A large study, cited here, makes the link between soda consumption and depression. The findings were good enough to make the front page of yahoo.com.
The sample size, and some of the adjusted covariates used, appear to make sense. My gripe isn’t with the statistical methods. However, one of the main issues with the release of this information for public consumption is what happens when non-statisticians try and interpret statistical findings. In this article, the author clearly identifies that only an association between the two variables (soda consumption and depression) was found, and that more work is needed on the subject. The article’s editor, however, goes a step further. The title of the article implies a direct causal effect, and the subtitle to the picture (shown below) adds to this issue, stating “Is your diet soda making you depressed?”
Because this article doesn’t address causation, such headlines and descriptions should be reworked, especially when read by the masses. What we know? Soda is linked to depression in this sample. What we don’t know? If soda consumption is causing this depression.