January 16, 2010

correlation is not necessarily causality; even if it is, it is not necessarily in the direction you presume

While the study on
learning to juggle was a longitudinal intervention study – the same people were studied before and after an intervention (in this case, learning & practicing to juggle) – the new study on ballet was a cross-sectional study, in which ballet dancers were compared to age-matched control (non-dancer) participants.

Correlation is not necessarily causality, so intervention studies are stronger than cross-sectional studies.

In the case of the ballet study, it is (formally) possible that the reductions in brain tissue seen in the professional ballet dancers were the cause of, rather than the result of, their choice & pursuit of that profession. Which reminds me that my Ph.D. advisor used to point out that cross-sectional studies purporting to show health benefits of exercise were flawed, because their results could be explained instead by healthier people being more likely to exercise. And which also reminds me of a bunch of drummer jokes...

No comments:

Post a Comment


Blog Archive