We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Online ordering is currently unavailable due to technical issues. We apologise for any delays responding to customers while we resolve this. For further updates please visit our website: https://www.cambridge.org/news-and-insights/technical-incident
Marc Rotkovic (Princeton University) and Dustin Tingley (Harvard University)
Selection Committee: Michael Peress (Stonybrook), Suzanna Linn (Penn State), Brandon Stewart
Citation
Ratkovic and Tingley "Sparse Estimation with Uncertainty: Subgroup Analysis in Large Dimensional Designs" proposes a new method, LASSOPlus, for uncovering conditional, subgroup effects in experiments with multiple treatments and pretreatment covariates. LASSOPlus does three important things: it selects relevant effects through a sparse model, automatically selects the degree of regularization and, in contrast to many prior methods, provides uncertainty estimates through a Bayesian posterior. The paper presents the model, uses simulation evidence to demonstrate its performance relative to a set of alternative strategies, and applies the method to prior work by Bechtel and Scheve (2013). The authors also provide the sparsereg R package. The methodological contribution of the paper coupled with a strong readily available implementation ensures that this work will have a big impact on the field.