[the above should read "stacy et al"]

Assertion
The authors cite a prevalence of pathological gambling (PG) of 0.3% to 1.3% in the general population and find a prevalence of 1.5% among those taking Mirapex. They conclude that there is an association between dopamine agonists and pathological gambling.

Problem
The fact that the authors put forward the two prevalences in the context of the assertion that there is an association between the drug and the behavior naturally implies that the difference between the two numbers is statistically significant such that it confirms their conclusion.

But the authors do not SAY it is statistically significant.

That could be because the difference between 0.4% to 1.3% and 1.5% is not statistically significant.

0.3% is the only portion of the range for the general population for which 1.5% represents a statistically significant difference. (calculations verified by a statistician)

I have looked around quite a bit and the absolute lowest prevalence I have found for PG in the general population is 0.4%, and that was in Canada. I would check the authors’ source, except they do not provide one.

This means not that the authors found an association, but that they failed to find one.

Also, in spite of the fact that all PGers were also taking levodopa, it is exonerated.