TIME VARIATIONS IN INTERVIEW QUALITY - An Analysis Based Upon Interviews In a Block-Quota Sampled Field Survey

From 1278 interviews and over 3200 interview attempts, data were obtained to assess the effect of interviewer experience and the effect of various designated periods in time of interviewing and solicitations for interviewing upon the quality of the interviews obtained. The data used were from a block-quota sampled survey of Midwest United States semi-urban adult residents. The indexes used in measuring interview quality were divided into two basic categories: (1) the solicitation indexes, which served to indicate the degree to which a representative sample of the survey population had been obtained; and (2) the interaction quality indexes, which served to indicate the degree to which accurate and sufficient information had been registered in the interview schedules. Two hypotheses were tested. The first was that interview quality underwent no significant variation on the basis of the level of interviewer experience. This hypothesis was rejected since, in six of the eleven quality indexes, the proportions of mean values falling above the mean of the means for the first twenty-five interviews conducted as compared to the second twenty-five interviews conducted were found to be different at least beyond the 10% level of significance (two beyond the 2% level). In addition, results were not found to be generally inconsistent with available results from other interview quality studies. The second hypothesis was that interview quality underwent no significant variation on the basis of the time of day or the day of the week the interviews were conducted. Like the first hypothesis, it was rejected, though with less certainty of the meaningfulness of doing so. Statistically, many variations in mean index values were found significant beyond the 5% level (four were significant beyond the 0.3% level). Nevertheless, the few results from other studies which were more or less comparable were not in accordance with findings in the present study, and the various quality indexes of this study appeared to fluctuate in mean values rather independently, thus not being in any obvious agreement with one another. A few minor practical suggestions for future interview surveying projects seemed justified and were proposed in the case of findings relevant to the first hypothesis. Directions for useful future research were outlined regarding the further assessment of the effects of interviewer experience on interview quality. Practical suggestions pertaining to results from testing the second hypothesis were judged to still be premature; but some suggestions for future research into the effects of time of interviewing on interview quality were proposed