Published on Development Impact

Weekly links September 2: p-values and uncertainty, training policymakers in causal thinking, forgetting everything I read, and more…

This page in:

The blog will be back with new posts next week. In the meantime, here are some of the interesting things I read or listened to during the summer.

·       Training Pakistani policymakers in causal econometrics makes them more likely to say quantitative evidence is important and to show interest in conducting randomized trials –on VoxDev,  Sultan Mehmood, Shaheen Naseer and Daniel Chen summarize their recent paper.

·       In the recent journal of economic perspectives, Guido Imbens discusses statistical significance, p-values, and the reporting of uncertainty – “Although hypothesis testing is routinely used in economics, I would submit that many of the substantive questions are primarily about point estimation and their uncertainty, rather than about testing.”… “Statistical significance has been over-emphasized in empirical research. In many cases where decision makers are faced with deciding whether to implement a new policy or not, confidence intervals are a more useful way of communicating uncertainty of point estimates.”….” in …cases where the questions of interest are naturally formulated as hypothesis tests, …it is my view that p-values are a reasonable and standardized way of communicating the strength of the evidence. Summarizing the strength of that evidence by a binary indicator—whether a statistically significant at the 5 or 1 percent level—seems to serve little purpose.”

·       In the same issue, Max Kasy discusses publication bias, and one rationale for not wanting to publish a lot of null results: “there is a deep tension between relevance for decision-making and replicability in the design of publication rules. … this type of logic holds more generally in any setting where published research informs decision-makers and there is some cost which prevents us from communicating all the data. Such a cost clearly must be present; otherwise it would be optimal to simply publish all data, without any role for statistical inference, researchers, or journals. Given such a cost, it is not worthwhile to publish “null results”—that is results that do not change decisions relative to the default absent publication. Surprising results, on the other hand, especially those that lead to large changes of optimal decisions, are of great value to decision-makers, and should thus be preferred for publication….similar conclusions hold when our goal is to maximize social learning, subject to attention constraints.”

·       The IGC blog has a series of posts summarizing papers presented at the BREAD economics of Africa conference.

·       On the OPM blog, Pierre Bachas and Alípio Ferreira Cantisani explain what happened when they experimented with having algorithms to choose which firms to tax audit rather than tax inspector discretion – the algorithms chose smaller firms, that were slightly less likely to show irregularities after audit than the firms chosen by inspectors.

·       Tips for working with large datasets in Stata from Pere Taberner

·       I always enjoy listening to Sendhil Mullainathan and his discussion with Steve Levitt on the People I (mostly) admire podcast is no exception. Happy to hear I’m not the only one who forgets most books I read, and loved the discussion of the pros and cons of admitting that often you are unsure whether or not the advice you are giving is good.

·       Is Esther Duflo Veronica Mars? Another fun podcast I listened to recently is We are Supported by, which has an episode in which Kristen Bell and Monica Padman interview Esther Duflo. A great example of how to communicate what development economists do to a wider audience. Indeed, one of the points she makes is the importance of communication and dissemination of results to policy audiences, and that publishing new findings is not going to change much by itself. She also discusses gender issues in the profession a lot, the importance of role models, and that people are more persuadable when given relevant information than we often think.

·       The International Economic Association has started a new feature on its website where it highlights the work of a different economist from around the globe each month or so. The first is Kanika Mahajan from Ashoka University in India.

·       On VoxDev, Stephen Anderson and I summarize our forthcoming JPE paper on hiring marketing and accounting specialists in Nigerian SMEs. We also discuss policy implications, including the idea that government subsidies for these services might be “market-making”.

·       Over at the CGD blog, Crawfurd et al. find that one-on-one phone tutorials by teachers did not help students in Sierra Leone learn during school closures.


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