• Presentation

Maximizing Questionnaire Completions and Minimizing Item Nonresponse in a School-based Web Survey of High School Students

Citation

Burns, L. J., Pratt, D. J., Joshua, S. P., & Dever, J. A. (2011, May). Maximizing Questionnaire Completions and Minimizing Item Nonresponse in a School-based Web Survey of High School Students. Presented at AAPOR 2011, Phoenix, AZ.

Abstract

The High School Longitudinal Study of 2009 (HSLS:09), conducted for the National Center for Education Statistics (NCES), tracks more than 21,000 U.S. 9th graders through their high school years and beyond. The main focus of HSLS:09 is to understand students‘ academic goal-setting and decision-making regarding postsecondary educational destinations and career choices. The survey emphasizes especially the STEM areas (science, technology, engineering and mathematics). AAPOR 66th Annual Conference Abstracts Page 122 of 274 School-based surveys of secondary school students are subject to strict time limitations not placed on many other types of surveys. The questionnaire and, in this case, math assessment must be completed by students within the time set aside during the school day by the school‘s administration. This creates a challenge for questionnaire developers who wish to maximize the amount of information that is collected while maintaining high response rates for items at the end of the questionnaire. HSLS:09 mitigated the degree of item nonresponse on the web-based questionnaire with two approaches. First, the order in which modules at the end of the questionnaire were administered was varied; students were randomly assigned to one of three groups. Second, students who had not been able to complete the questionnaire in the first part of the in-school session were able to resume work on the questionnaire if time allowed after the math assessment had been completed. In this paper we evaluate the effectiveness of these approaches by comparing item nonresponse rates with other NCES school-based surveys that used paper and pencil questionnaires and therefore were not able to employ these strategies.