Purpose: To compare the estimates of HIV-related risk derived from a sample drawn through random digit dialing and that component of the sample drawn from households listed in a telephone directory.
Methods: The prevalence of key outcomes, and their 95% confidence intervals, was estimated for the total sample and for that component of the sample drawn from households listed in a telephone directory.
Results: On all outcome measures the sample derived from listed households was more conservative. With few exceptions, the estimates derived from the overall sample and from listed households were not significantly different.
Conclusion: A continued reliance on random digit dialing seems unwarranted.