|
Abstract:
|
Randomized and observational studies each have strengths and limitations for estimating parameters in a target population of interest. Estimates from randomized data may have internal validity but are not generalizable to the target population. Observational data may be more likely to have external validity, but are typically affected by unmeasured confounding. Intersecting with issues of generalizability are fairness considerations. Who is in the target population and what societal biases are reflected in the data are critical issues. This talk will discuss new statistical machine learning developments in these areas and their relevance for biobehavioral research, including an application in mental health policy.
|
