Abstract
Response format and method of estimation effects on confirmatory factor analysis. Five-point Likert scale and continuous response format are compared based on results from a 640 high school students sample. This work focuses in the 20 items State Self Esteem Scale (SSES) developed by Heatherton & Polivy (1991). Confirmatory factor structures are tested according to the construct validity models (single factor, three factors and multitrait-multimethod model). For these models, maximum likelihood and distribution free methods solutions are provided. After goodness-of-fit assessment, some psychometric issues are addressed: unequal method effects' performance depending on both, the response format and the estimation method being used.
