Though this might be true about the article itself as posted, this is not a bit of bad journalism. The author Sydney Finkelstein, is a Professor of Management and spent 6 years studying 50 companies and conducting some 200 interviews, here's a link to the journal article : http://www.jacksonleadership.com/pdfs/7Habits_IveyBusinessJo...
So to answer your "If you really wanted to check the effect of a certain character trait, you'd need to try to (carefully) compare how CEOs with the trait performed compared to CEOs without the trait, and make damn sure your sample sizes and methods are sound before making any confident assertions." It seams like 6 years, 200 interviews, and research in 50 companies would seem to be a fairely extensive study IMHO.
Furthermore, just because someone is an academic authority hardly makes them less liable to fall for the cognitive biases that davidkatz mentions (that we color character traits into stories that fit the result). It goes without saying that we all have a strong bias to attribute other people's actions to their personality rather than conditions in their environment over which they have no control (http://en.wikipedia.org/wiki/Fundamental_attribution_error). We generally don't apply this error to ourselves.
The classic example is that if someone trips over a rock, you'll tend to have the impression that they're clumsy. If you trip over a rock, you won't think you're clumsy, you'll just say to yourself that the rock was sticking out.
When someone says: 'here are 7 habits of unsuccessful CEOs' and does not even attempt to account for the problem of 'how do we know it's really these habits that caused the company to fail?', but rather just flat out assumes it in the face of decades of findings that suggest that this is an easy trap to fall into, I for one can't take them seriously, and academic credentials have almost nothing to do with it.