Statistical tools

I follow Language Log, because since the 20th century the linguistic emphasis in philosophy has been sufficiently great that I frequently get the linguists' jokes.  It also often ends up covering interesting topics.  Mark Liberman has recently been criticizing the shoddy science in Loanne Brizendine's The Female Brain, and his most recent post on the topic introduced me to a statistical concept with which I was heretofore unfamiliar, the cohen's d effect size.  I suppose it is unsurprising that statisticians would have developed lots of different ways of measuring the sizes of effects.  It is also not surprising, though it is depressing, that these more informative statistical tools are so rarely employed in any examination of the significance of gender differences in the brain.  I'm glad some people are keeping an eye out for the shoddy science; if only they got as much press.

Comments

[Chris]

You know, effect size is pretty regularly employed, if not by the individuals conducting the research, then by reviewers, but that doesn't mean that a paper with a small effect size won't get published, and that people won't make too much out of it because they don't consider effect sizes.

Interestingly, in most of the social sciences, calculating power is routine, and in order to calculate power, you have to have some estimate of effect size. So, before studies are conducted, they have some idea about what effect size to expect.