|
Abstract:
|
Like other longitudinal household surveys, the Survey of Income and Program Participation (SIPP) experiences challenges maintaining sample balance across waves. To combat this, SIPP implemented an adaptive survey design experiment in Wave III aimed at facilitating field interviewers' work and increasing sample balance by prioritizing under-represented cases and de-prioritizing over-represented cases. Because SIPP interviewers attempt to interview all original sample members, the experiment also prioritized movers, historically hard-to-interview cases. The results of R-indicators and business rules determined the prioritization of sample cases, and case management systems supplied field interviewers with cases labeled "low," "medium," or "high" priority on their laptops. All SIPP interviewers were randomly assigned to treatment or control. Control interviewers saw only medium priority cases, while treatment interviewers were eligible to see high, medium, and low priority cases. Results compare average number of contact attempts, contact rates, and completion rates daily across low, medium, and high cases as well as between treatment and control FRs.
|
