A recent paper from University College London suggests that antidepressants have a small effect on people's levels of depression, but publication bias often results in negative studies being overlooked or spun in a positive way.