A Cautionary Note on the Use of Matching to Estimate Causal Effects: An Empirical Example Comparing Matching Estimates to an Experimental Benchmark

We randomly selected one individual to represent each of the households that had a listed phone number in the file; We then randomly assigned subjects to two different treatment groups: 16,047 subjects were slated to receive a phone call encouraging them to vote on Election Day, and a different 16,037 subjects were slated to receive a phone call encouraging them to buckle up their seatbelts while driving. The remaining 2,747,231 subjects were assigned to the control group and did not receive a phone call.