Consistency and Accuracy in the Measurement of Daily Contacts: Surveys and Contact Diaries Compared
Survey research has relied on statistical tools to analyze how consistent behavioral items are. It would be more accurate, however, to examine the items with direct evidence, whenever such evidence is available. Using "daily contact" as the subject, this paper evaluates survey results against actual contact records complied from contact diaries. Data included three separate sets that 52 informants completed in 2004 in the following order: (1) a self-administered survey (the first survey), (2) diary-keeping of all interpersonal contacts for three months (n=4,776 days of data about average daily contacts), and (3) a telephone survey (the second survey, taken two months after the informants started keeping the contact diaries). The surveys contained a single-item question about the range of daily contacts (that included six ordinal response categories) and derivative questions about the composition of these contacts. Measured against their actual contact records, half of the informants made a precise estimate in the first survey about the range of their daily contacts. In the second survey, about three-fourth of the informants estimated their daily contacts correctly. The informants also assessed the gender composition among their daily contacts more precisely than they did with the proportion of the contacts that they initiated. Men and women tend to perceive daily contacts differently: men overestimate the volume of their daily contacts and the proportion of initiative contacts; on the contrary , women underestimate their daily contacts and the initiative contacts. Such subtle differences seem to reflect differential expectations of the gender role in the larger society. Using the diary records as the criterion, the study reveals evidence that justifies the use of ordinal response categories in the single-item measurement of daily contacts in large-scale surveys.