Influence of Skip Patterns on Item Non-Response in a Substance Use Survey of 7th to 12th Grade Students

• NCJ Number: 229334
• Journal: Journal of Drug Education Volume: 39 Issue: 2 Dated: 2009 Pages: 167-180
• Author(s): Kele Ding, Ph.D.; R. Scott Olds, Ph.D.; Dennis L. Thombs, Ph.D.
• Date Published: 2009
• Length: 14 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This retrospective case study assessed the influence of a non-response error on a substance-use (cigarette, alcohol, and marijuana) survey of 7th-12th grade students on their subsequent response to questionnaire items that assessed adolescent alcohol and marijuana use.

Abstract

"Item" non-response as defined in this study is the "failure to obtain substantive answers to individual survey questions in the course of filling out the questionnaire." Item non-response is believed to have many causes, including survey mode, question topic, question structure, question difficulty, respondent attributes, and institutional requirements and policies. Questionnaires with formats that instruct youth to skip sets of items based on no prior involvement in drug use may create frustration and facilitate item non-response as well. Skip-patterns survey designs are intended to reduce response burden by allowing the respondent to skip irrelevant items. The current study found that a large number of 7th-12th grade respondents committed item non-response error after they completed the first skip questions (i.e., in the cigarette-use section). Compared to the non-response errors in other sections of the questionnaire, the errors from the cigarette-use items were much higher, especially at the high end (50 percent-100 percent). It is reasonable to attribute this level of non-response error to the first-time appearance of the skip pattern in the questionnaire. This study suggests that skip patterns may be responsible for underestimating rates of adolescent alcohol and marijuana use. These effects appear to be much less pronounced, however, among students with high academic standing. Researchers whose goal is to establish the parameters of substance use in local communities and schools should carefully consider the design of self-report questionnaires. 5 tables, 20 references, and appended sample page of skip pattern questions on cigarette use