Ying Wu1, Susan A Newfield. 1. Research Office, West Virginia University School of Nursing, PO Box 9640, Morgantown, WV 26506-9640, USA. ywu@hsc.wvu.edu
Abstract
BACKGROUND: This study assessed whether data-collection formats, computerized versus paper-and-pencil, affect response patterns and descriptive statistics for adolescent health assessment surveys. Youth were assessed as part of a health risk reduction program. METHODS: Baseline data from 1131 youth were analyzed. Participants completed the questionnaire either by computer (n = 390) or by paper-and-pencil (n = 741). RESULTS: The rate of returned surveys meeting inclusion requirements was 90.6% and did not differ by methods. However, the computerized method resulted in significantly less incompleteness but more identical responses. Multiple regression indicated that the survey methods did not contribute to problematic responses. The two survey methods yielded similar scale internal reliability and descriptive statistics for behavioral and psychological outcomes, although the computerized method elicited higher reports of some risk items such as carrying a knife, beating up a person, selling drugs, and delivering drugs. CONCLUSIONS: Overall, the survey method did not produce a significant difference in outcomes. This provides support for program personnel selecting survey methods based on study goals with confidence that the method of administration will not have a significant impact on the outcome.
BACKGROUND: This study assessed whether data-collection formats, computerized versus paper-and-pencil, affect response patterns and descriptive statistics for adolescent health assessment surveys. Youth were assessed as part of a health risk reduction program. METHODS: Baseline data from 1131 youth were analyzed. Participants completed the questionnaire either by computer (n = 390) or by paper-and-pencil (n = 741). RESULTS: The rate of returned surveys meeting inclusion requirements was 90.6% and did not differ by methods. However, the computerized method resulted in significantly less incompleteness but more identical responses. Multiple regression indicated that the survey methods did not contribute to problematic responses. The two survey methods yielded similar scale internal reliability and descriptive statistics for behavioral and psychological outcomes, although the computerized method elicited higher reports of some risk items such as carrying a knife, beating up a person, selling drugs, and delivering drugs. CONCLUSIONS: Overall, the survey method did not produce a significant difference in outcomes. This provides support for program personnel selecting survey methods based on study goals with confidence that the method of administration will not have a significant impact on the outcome.
Authors: M Elle Saine; Julia E Szymczak; Tyler M Moore; Laura P Bamford; Frances K Barg; Jason Schnittker; John H Holmes; Nandita Mitra; Vincent Lo Re Journal: J Viral Hepat Date: 2020-06-23 Impact factor: 3.517
Authors: M Elle Saine; Tyler M Moore; Julia E Szymczak; Laura P Bamford; Frances K Barg; Nandita Mitra; Jason Schnittker; John H Holmes; Vincent Lo Re Journal: PLoS One Date: 2020-02-05 Impact factor: 3.240