I believe that if the theoretical instrumental model you can depart with the aim of confirming by AFC.

Hello Anh,

If you have an a priori expectation of what the structure should be for a given measure, then, yes: you can (and should) go with CFA (confirmatory factor analysis). If the proposed model offers poor fit to your data, then it's back to the theoretical drawing board for more work on the measure (with EFA perhaps being one of your tools).

You can, as well, test for factor structure invariance across samples (males, females) as well in a CFA/SEM framework, if this represents one of your research questions.

Was it the case that you had a much larger sample size for male respondents that led you to run both EFA and CFA for that portion of your study?

Good luck with your work.

To test the differences between men and women, you need to specify the same model for both sub-populations, and then assess whether it fits equally well in each. You can only compare a 7-factor model to a 5-factor model if one is a subset of the other. Further, you cannot include different variables in your two models if you want to do a systematic comparison

In terms of having an N of 150, a five factor model implies 10 correlations among the factors plus 20 loadings, or 30 parameters to estimate. To compare two different sub-groups, you double that to get estimated 60 parameters. Most advice on sample size in CFA recommends a minimum of 5 observations per parameter, so your sample is basically too small to estimate such a complex model.