A Weight-Trimming Approach to Achieve a Comparable Increase to Bias Across Countries in the Programme for the International Assessment of Adult Competencies

• Activity Number: 41
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Wendy Van de Kerckhove*+ and Leyla Mohadjer and Thomas Krenzke
• Companies: Westat and Westat and Westat
• Keywords: Weighting ; extreme weights ; calibration

Abstract

Weight trimming can be used to reduce sampling variance and the impact of influential cases. However, it also introduces bias into the survey estimates. In surveys with an emphasis on comparing estimates across countries or states, it is desirable that the impact on bias be consistent. This was the case for the Programme for the International Assessment of Adult Competencies (PIAAC), an international adult literacy survey. To limit the number of cases trimmed and help achieve a comparable increase to bias across countries, a modification was made to the common method of trimming weights that are over k times the median. Rather than using a constant value of k, the factor was based on a function of the coefficient of variation (CV) of the country's weights. This paper will describe the trimming procedure for PIAAC and evaluate the resulting effect on bias and variance.