My first reaction was that this was one serious peever. But then I read that the *Salon *article is based on a journal article. Unfortunately, U of T doesn’t yet provide access to this issue of* Computers in Society* (I suspect I will get access in the next few days) I have two questions: What is the sample size? What is the size of the effect Coulehan is writing about?

How many people and what kinds did they survey to determine that period use was aggressive? Is the number sufficient to generalize to the population as a whole? Often in studies like this, those surveyed are university students (they’re easy to round up and willing to do stuff for the small sums of money researchers pay), but students are not representative.

The second question about effect size is one that reporters almost always fail to ask. There is a difference between “statistical significance” and, what in the medical field is called, “clinical significance.” An effect may be real and measurable (statistically significant) but so small that it has no practical effect (clinical significance).

And the article flies in the face of the, admittedly unscientific, findings of dating sites like OKCupid that have determined good grammar is a turn-on.