When is Profiling Discriminatory?

Dare I suggest that profiling is discriminatory if it is inefficient? What about suggesting that it is not discriminatory if it is efficient? I wouldn't want to go too far in these directions without making it clear that I do not approve of efficient pusuit of discrimination (if that makes sense).

I have suggested as much before. For example, see here and here and here. In yesterday's piece, I pointed out that racial profiling in Kingston, Ontario, might not necessarily be discriminatory if the targeted group is, indeed, more likely to commit a crime (a possibility which seems to have been ignored by the study's author, Scot Worley; also, see here and scroll down to his profile [thanks to John Chilton for these two links, provided in his comments to yesterday's piece]).

But now consider domestic abuse. Grant Brown writes about his research in this May 31, 2005 letter to the National Post [p 21, $ req'd; thanks to Jack for the material]:

I found that gender was the single-most significant predictor of outcomes, more so than even prior criminal records, the injury level sustained by the accuser/victim, intoxication of the accused or whether children were present at the time of the incident, among other things.

For example, men were 20 times more likely to be charged with domestic violence when neither party was injured, even though hundreds of sociological studies over the years have shown that women, by their own admission, are at least as likely as men to commit non-injurious violence against their partners.

Women who seriously injured their partners were less likely to be taken into custody by the police than men who caused no injury to their partners. And men received significantly harsher sentences after conviction, when all other relevant variables were taken into account.

Because these conclusions do not align with the accepted dogma, they have been rejected and ignored by the Edmonton Police Service, even though they are vastly more robust than the allegations of racial profiling eliciting mea culpas from the Kingston police chief. Gender profiling in domestic disputes is pervasive in Canada; the establishment’s selective sensitivity perpetuates it.

Why are men targeted more than women? Is it because, as Grant Brown suggests, Edmonton Police discriminate against men in domestic abuse cases? Are his data the result of efficient profiling or discrimination?

Or is it because men, on average, are much bigger than women and are more likely to hurt women more seriously, conditional on hurting them in the first place? Or is there some other explanation? If someone has a link to his study, please post it in the comments or let me know via e-mail.