Diane: Basically, yes. On just my first thoughts, it seemed inconsistent that child abuse should not show the same trends as "social negligence scale" for instance. Certainly there is a 'noise' out in the emergency rooms etc; that child abuse has little respect of either poverty, social class or other 'social trends'. In the specific instance (ie one patient crying in front of you) that may be true. But I would have thought that a deeper link would be shown. But what do I know? Playing with these very very large data sets must be a completely different experience - or what is that fancy German phrase 'weltanshaung" or something. I realised after I wrote - that the conjectures on 'statistical power' might be completely off. Because the question of 'power' only applies to 'negative answers'. So if the PI's have shown a link it is all OK...?!? Assuming ... that the alpha correction is OK. I think anyway..... ? Thanks for your kind reply. H