Applied Statistics Workshop (Gov3009)

Date: 

Wednesday, March 11, 2015, 12:00pm to 1:30pm

Location: 

CGIS Knaffel 1737 Cambridge St, Cambridge Room K354
The Applied Statistics Workshop (Gov 3009) meets all academic year, Wednesdays, 12pm-1:30pm, in CGIS K354. This workshop is a forum for advanced graduate students, faculty, and visiting scholars to present and discuss methodological or empirical work in progress in an interdisciplinary setting. The workshop features a tour of Harvard's statistical innovations and applications with weekly stops in different fields and disciplines and includes occasional presentations by invited speakers. There is a free lunch provided. Presentation given by James Robins Title: The Foundations of Statistics and Its Implications for Current Methods for Causal Inference from Observational and Randomized Trial Data Abstract: The foundations of statistics are the fundamental conceptual principles that underlie statistical methodology and distinguish statistics from the highly related fields of probability and mathematics. Examples of foundational concepts include ancillarity, the conditionality principle, the likelihood principle, statistical decision theory, the weak and strong repeated sampling principle, coherence and even the meaning of probability itself. In the 1950s and 1960s, the study of the foundations of statistics held an important place in the field. However its central role faded with the revolution in computing that offered the ability to actually do more than just philosophize about how to analyze complex high dimensional data. I discuss how these principles both inform and are informed by modern approaches to causal analysis. Among other examples, I discuss from a foundational perspective are (i) methods for model and/or covariate selection including the issue of whether detailed balance on covariates is needed after one stratifies on the true or estimated propensity, (ii) the conflict between the minimization of MSE versus accuracy of confidence intervals as inferential goals and (iii) the question of a whether principled Baysesian inference must ignore the propensity score even when it is known.