To promote high levels of performance by NCVS interviewers, an intervention of refresher training and performance monitoring was phased in by a randomized experiment. Interviewer teams were randomly assigned to two cohorts. Those in the first cohort were trained in late 2011, and after training, supervisors monitored interviewer performance using an expanded set of data quality indicators. For the second cohort, training and monitoring were delayed until 2012. Effects of the intervention were estimated by comparing outcomes for the two cohorts in the latter months of 2011, using statistical models that account for the experimental design. We estimate that the intervention

• raised the apparent rate of household property crime by 28%, a result that is statistically significant, and
• raised the apparent rate of violent crime by 19%, a result that is not statistically significant.

The effects of the intervention were larger and more significant for crimes that had not been reported to police. These effects, which were evident in the first cohort in the latter months of 2011, did not affect any of the NCVS crime estimates published for 2011, because those estimates were based only on pre-intervention interviews.