Spring 2014 Seminar Series

DATE: 2/27/2014
LOCATION: E51-149
TIME: 4:15pm
Reception immediately following

SPEAKER:
Abba Krieger

TITLE
Matching on-the-fly: Sequential Allocation with Higher Power and Efficiency

ABSTRACT
In clinical trials and online experimentation (e.g., MTurk) one wants to assign patients to “treatment” or “control” at the time the patient enters the system. The gold standard is random assignment, but if the sample size is modest, then this might result in imbalances, either in the number assignment to each arm, or the values of the covariates between treatment and control. We present a dynamic allocation procedure that maintains the mathematical properties of randomization and that leads to increase in power and efficiency when measuring an average treatment effect. The simple idea is to consider that subjects arrive iteratively and are either randomized to one of the treatment arms, or paired via a matching criterion to a previously randomized subject and administered the alternate treatment. We develop estimators for the average treatment effect that combine information from both the matched pairs and unmatched subjects. Simulations illustrate the method’s higher power and efficiency over several competing allocation procedures for simulated data as well as data from a clinical trial. The approach has tuning parameters. A discussion of the optimal approach to setting these parameters will also be discussed.