> I don't get this. If, say, the actual effect of lead on crime is 0.33 on their scale (a "large" effect size) then you'd expect to find papers clustered around that value<p>You'd expect them to be clustered around that value symmetrically, right? That's clearly not the case in the diagram that the blog post author quoted. There's a cluster around very weak positive effects, and then a long tail of strong positive effects but no matching long tail of weak negative ones. This suggests either that the negative results were truncated out, negative results have been hacked to positive ones, or there is some confounding factor at play. And I think the modeling in the paper is just an attempt at finding the confounding factor, and not finding one.<p>That said, the author's arguments around negative results being publishable in this field + studies not disappearing seem pretty strong.<p>(Or, at least this is my reading of the situation as a total amateur.)