Reanalyzing the 16pf-5 second order structure:: Exploratory versus confirmatory factorial analysis

In the present study, the second order structures of the 16PF-5 for 3, 4, 5 and 6 factors are analysed in a sample of 636 undergraduate students and their friends and relatives, with a mean age of 25.09 years (sd: 9.20). A two-stage analysis with Exploratory and Confirmatory Factor Analysis (EFA and CFA) was performed assigning subjects either to a calibration sample (314 subjects), or to a validation sample (322 subjects). First, EFAs were conducted for each of the 4-factor structures with the calibration sample, and second, CFAs were specified and assessed also for the 4-factor different structures. Four different situations were tested, simple structure, secondary loadings >.30, modification indices between covariance error terms > 20, and modification indices between covariance error terms > 15. Results show that the 16PF-5 simple structure yields poor fit indices throughout all models, although when incorporating secondary loadings and correlating the variances of the error terms with modification indices > 15, the 5-factor model appears to be the best representation of observed data. The usefulness of CFA in the study of personality structural questionnaires is discussed in the light of these results.