I could be mistaken, but the correlation that I believed to exist was that refereeing was common in fields where works where fully self contained. E.g. with mathematics you do not need to additional research to (in)validate a work, as long as the steps are correct the conclusion is correct. Contrast this to a field like medicine where you would need to redo all the experiments as well to truly 'referee' a result, whilst peer review 'simply' means somebody went over your work and gauged it's reliability.
Now, I could well be mistaken and it's definitely not a clear cut line, but that seemed to be the general trend I observed. Then again, I don't know that great a number of fields that use the term 'refereed' over 'peer-reviewed' in the first place, so I could well be mistaken.