Propensity score methods attempt to make observational data look like a randomized trial. But there are some big limitations that need to be considered before we jump on the bandwagon. For the video version, click here.