Mon, Feb 13, 2017, noon:
Daniel Almirall, "Getting SMART about adaptive interventions"
As longitudinal measurement becomes more intensive – particularly with the increase in Internet-based interviewing – convincing respondents to continue participating may become more difficult. We analyze data from a weekly longitudinal panel to identify factors in the rate of continued participation and the probability of on-time response to weekly surveys. We find the same individual-level characteristics that typify continued participation in less frequent longitudinal data collection (e.g., race, SES) also predict continued participation in this study. These variables, along with personality characteristics, also influence on-time responses to weekly surveys. Minority, low-SES, extroverted, and less conscientious respondents are less timely. But we also find that some factors central to the study – for example, having many sexual partners during the study – were associated with both more completed interviews and late interviews. Changes in behaviors key to our study – such as sexual partners, contraceptive use, and pregnancy – are associated with a delay in the subsequent interview.
Country of focus: United States of America.