Postdoc with Liz Stuart on propensity score methods when the covariates are measured with error

Liz Stuart sends this one along:

Johns Hopkins University Post-Doctoral Fellow Opening

The Department of Mental Health invites applications for a post-doctoral research fellow in Statistical Methods for Mental Health Research, to be supervised by Dr. Elizabeth Stuart. The successful applicant for this position will work on an NIH funded study to develop and evaluate propensity score methods when the covariates are measured with error, or when datasets contain similar, but not identical, measures of covariates of interest. The post-doc will work with a team of researchers on development of new statistical methods, analysis of existing data, Monte Carlo simulations, and will take the lead on scientific publications. Opportunities will be provided for training in causal inference, missing data, latent variable methods, the evaluation of statistical methods, and the application of statistics to mental health and education. The three particular motivating examples will be: early intervention for children with autism, prevention of perinatal depression, and prevention of dementia. The position also offers close contact with the Department of Biostatistics at Johns Hopkins Bloomberg School of Public Health.

Candidates must have a thorough grasp of basic and advanced multivariate analyses, experience programming in R, and a doctoral degree in statistics, biostatistics, education, psychology, or a related discipline (ABD candidates will also be considered). Experience with propensity scores, multiple imputation, Monte Carlo simulations, and a track record of publications in the social sciences are desired. This is a two year position and salary will depend on qualifications. Considerations of applications will continue until the position is filled.

Applicants should submit a cover letter, CV, and references to: Elizabeth Stuart, Department of Mental Health, Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, [email protected], www.biostat.jhsph.edu/~estuart

This looks like important stuff. On conceptual grounds, I still have warm feelings toward a Gaussian process version of Jennifer’s include-’em-all-and-let-Bart-sort-’em-out approach, but I know that propensity score methods are popular, and just about every model is improved by throwing in a measurement model. Also, Liz is great, in both her applied and theoretical work.