A new report issued by IES: A key purpose of rigorous evaluations of education programs and interventions is to inform policy choices. Typically, such assessments focus on the overall or average treatment effect of the intervention on key outcomes. However, there are also important program and policy questions that pertain to variation in treatment effects across subgroups of study participants...
Illustrations of spurious correlations
Great website:
New – Journal of Observational Studies
A new open-access journal dedicated to observational studies. Normally I am dubious of new journals, until I looked at the ed board: Imbens, Imai, Cook, and bunch of other rock stars. Looks to be a good journal: There are several journals that deal with observational studies in specific fields, but this is the first journal that is devoted to observational studies across the wide range of fields...
Whitehurst reviews the evidence on pre-k programs …
… and the results are depressing.
Causal inference workshops this summer
The basic is at Northwestern, the advanced at Duke; more info here.
I attended the first version of the basic workshop several years ago. It was a bit disjointed (not surprising, given that it was the first iteration). The best aspect of the workshop are the big guns they bring in – I heard lectures by Rubin and Imbens. Well worth the cost.
Matching and complex survey data
Several people have asked me about using propensity score matching with complex surveys, and whether/how weights, PSUs and strata should be taken into account. This paper explains how to handle these issues.
A brief introduction to propensity score matching
I recently started a quant methods working group at the College of Education here, and gave a presentation on propensity score matching for people who had heard about it and wanted to know more.
Causal inference and prestige
In class this week, I mentioned that among researchers there is an implicit prestige hierarchy of analytical techniques, in this order: Random assignment of treatment Regression discontinuity Difference-in-difference (with a change in treatment that is clearly exogenous) Fixed effects panel models and instrumental variables Matching OLS, HLM, logistic regression, etc.(any version of the general...