· Why mediation analysis will generally be invalid by Uri Simonsohn on data colada. “Examples of intuitive shortcomings of mediation: the mediator of interest to the authors correlates with other mediators that are not of interest and thus not included in the regression. If the excluded mediators correlate with the included one, a very likely scenario, the included mediator picks up spurious mediation from the other ones. This bias also affects virtually every published mediation analysis. In addition, non-linearities and measurement error can produce spurious mediation, and there may be reverse causality with the 'dependent variable' causing the change in the mediator….this post focuses on a counterintuitive one…The problem of interest to this post is that if there is any variable, besides X, that correlates with M and Y (a very likely scenario), mediation is invalid.”
· Scott Cunningham interviews Rajeev Dehejia in a recent podcast, and talks about how the standard Table 1 (comparing characteristics of treatment and control) is so important. “Rubin and Imbens one semester co-taught an innovative new class on causal inference and Rajeev was one of the students who took it that year. Together with his classmate, Sadek Wahba, the two students decided after the class concluded to not so much replicate Lalonde, but rather extend the analysis using the more up-to-date methods learned from Imbens and Rubin. They chose the propensity score and published two papers reevaluating the Lalonde data …. The propensity score analysis ultimately did much better than what Lalonde’s analysis had done. A lot of gains were made simply from recognizing the serious common support violations rampant in all six of those datasets”
· On VoxDev, Tanika Chakraborty and Rajshri Jayaraman summarize their work on how midday meals improved learning outcomes in Indian schools. “the Indian Supreme Court issued a directive in 2001 ordering Indian states to institute free midday meals in government primary schools. Over the subsequent five years, state governments across India introduced midday meals. Staggered implementation of the programme in primary schools generates plausibly exogenous variation in the length of potential exposure to the programme based on a child's birth cohort and state of residence. Children only enjoyed the programme to the extent that they were of primary school going age (6 to 10 years old) and lived in a state which had instituted midday meals in primary school… primary school-aged children’s learning outcomes increase, albeit at a decreasing rate, with exposure to midday meals… In the final year of exposure, the effect tapers off slightly to 0.23 points for reading and 0.12 points for math”
· Are you directly involved in collecting data in low income countries? A research project is looking at ethical implications and issues faced, and is looking for those involved (e.g., as data collector, enumerator, PhD, Postdoc, PI or similar positions) to answer a survey to help them gain a better understanding of the multiple challenges that research staff faces when collecting data in low- and middle-income countries, and ultimately contribute to the wellbeing and safety of all research staff. Contact research-ethics-survey@uni-goettingen.de with any questions.
Join the Conversation