Published on Development Impact

Weekly links July 31: Household surveys in crisis, Doing Business, machine learning, and more…

This page in:
  • Household surveys in crisis – Bruce Meyer and co-authors in a new NBER working paper highlight the many issues due to declining cooperation of respondents in the U.S.:
    • Unit nonresponse: Households have become increasingly less likely to answer surveys at all: nonresponse rates for major U.S. surveys like the CPS, SIPP, and GSS now exceed 20 percent
    • Item nonresponse: Those that respond are less likely to answer certain questions: imputation rates average about 25 percent for transfer income
    • Measurement error: when they do provide answers, they are less likely to be accurate: compared to admin data, for the main cash welfare programs in the U.S., less than half the dollars given out are captured in surveys, with the downward bias increasing about one percent per year
    • They discuss several reasons for this trend, and suggest a key reason is that respondents are over-surveyed “We suspect that talking with an interviewer, which once was a rare chance to tell someone about your life, now is crowded out by an annoying press of telemarketers and commercial surveyors.”
  • The latest issue of the Journal of Economic Perspectives is now out, with a mini-symposium on pre-analysis plans (Olken, Coffman and Niederle) and on Doing Business (Besley, Hallward-Driemeier and Pritchett)
  • Marc Bellemare on his approaches to dealing with outliers
  • From the Economist, a summary of scientific research showing a correlation between more self-control and subsequent worse health outcomes for people from poor backgrounds
  • Slides from Guido Imbens and Susan Athey’s NBER summer institute lectures on machine learning
Development Impact will be on summer break for August, with the possible exception of occasional links


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000