IPA’s weekly links

Guest post by Jeff Mosenkis of Innovations for Poverty Action.


  • An interesting new Center for Global Development working paper looks at the effects of family planning services becoming available across Malaysia. They find that it helped girls earn more later in life, even those too young to benefit directly from the program. Girls just born grew up to earn more (and were more likely to have their elderly parents move in with them), presumably through a generalized improvement in women’s empowerment in the area (h/t David Batcheck).
  • Lucia Diaz-Martin, Rachel Glennerster, and Ariella Park warn that using whether women work or not as a proxy for empowerment in research can be problematic, because it matters why they’re working
  • It’s never a good omen for a body of literature when David Roodman takes his microscope to the code. In this case, just in time for the holidays, he releases a Worm Wars prequel. GiveWell rates deworming as one of the most effective charitable causes, but the case is based on a just few data sets. One is Hoyt Bleakley’s historical finding from the 1800s that hookworm eradication in the American South resulted in improved education. Roodman went back to re-examine the case with the benefit of a 100x larger sample, thanks to more recently added census data. He now thinks the improvements in education observed might have been a continuation of a trend that had already been in progress.
    • Despite his findings, GiveWell still thinks the case for current deworming efforts in developing countries is strong and continues its recommendations in support.
  • But in a remarkable development, hookworm is back in the American South.
  • Some in government are pushing for policy to be made based on “sound science.” FiveThirtyEight explains this is being used as a Trojan horse, turning the scientific process against itself. Historically this tactic has been used by tobacco and energy industries to delay policy until scientists have conclusively proven something or under the guise of needing more rigorous studies to establish certainty. Conveniently, it turns out that topics like smoking and climate science can’t be subjected to randomized controlled trials.
  • Nature asked several statisticians how to improve science. Opinions varied on if “to p or not to p” is the problem or it’s more deeply embedded in norms of how statistics are used in different fields.


And, in 1988 The Lancet asked clinical trial authors to do a subgroup analysis to find out for which patients (out of their 17,000) had benefitted most from using aspirin to prevent heart attacks:

Peto, a statistical rigorist, refused — such analyses would inevitably lead to artifactual conclusions — but the editors persisted, declining to advance the paper otherwise. Peto sent the paper back, but with a prank buried inside. The clinical subgroups were there, as requested — but he had inserted an additional one: “The patients were subdivided into 12 … groups according to their medieval astrological birth signs.” When the tongue-in-cheek zodiac subgroups were analyzed, Geminis and Libras were found to have no benefit from aspirin, but the drug “produced halving of risk if you were born under Capricorn.” Peto now insisted that the “astrological subgroups” also be included in the paper — in part to serve as a moral lesson for posterity.