I am writing a paper doing multiple regression on self report data. We didn’t learn about this in our quantitative methods lectures, and my supervisor tells me she is not too good at the methods/statistics part. So I have been reading about multiple regression in different books and seen some youtube videos, and of course I want the assumptions to be met before doing the analyses. I find it a little confusing, since different authors and video makers say somewhat similar things, but in different ways and with different ways to test the different assumptions, and not everybody writes about multiple hierarchical regression.
What I find most confusing is testing for independent errors, and since I’m using questionnaire data, I’m not performing the Durbin Watson. I did a ZPRED * ZRESID plot, and it looks like below.
As far as I understand, optimally the dots should be spread out, which they aren’t. (Is that correct when there are many IVs?)
For the individual errors assumption, what does it mean that they are spread out like this? The plot has the same shape for two populations with N=700 and N=7000. R^2 total is about 0,25.
Why are they all slanted, not vertically straight?
If anyone has any knowledge that might help me understand this, thanks for your time 🙂