A Study of Nonresponse Adjustment for Subject Missing: A TEPS Case Study

• Activity Number: 244
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Hong-Long Wang*+ and Ting-Hsiang Lin and Lyinn Chung and Shu-Yi Lei
• Companies: National Taipei University and National Taipei University and National Taipei University and National Taipei University
• Keywords: subject missing, missing mechanism, MCAR, MAR, TEPS, weighting adjustment.

Abstract

Subjects missing in a survey may be not random and cause inference bias. Non-response adjustment maybe needed to reduce the bias. However, missing mechanism will be needed before any non-response adjustment. In this paper, through a study of "Taiwan Education Panel Survey" (in short as TEPS), we will explore missing mechanism for subjects missing on 2nd wave data of TEPS. Distinguish missing completely at random (MCAR) from missing at random (MAR). To explore the variables causing MAR, we randomly select 100 data sets from the complete part. Each of the selected data set, with the sample size twice of the missing part, will combine with missing part and run logistic models to find the significant variables. Stability of the model will be discussed. For the non-response adjustments, we will apply weighting adjustment and list-wise delection and compare the result with the baseline.