Testing Measurement Invariance of a Brief Sense of Community Scale Among Black, Non-Hispanic and Hispanic Youth: Identifying Differences in Item Intercepts and Latent Factor Means

Background and Purpose: Although the Brief Sense of Community (BSCS), an 8-item tool, has been used with racially and ethnically diverse groups, validation of this instrument among those groups is scarce. Only one study has validated this scale among youth of color, which included Black, non-Hispanics and Hispanics. However, the analysis was done as a single group. As a result, the measurement equivalence of the BSCS across Black non-Hispanics and Hispanics is unknown. Since the literature indicates that culture may have an influence on individuals’ sense of community, it is critical to validate the BSCS in a multi-group factor analysis simultaneously across the two groups. Model equivalence is critical in cross-cultural comparison using the same instrument. However, this level of scrutiny has not been performed for the BSCS. Thus, this study seeks to: (1) test the factor structure of the BSCS across Black, non-Hispanics and Hispanics; and (2) assess between-group differences on measurement invariance.

Methods: This study used data collected in 2016 from a self-administered online survey provided to college students in a Northeastern university in order to determine their experiences and attitudes related to sexual assault. Participants were also asked the extent to which they felt part of their campus community by using the BSCS, with a minor adaptation to indicate that the university campus was the community of interest. The BSCS assessed participants’ needs fulfillment, group membership, influence, and emotional connection. Almost two-thirds of participants were female (65%). Nearly 22% were Asian, non-Hispanic; about one-fifth (19.77%) was Black, non-Hispanic; a little over one-quarter was Hispanic (26.23%) and 28.09% were White, non-Hispanic.

Multi-group Confirmatory Factor Analysis was used to assess the validity of the BSCS as a first order, four factor structure for Black, non-Hispanics and Hispanics simultaneously, and to test the equality of the BSCS parameters across the two groups.

Results: The configural invariance model indicated that the BSCS four-factor structure is supported across Black, non-Hispanics and Hispanics (X²= 68.707; DF= 28; CFI=.988; RMSEA=.045). The measurement weights model indicated equivalence of factors loading across the two groups (X²= 3.737; DF= 4; p=.443). The measurement intercepts model (X²= 31.130; DF= 12.31; p<.05), structural covariances model (X²= 36.878; DF= 22; p<.05), and the measurement residual model (X²= 44.938; DF= 30; p<.05) indicated that the two groups were not equal based on intercepts, variances and covariances, and error covariances. That is, Black, non-Hispanics had statistically significant higher item intercepts than Hispanics and their mean intercept was not equal for two latent constructs: group membership and influence.  

Conclusions and Implications: The multidimensional construct of the BSCS was confirmed for Black, non-Hispanics and Hispanics. Similarly, the factor loading for each of the BSCS dimension was equal across the two groups. Since equivalence was not upheld for all models in the analysis, caution should be exercised when testing for mean differences between Black, non-Hispanics and Hispanics as their mean intercepts are not equal. This study highlights differences among Blacks, non-Hispanics and Hispanics on their understanding of group membership and influence, warranting further exploration.