Emil O W Kirkegaard

Reference appreciated-- my point concerned the interpretation of certain items listed under "personal rights" and "press freedom," but I think the most likely conclusion is that my preferences simply differ in certain slight but relevant respects from the western liberal norm.

I would rather ban street protests than have to cross one to get into my building to go to work, for instance. I would rather ban libel and enforce the ban stringently than allow the media to more actively mislead the public. Things like that.

If so, then considering the likely audience, the factor score is probably correct for the average reader and the perceived deviation is merely relative to my subjective policy preferences.

As I said, it doesn't matter if you exclude the political indicators you don't like.

In approximate terms, certainly. I am aware the effect is small, but it _is_ directed in the manner that I described. Robustness, yes, but not immutability.

That the results are relatively robust is expected-- any perturbation of the input variables produces only a small change in the factor loading. That they would be identical, however, is mathematically impossible in lieu of perfect collinearity. It is a minor point.