Can I run factor analysis on items that are measured on two different 5 likert scale, that is one scale is measuring frequency ( i.e never , rarely, occasionally, frequently and always) and other agreement ( strongly disagree, disagree, not sure, agree and strongly agree)
$\begingroup$
$\endgroup$
4
1 Answer
$\begingroup$
$\endgroup$
1
Well, you could do this but why would you?
Factor analysis starts with the correlation or covariance matrix of the items you include. Then it tries to find patterns in order to create latent variables.
What will you do if you find a latent variable that includes both frequencies and agreements? Will that be of any interest? Will you be able to use that latent variable to do anything?
Probably the answers here are 1) Be confused 2) and 3) No.
If that's the case, there is no reason to do the FA this way.
answered Apr 17, 2016 at 11:21
-
$\begingroup$
you find a latent variable that includes both frequencies and agreementsWell, this would sound a bit vague. One thing is to argue that incompatible stimulus measurement units may distort or impede proper correlation between the items. Another thing is to argue that a latent trait can or should itself have "units". $\endgroup$– ttnphnsApr 17, 2016 at 12:51
1 2 3 4 5, for a 5-level likert scale. $\endgroup$