An experimental comparison of two self-report methods for measuring lambda

Criticisms of the RAND Second Inmate Survey have implied that missing and ambiguous responses and problems related to trying to measure self-reported crime rates over extended periods of time may have led to inflated estimates of λ. In the present study, the authors randomly assigned prison inmates to two groups, one to be interviewed using the RAND method for measuring crime rates, and one to be interviewed using the authors' modified month-by-month reporting method. The authors expected that their month-by-month method would produce lower estimates of λ. They found that the two distributions of λ did not differ significantly from each other, suggesting that the RAND results are very robust.