search

aiorazabala / qmethod

R package to analyse Q methodology data
GNU General Public License v2.0
34 stars 18 forks source link

fix: principal::psych flips negative columns on rotation, q.mrot.do does not #268

Closed maxheld83 closed 9 years ago

maxheld83 commented 9 years ago

another problem that came out of testing (luckily), where it's hard to replicate whatever psych does.

It turns out that sometimes, psych appears to reflect columns in the loadings matrix, whereas q.mrot.do does not do that.

Similar to #263

maxheld83 commented 9 years ago

above commit does not actually solve this problem, just avoids it in testing -> suboptimal

maxheld83 commented 9 years ago

on second thought, this might not be a bug or problem.

Contra re-ordering in #263, it might well make sense to have "inconsistent" results here, i.e. q.mrot.do not flipping and qmethod() with flipping.

With automatic rotation, it is reasonable to assume that users want mostly positively loading factors (negative ones are hard to interpret).

With manual rotation, however, that is less clear – people might very well prefer an average-negative loading factor, so long as they are aware of it.

So, we can just keep this as is, so long as all the tests take care of it (they do).

I've also added manual flipping in case reorder = FALSE.