Hello, 

I would like to run a EFA on my questionnaire to determine construct validity, however, my data is dichotomous. 

I have checked the literature and it has been suggested to run a polychoric correlation matrix instead and then use factor analysis on the matrix generated. 

I have generated a polychoric matrix using the following steps: ANALYSE -> CORRELATIONS -> HETEROGENEOUS CORRELATIONS 

My question is how to run a EFA using the polychoric matrix that I have generated?

When I try to run the factor analysis SPSS uses the Pearson's correlation matrix (ANALYSE -> DIMENSION REDUCTION -> FACTOR). How would I select my Polychoric matrix to run the factor analysis instead of Pearson's? 

I am quite new to SPSS, any help would be appreciated.

Thank you.

Please note that tetrachoric or polychoric correlation matrix is not always positive definite (p.d). FACTOR procedure, IIRC, will need a p.d. matrix for input.

Please look in Command Syntax Reference, FACTOR command, for how to input a correltion matrix fo factor analysis. You will be able to do extraction and rotation oc the factors, but not computation of factor scores (because you "don't have" data to input.

You might want also to check this answer and further links therein https://stats.stackexchange.com/questions/215404/is-there-factor-analysis-or-pca-for-ordinal-or-binary-data/215483#215483
And your tetrachoric or polychoric correlation matrix should happen to be positive definite ("gramian") in order to submit it to FACTOR successfully.