A Reply to Gambino's "An Epidemiologic Note on Verification Bias: Implications for Estimation of Rates"

While two-stage designs in problem gambling research are expensive and relatively difficult to execute, they have the potential to yield more accurate prevalence estimates as well as valuable additional information about gambling and problem gambling in the community. However, gains in precision are heavily dependent on the accuracy of the screening methods used and the sample sizes involved. Sound practice requires reliable information about the variances and confidence levels associated with different screens. In the spirit of contributing to the ongoing dialog about ways to improve the measurement of problem gambling, we examine several reasons to question whether the revised estimates offered by Gambino are in fact an improvement.