Talk:Differential item functioning

Is the DIF graphic correct? edit

I think the graphic's author has tried to show two parallel logistic curves for uniform DIF and non-parallel logistic curves for nonuniform DIF. Unfortunately, that's not what it appears like, and it's likely to confuse some researchers (since two logit curves with the same slope may look nonuniform since they aren't parallel on the *probability* scale).

Should we generate a parallel and nonparallel logistic slope using, say, R to have a better depiction? Or have I misunderstood the differences here?