using the Statistical Package for Social Sciences (IBM SPSS) v. 23.

Results

The standardised factor loadings and descriptive statistics and goodness-of-fit indices of both the Chilean and Spanish samples are displayed on Table 1. The CFA from both countries showed fit indices which provide support for a one-dimensional factor structure and reliability. Although the x² test was significant, it is well established that the x² test is sensitive to sample size (^{MacCallum et al., 2006}).

Table 1 Descriptive statistics of the Satisfaction with Life Scale scores in a sample of Chilean and Spanish university students 

a x² (df) = 477.6 (10), p < .01, RMSEA = .041, CFI = .997, TLI = .994. Cronbach’s a = .877

b x² (df) = 426.4 (10), p < .01, RMSEA = .027, CFI = .999, TLI = .998. Cronbach’s a = .808

Table 2 shows the results and comparisons of model fits for cross-cultural measurement invariance of the SWLS. The x² test was also significant in Model 0 (^{MacCallum et al., 2006}). Therefore, considering the fit indices, the overall fit of the baseline model was acceptable, which established the configural invariance of the SWLS in Chilean and Spanish university students. This finding supports the presence of a single-factor model in the two countries, and thus indicates that both university student groups conceptualise life satisfaction in the same one-dimensional structure. This model served as a basis for comparison for the metric measurement invariance model.

Table 2 Factorial Invariance for the Satisfaction with Life Scale model in a sample of Chilean and Spanish university students 

x² = Chi square. df = Degrees of freedom. D_SB x² = Delta of Satorra-Bentler adjustment of Chi-square. D_df = Delta of degrees of freedom.

p_D_SB x² = Statistical significance of Delta of Satorra-Bentler adjustment of Chi-square. RMSEA = Root Mean Square Error of Approximations. CFI = Comparative Fit Index. TLI = Tuker-Lewis Index.

Consequently, factor loadings were constrained to be equal throughout groups to test for metric invariance (Table 2).

The SB x² difference t est between configural invariance (Model 0) and metric invariance (Model 1) models was not significant, demonstrating support for metric invariance. Therefore, the factor loadings were invariant among university students from both countries. Thus, each item on the SWLS contributes in a similar degree to the latent variable among the university student samples from these two countries.

Next, equality of thresholds throughout groups was imposed on the model to test for scalar invariance (Table 2). The SB x² difference test between the metric invariance (Model 1) and scalar invariance (Model 2) models was not significant, demonstrating a scalar invariance. This finding indicates that group mean differences in the latent construct capture all of the mean differences in the shared variance of the items (Figures 1 and 2). Reaching a scalar invariance allows us to compare group means on the latent factor (^{Putnick & Bornstein, 2016}).

Figure 1 Path diagram of the Satisfaction with Life Scale for a sample of Chilean university students. 

Figure 2 Path diagram of the Satisfaction with Life Scale for a sample of Spanish university students. 

The CFA from both male (x² (df) = 759.6 (10), p < .01, RMSEA = .070, CFI = .996, TLI = .991) and female university students (x² (df) = 978.4 (10), p < .01, RMSEA = .046, CFI = .999, TLI = .997) also presented fit indices which provide support for a one-dimensional factor structure and reliability (male Cronbach’s a = .826, female Cronbach’s a = .877). The results of the multigroup analysis by gender are displayed on Table 3, showing that the model achieved configural, metric and scalar invariance (Figures 3 and 4.).

Table 3 Factorial Invariance for the Satisfaction with Life Scale model according to gender in a sample of Chilean and Spanish university students 

x² = Chi square. df = Degrees of freedom. D_SB x² = Delta of Satorra-Bentler adjustment of Chi-square. D_df = Delta of degrees of freedom.

p_D_SB x² = Statistical significance of Delta of Satorra-Bentler adjustment of Chi-square. RMSEA = Root Mean Square Error of Approximations. CFI = Comparative Fit Index. TLI = Tuker-Lewis Index.

Figure 3 Path diagram of the Satisfaction with Life Scale for a sample of male university students from Chile and Spain. 

Figure 4 Path diagram of the Satisfaction with Life Scale for a sample of female university students from Chile and Spain. 

The Spanish (Mean = 20.4, SD = 4.9) and the Chilean (Mean = 21.4, SD = 5.1) samples did not differ in the average scores for the SWLS (t = 1.588, p = .114). Likewise, the male (Mean = 20.9, SD = 5.0) and the female (Mean = 21.1, SD = 5.1) samples did not differ in the average scores for the SWLS (t = -.262, p = .794).

