First the link to this week’s complete list as HTML and as PDF.
***
I wish psychologists would extend the careful preparation of their experiments and their specificity to precise and limited circumstances as claimed by Bryan et al. to the reporting of their results, which quite to the contrary often come out generalised and overblown as Gerber et al. rightly point out.
***
Vogt et al.’s title is misleading. They do not report on an observed change in attitude but on a trial aimed at changing it, and according to Christakis a failed or at least unproven trial at that. Nice try though, but why can’t researchers in the humanities ever be honest about their results?
***
To me Lyons et al. looks like another example for extreme overfitting of widely scattered and uncertain data as Telford et al. rightly point out.
***
Allen et al. are highly unconvincing. Their data points are very unevenly distributed on the map (fig 1a) and many of them cluster right on the border between areas. The area maps (figs 1b–d) do not support their conclusion. Both regressions in figs 2b and 2c are determined by one single data point alone, probably the same one in both cases. While fig 2a does look convincing, it’s insufficient for the case. As others have shown, it’s probably not productivity as such but its spacial clustering and defensibility that’s the determinant factor here.