IPA’s weekly links

Guest post by Jeff Mosenkis of Innovations for Poverty Action.

  • In a story circulating widely now, Science journalist John Bohannon, explains how he planted a fake study in the news showing chocolate can help with weight loss. The study was really done, but “p-hacked” – they compared two diet groups (one including chocolate) to a control and measured lots of outcomes so at least one one would come out as statistically significant by chance, then published in a fee-for-publication journal and put out a press release about it, which several outlets picked up without asking any critical questions.

3 points:

-There may ethical questions around deliberately misleading the public, especially about health.

-None of the outlets who picked it up were particularly big names.

-Dean Karlan (in a very cursory glance), points out that they were already dieting and 1.5 oz of chocolate is very few calories, so having that outlet could have kept people from eating other sweets (as another study suggests). The findings were in weight loss, the area one would hypothesize, rather than other areas like sodium level, so the findings might actually be legit, which would be the greatest con of all.  (h/t Everybody.)

  • Psychologist Michael Inzlicht says that researchers have to start being honest with themselves about cherry picking findings, starting with himself. He subjects his own results to three tests anybody can do: First a “p-curve” which checks your results against a pre-set rule to see if your findings are consistently too good (paper and calculator here). He did OK there, but fared worse on the Test for Insufficient Variance (TIVA) which involves converting p values to z-scores (explained here), and Replication Index (or R-index) which tries to determine the probability a finding can be replicated using power and frequency of success rates across a series of studies. However supporting Chris’ Drama Queen rule, Inzlicht wasn’t alone:

Uli Schimmack … also reanalyzed a set of psychological studies appearing in Science, finding that they achieved a painfully low R-index of 33.9%. These low values raise concerns about the empirical support for the underlying articles.


  • A mystery phone without many features is taking Ghana by storm, entirely by word of mouth, according to Quartz. It’s bulky, ugly and doesn’t work well, but its main feature is “power banking” – a huge battery that lets people save up power and charge their other devices during frequent outages (there’s a mobile banking joke in there somewhere).
  • NPR looks at Somali refugees who’ve spent the past 24 years in a refugee camp in Kenya (the subject of a forthcoming book by Ben Rawlence). A generation has grown up there with western-style education and job training, and under rules designed to promote gender equality. Rawlence points out that in many ways they look more like a middle class than those still in Somalia. A group of the refugees were recently taken on a U.N.-sponsored visit back to Somalia and were unimpressed, according to the story.

And, from Ian Bremmer, who titles this “Theory vs. Practice”: