Hardly, Graham
Very good point. But the AC had one point you ignored: sample set size. Regressing a line from 20 sample points is a bit like reconstructing a skeleton from a single bone cell -- it's very rarely enough data to produce an accurate or precise result.
The other problem with this article, of course, is that there's no information on methodology (which would help determine if the sample set was large enough, as well as answer questions about the potential for self-reporting or distortion of results by the subjects, the measurements used to determine "normal" vs. "aggressive" sexuality, etc) or on sample source (which would answer questions regarding potential bias -- for example, if the professor was a little lazy, he might have not bothered to even leave the college grounds to get his subjects, leading to a greater potential for confounding variables.)
This is fairly common in reporting -- scientific studies and their results are presented without the necessary backup data and documentation to allow the general public to assess the validity of the research. In many cases, the scientist themselves refuse to release their raw data or their methodology.
But this is anathema to pure science. It leads the general public to distrust scientific findings in general, and allows (potentially well-meaning) scientists (e.g, CRU) to adjusts results and in extreme cases data to make their findings more compelling -- leading to the inevitable exposure of the fiddling and increased distrust of the findings and science in general, regardless of the accuracy of the underlying hypothesis.
Without the supporting data and methodology, any result is no better than a personal opinion -- yet publications continually give voice to these results without even bothering to check the data and methodology, if they're even available to check.
So yes, until they show their work, these are all crap studies.