ACSPRI Conferences, RC33 Eighth International Conference on Social Science Methodology

Font Size:  Small  Medium  Large

Calendar Interviewing in Life Course Research: Associations between Verbal Behaviors and Data Quality

Robert F. Belli, Ipek Bilgen, Tarek Al Baghal

Building: Law Building
Room: Breakout 8 - Law Building, Room 100
Date: 2012-07-11 03:30 PM – 05:00 PM
Last modified: 2012-06-12


Retrospective survey reports have been used widely in the social, behavioral, and health sciences to help draw inferences regarding life course trajectories. A key issue with retrospective reports is response errors, and that these errors challenge the validity of both the data and the inferences drawn from them. An emerging calendar interviewing method shows considerable promise by reliably yielding better quality retrospective reports in comparison to conventional standardized questionnaires. These improvements in data quality are hypothesized to be the result of calendars utilizing what is known about the cognitive structure of autobiographical knowledge to provide effective retrieval cues, and by encouraging conversational aspects calendar interviewing that may also promote the acquisition of quality data. This research is motivated to explore the adequacy of these hypotheses by examining the associations between retrieval and conversational verbal behaviors and data quality measures. A verbal behavior coding scheme was applied to the transcripts of 165 calendar interviews that collected life-course information on residence, marriage, employment, and unemployment from respondents in the Panel Study of Income Dynamics (PSID). Confirmatory factor analyses revealed three latent factors each for interviewer and respondent behaviors. The three factors for interviewers consisted of retrieval probes, rapport behaviors, and conversational behaviors intended to satisfy questionnaire objectives. As for respondents, the three factors consisted of retrieval strategies, rapport, and conversational behaviors indicative of interviewing difficulty. Annual absolute and signed errors between retrospective calendar reports and reports collected for up to 30 years in the history of the PSID were used as measures of data quality. Regression analyses that treated the data quality measures as dependent variables have revealed interactions between levels of behaviors and levels of retrieval difficulty that is estimated on the extent to which one’s past is complicated as revealed in the panel data. Retrieval behaviors are associated with decreases in absolute error and underreporting in circumstances in which the retrieval task is more difficult, but increases in these errors when retrieval difficulty is low. Rapport behaviors are associated with reduced error and underreporting of difficult employment histories but increased error and underreporting with unemployment histories. Patterns of results for conversational and interviewing difficulty behaviors are inconsistent. Results do not completely confirm hypotheses, but nevertheless lead to insights regarding how to improve interviewer training in calendar interviews.