| Abstract: | Six hundred and twenty-six participants from the Panel Studyof Income Dynamics (PSID) were interviewed via telephone retrospectivelyabout life course events by either computer-assisted (CATI)event history calendar (EHC) or standardized CATI conventionalquestionnaire (CQ) methods, randomly assigned. Experimentalretrospective reports, for a reference period up to 30 years,were validated against reports provided annually from the PSIDcore interviews. Data quality outcome measures included variablesassociated with marriage, cohabitation, employment, unemployment,residential changes, and cigarette smoking. The EHC providedhigher quality retrospective reports for cohabitation, employment,unemployment, and smoking histories; the CQ provided betterdata quality for marriage history, although what variable wasbeing measured, instead of which method was being used, hadthe biggest impact on differences in data quality. Both EHCand CQ interviews lasted on average around one hour, with theEHC interviews being on average 10 percent longer. Interviewerspreferred the EHC interviews. In both EHC and CQ conditions,respondents generally enjoyed the interviews, and did not findquestions difficult. The costs and benefits of both EHC andCQ methods in the collection of life course retrospective reportsare discussed. Received for publication December 19, 2005. Revision received September 25, 2007. Accepted for publication September 25, 2007. |