Discussion

This study examined the measurement invariance of the SWLS among university student samples from Chile and Spain using multi-group confirmatory factor analysis. Previous studies have assessed cross-cultural invariance in the SWLS in developing Spanish-speaking countries (^{Dimitrova & Dominguez, 2015}, ^{Schnettler et al., 2017}), by comparing developed and developing countries (^{Casas et al., 2012}). This study expands on the evidence of cross-cultural invariance of the SWLS by examining this scale on university student samples from developed and developing Spanish-speaking countries. The results suggest configural, metric and scalar invariance of the SWLS among university students in Chile and Spain, indicating that the SWLS is a valid instrument for cross-cultural comparisons regarding undergraduate populations from these countries.

These findings are consistent with results obtained by ^{Tucker et al. (2006)} when comparing US and Russian student samples, and also to those obtained by ^{Ponizovsky et al. (2013)} from immigrants from the Former Soviet Union in Israel, Bulgaria, and Germany. However, these results contradict those obtained by ^{Dimitrova and Dominguez (2015)} in adults from three Spanish-speaking countries in America (intercepts of items 2 and 3 non invariant), those by ^{Atienza et al. (2016)} in adolescents from Spain and Portugal (intercept for item 5 not invariant), and those by ^{Whisman and Judd (2016)} in middle-aged and elderly adults from the US, England, and Japan (intercept for item 4 not invariant). These studies reported only partial scalar invariance. Further research is necessary to explore the characteristics that account for scalar invariance levels in different groups (e.g., students, immigrants).

The findings from this study suggest a greater influence of common characteristics in Mediterranean and South American countries, than that of their cultural differences (^{Arenas-Gaitán et al., 2011}). Among these commonalities are language and the cultural tendency of Chilean and Spanish young people to live with their parents for a longer period, often during their university studies (^{Crocetti& Meeus, 2014}; ^{Schnettler et al., 2016}). There are antecedents of full cross-cultural scalar invariance among groups with marked cultural differences and different languages, such as those reported by ^{Tucker et al. (2006)}, and among immigrant groups in the same country (^{Ponizovsky et al., 2013}). At the same time, only partial scalar invariance has been supported among samples of different ages who speak the same language (^{Dimitrova & Dominguez, 2015}; ^{Schnettler et al., 2017}) and share a similar culture (^{Atienza et al., 2016}), as well as among countries with marked cultural disparities (^{Whisman & Judd, 2016}). Against this evidence, the present study shows that university populations from Chile and Spain share cultural conceptualisations regarding satisfaction with life.

SWLS items 4 and 5 did not differ between Chilean and Spanish participants. There are mixed findings regarding the cross-cultural discrepancies in these items, as they suggest a sense of accomplishment that younger participants might not be able to report. Some authors argue that item 4 might be positively linked with age and item 5 might be negatively linked with age (^{Atienza et al., 2016}; ^{Casas et al., 2012}). Moreover, it has been suggested that items 4 and 5 are related to the past while the other three items focus on the present (^{Oishi, 2006}; ^{Whisman & Judd, 2016}). Further research should evaluate the cross-cultural measurement invariance of the SWLS both among countries with large cultural differences and different languages (e.g., Western and Eastern countries), as well as in countries with similar cultures. These studies should consider, for instance, samples from Mediterranean countries and countries in which emerging adults are expected to leave home in their early twenties, as is the case in the US and in Northern Europe (^{Crocetti & Meeus, 2014}). In addition, because the level of economic development differs among South American countries, as well as conditions of access and funding for higher education, future studies should consider university student samples from different South American countries to evaluate cross-cultural measurement invariance in the SWLS.

Lastly, measurement invariance across genders was also assessed regardless of the country of origin. The results also suggest configural, metric and scalar invariance of the SWLS in male and female university students from Chile and Spain. These findings agree with the results of previous studies in university student samples showing that this scale is invariant across genders (^{Hultell & Gustavsson, 2008}; ^{Shevlin et al., 1998}; ^{Tomás et al., 2015}). It should be highlighted, however, that our results indicate that the SWLS is invariant across genders even when participants are of a different nationality.

Conclusion

Despite its limitations, this study provides evidence that the Satisfaction with Life Scale (SWLS) is interpreted correspondingly by university students in Chile and Spain alike, regardless of gender. SWLS measurement invariance has been widely tested across diverse cultural contexts and populations, but studies focusing on samples from adults in higher education have received less attention. The similar SWLS structure shown in university samples from two Spanish-speaking countries is an opportunity for significant latent mean comparisons in research, detecting variables that can influence life satisfaction and overall SWB. On a practical level, an instrument that measures life satisfaction consistently throughout student populations can inform the creation of common policies for improving the living conditions of university students both in developed and developing Spanish-speaking countries.