Estimating Dynamic Treatment Effects from Project STAR (Working Paper 36)

This paper considers the analysis of data from randomized trials which offer a sequence of interventions and suffer from a variety of problems in implementation. In experiments that provide treatment in multiple periods (T>1), subjects have up to 2^{T}-1 counterfactual outcomes to be estimated to determine the full sequence of causal effects from the study. Traditional program evaluation and non-experimental estimators are unable to recover parameters of interest to policy makers in this setting, particularly if there is non-ignorable attrition. We examine these issues in the context of Tennessee's highly influential randomized class size study, Project STAR. We demonstrate how a researcher can estimate the full sequence of dynamic treatment effects using a sequential difference in difference strategy that accounts for attrition due to observables using inverse probability weighting M-estimators. These estimates allow us to recover the structural parameters of the small class effects in the underlying education production function and construct dynamic average treatment effects. We present a complete and different picture of the effectiveness of reduced class size and find that accounting for both attrition due to observables and selection due to unobservable is crucial and necessary with data from Project STAR

Hall conditions for edge-weighted bipartite graphs

A weighted variant of Hall's condition for the existence of matchings is shown to be equivalent to the existence of a matching in a lexicographic product. This is used to introduce characterizations of those bipartite graphs whose edges may be replicated so as to yield semiregular multigraphs or, equivalently, semiregular edge-weightings. Such bipartite graphs will be called semiregularizable. Some infinite families of semiregularizable trees are described and all semiregularizable trees on at most 11 vertices are listed. Matrix analogues of some of the results are mentioned and are shown to imply some of the known characterizations of regularizable graphs.

Familiarity and Preference for Pitch Probability Profiles

We investigated familiarity and preference judgments of participants toward a novel musical system. We exposed participants to tone sequences generated from a novel pitch probability profile. Afterward, we either asked participants to identify more familiar or we asked participants to identify preferred tone sequences in a two-alternative forced-choice task. The task paired a tone sequence generated from the pitch probability profile they had been exposed to and a tone sequence generated from another pitch probability profile at three levels of distinctiveness. We found that participants identified tone sequences as more familiar if they were generated from the same pitch probability profile which they had been exposed to. However, participants did not prefer these tone sequences. We interpret this relationship between familiarity and preference to be consistent with an inverted U-shaped relationship between knowledge and affect. The fact that participants identified tone sequences as even more familiar if they were generated from the more distinctive (caricatured) version of the pitch probability profile which they had been exposed to suggests that the statistical learning of the pitch probability profile is involved in gaining of musical knowledge.