A Study of Modeling Longitudinal Nonresponse Using Paradata in the Survey of Labour and Income Dynamics (SLID)

• Activity Number: 321
• Type: Invited
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Beatrice Baribeau and Wisner Jocelyn
• Companies: Statistics Canada
• Keywords: Paradata ; Nonresponse adjustment ; Longitudinal survey

Abstract

This paper discusses and evaluates methods proposed for nonresponse adjustments including paradata, for the Survey of Labour and Income Dynamics (SLID). The proposed adjustments include variables such as the number of calls to a household and time of call. Methods of evaluation of the nonresponse adjustment include correlation of key auxiliary variables to SLID variables of interest, correlation of key auxiliary variables to response, evaluation of the stability of the weights, examination of pseudo-relative-bias as an indication of potential bias, impact on key estimates and standard errors, and validation of the proposed nonresponse models. Results show both pros and cons for the use for paradata in the nonresponse adjustment for SLID, such as a trade-off of reduction in pseudo-relative bias for an increase in variance.