Strategic Use of Random Subsample Replication and a Coefficient of Factor Replicability

The problem of demonstrating replicability of factor structure across random samples is addressed. Procedures are outlined which combine the use of random subsample replication strategies with the correlations between factor score estimates across replicate pairs to generate a coefficient of replicability and confidence intervals associated with the coefficient. Data from the national norming sample of the Self Observation Scales are used in the illustrative example.