Which Dogs Bite?

[nicky187]

122. Sure do.

They're basing a policy decision on faulty inference. The results are (depending on how you phrased the research questions) Type I or Type II errors. In any event, the costs are going to be high in terms of loss of innocent life, and emotional distress.

I'd much rather see a study based on a random sample of dogs taken from veterinary practices or some other population register. Then if they check to see how many of those dogs have a history of biting, and why they bit, I'd be more pleased.

If that's not possible, then a retrospective case-control study could be done like this one, but could be designed better.

Even with flaws, the study has some merit, but this is what you get be trying to push inference beyond the limits of what is appropriate for the data and methods used.