Unintended consequences

Survey researchers commonly use RDD (random digit dialing) samples that are purged of listed business telephone numbers to increase interviewer productivity by removing numbers that are assumed to be ineligible for household surveys. This study investigates this practice and finds an unintended consequence: an increase in household noncoverage. The data come from national RDD surveys using samples that were not purged of listed business numbers. Phone numbers were flagged as listed businesses or not and respondents were asked about how their phone lines are used. Five percent of respondents were interviewed on lines classified as business numbers that normally would have been purged from the sample. But were these valid household interviews or should they have been excluded The data show that these are, in fact, primarily households. Ninety-four percent of these numbers rang at residences. Moreover, these phone numbers are used as household rather than business-only lines. Ninety-three percent said any adult in the household can answer the phone line in question. A more important finding is that business-line purging increases noncoverage. Sixty-five percent of those contacted on numbers that normally would have been purged from the sample said they had no other phone lines in the household - a noncoverage rate of 3.6 percentage points had these numbers been excluded from the sample. The study concludes with an examination of the characteristics of those interviewed on presumed business numbers, and considers the costbenefit implications of including these numbers in the sample.