aiorazabala / qmethod

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

validate q.mrot.choose against PQMethod #256

Closed maxheld83 closed 9 years ago

maxheld83 commented 9 years ago

so, looking good, but turns out PQMethod actually does re-order as per #273, so that's something to keep in mind.

maxheld83 commented 9 years ago

this is done.

I have tested 2 setups in both programs.

Manual rotations, resultant loadings and plots check out.

Notes:

YouTube video of the affair is currently uploading.

Documentation has been amended to reflect the validation.

maxheld83 commented 9 years ago

oh, and here is the result for the 2nd analysis, couldn't get the first one, but it's all in the video.

qmethod

maxheld83 commented 9 years ago

I should add: the validation of the interactive part of this feature, q.mrot.choose is actually relatively unimportant, because / as long as q.mrot.do does what it says it does (it has a bunch of unit tests).

Recall that q.mrot.do applies an arbitrary rotation matrix to a results object, and re-runs all of the qmethod steps accordingly. Crucially, this q.mrot.do, with, say, a manually supplied varimax rotation matrix must be the same as a qmethod(rotation = varimax) results object – if that is so, the rotation works fine. That is now part of the test.

q.mrot.choose merely calls q.mrot.do over, and over again, and it plots results from this results object, so in a way, users always get what they see (because q.mrot.choose merely returns the last rotation matrix).

So, the interactive function -- which cannot be programmatically tested -- is, luckily, not so important.

That being said, I have tested it against PQMethod, and it works fine.

It also meets the expectations (direction of rotation, changes of other plots, etc.).

maxheld83 commented 9 years ago

and here's the abovementioned video: https://www.youtube.com/watch?v=4Nk5gYmlzM8&edit=vd