Beruflich Dokumente
Kultur Dokumente
As you can see, these journalists (or their editors) attached extremely strong
titles to their science articles! An actual scientist wouldn't describe the results
of a study with such strong terms as "prove" or "They work." That's because
research in science is a steady accumulation of evidence--each study teaches us a
little bit more, but no study can "prove" a theory or a claim.
Do antidepressants work?
The "study" mentioned in the three headlines above was actually a meta-analysis of
522 clinical trials (that is, randomized controlled studies) of antidepressants.
Here's a summary and interpretation according to the Neuroskeptic blog:
a) Review: What does a meta-analysis do? Why might we value a meta-analysis over a
single study?
b) When the journalist describes the Standardized Mean Difference (SMD), they are
referring to a statistic very much like Cohen's d. As you can see, the conventions
for SMD are the same as for Cohen's d. Do you agree that the effect size of .31
could be considered "modest" according to these conventions?
c) I wrote above that "no study can 'prove' a theory or a claim." But what about a
meta-analysis--do you think meta-analyses are more likely to be able to prove a
theory? Are they definitive? (Why or why not?).
The thing is, �effective but only modestly� has been the established view on
antidepressants for at least 10 years. Just to mention one prior study, the Turner
et al. (2008) meta-analysis found the overall effect size of antidepressants to be
a modest SMD=0.31 � almost exactly the same as the new estimate.
Second, the Neuroskeptic cleverly points out that, a few years ago, the media
assigned the opposite headline to virtually the same result: