Positive effect bias in peer-reviewed studies

(via Marginal Revolution)
A new study published in the Archives of Internal Medicine points to a positive effect bias in medical studies. What this means is that studies that report positive effects from the treatment under study are more likely to be published that studies that report no effect form the treatment.

I suspect this holds for the social sciences, and is potentially more pronounced because of the quasi-experimental nature of many social science studies. Replicating something similar to the medical study in planning, and specifically in transportation-land use studies looking for causality towards a specific travel mode, would be valuable.