Evaluating Different Versions of the Multigroup Ethnic Identity Measure: Application of Modern Psychometric Strategies
Currently there are two competing versions of the Multigroup Ethnic Identity Measure (MEIM) used in the extant literature: a twelve item version (MEIM-12) and a revised six item version (MEIM-R). The factorial structure of both versions has repeatedly tested, with mixed results indicating that a correlated factor structure sometimes fit best, whereas a unidimensional factor structure fit best at other times. The primary purpose of this study was to test whether a bifactor model best represented the underlying structure of both the MEIM-12 and the MEIM-R. A secondary purpose was to provide further evidence of convergent and divergent for both measures and to test whether they were equally related to theoretically related constructs. Confirmatory factor analysis was used to compare competing models and differential item functioning using item response theory was used to examine measurement invariance. The prediction that a bifactor model would fit the MEIM-R and MEIM 12-item better than competing unidimensional and correlated two factor models was supported. Additionally, measurement invariance was established across individuals of Hispanic and non-Hispanic ethnicity for both measures. Total scores on the MEIM-12 were positively correlated with positive affect, living authentically, belongingness, stigma consciousness, and psychological flexibility, and were negatively correlated with satisfaction with life. Total scores on the MEIM-R were positively correlated with positive affect, living authentically, belongingness, stigma consciousness, and psychological flexibility, and were negatively correlated with satisfaction with life and psychological resilience. The MEIM-12 and MEIM-6 were differentially related to measures of resilience and belongingness, with the twelve item version being slightly more strongly related to these than the six item version. These findings definitively indicate that total scores, rather than subscale scores, should be calculated for both versions of the MEIM, and, moreover, that the six item version is sufficient with respect to most research applications.