This study examined the measurement invariance of the Geriatric Depression Scale Short Form (GDS-15) in older adults across five countries in Latin America and the Caribbean. Multiple group confirmatory factor analysis (MGCFA) and item response theory likelihood ratio tests (IRTLR-DIF) were used to test the measurement invariance of the GDS-15 by gender and cross-country comparisons. The sample for this study was made up of 7,573 older adults between the ages of 60 and 102. Data for the present study comes from the Survey on Health Well-Being and Aging in Latin America and the Caribbean (SABE)(Pelaez, et al., 2004). The underlying factor structure of the GDS-15 was examined within each country. A one-factor structure was found for the countries of Chile and Cuba and a two-factor was found for the countries of Argentina, Mexico and Uruguay. Results of the multiple group confirmatory factor analysis offered support for full measurement invariance of a one-factor structure by gender within the countries of Chile and Cuba. Partial measurement invariance was found for a two-factor structure by gender within the countries of Argentina, Mexico and Uruguay by gender. The IRTLR-DIF analyses by gender and cross-country comparisons revealed a lack of parameter invariance. Items exhibiting DIF by gender were on average more difficult for men to endorse than women. In summary the IRTLR-DIF procedure identified more non-invariant items than the MGCFA procedure. Implications and recommendations for future research are addressed.
Read
