BACKGROUND: Measuring historical physical activity in epidemiologic research depends on self-report. We aimed to describe data reporting errors women made in completing 2 validated questionnaires: Lifetime Physical Activity Questionnaire (LPAQ) and Occupational Questionnaire (OQ). METHODS: Participants--229 women aged 38 to 65 years--completed questionnaires on paper (n = 160) or by web interface (n = 69). One research assistant collected questionnaire data, identified potential errors and contacted participants to trouble-shoot errors. RESULTS: Women made mean 9.7 (SD 11.2) errors on paper and 7.1 (SD 6.2) errors on electronic versions of the LPAQ and 2.6 (SD 3.8) and 1.1 (SD 1.4) errors on paper and electronic versions of the OQ, respectively. Fewer mistakes were made on electronic versions of both questionnaires combined (8.5 ± 6.1) when compared with the paper versions (12.7 ± 13.1). Only ~2% of the sample completed all questionnaires without detectable errors. The most common errors were reporting activities or frequencies inconsistently between past year survey and the current age epoch, reporting more years than allowed by age epoch and missing information. CONCLUSIONS: Despite the implications of "self-report" questionnaires, we recommend researchers provide participants with additional instructions, either verbally or as written tip sheet or both, and follow-up after questionnaire completion to correct mistakes as needed.
BACKGROUND: Measuring historical physical activity in epidemiologic research depends on self-report. We aimed to describe data reporting errors women made in completing 2 validated questionnaires: Lifetime Physical Activity Questionnaire (LPAQ) and Occupational Questionnaire (OQ). METHODS:Participants--229 women aged 38 to 65 years--completed questionnaires on paper (n = 160) or by web interface (n = 69). One research assistant collected questionnaire data, identified potential errors and contacted participants to trouble-shoot errors. RESULTS:Women made mean 9.7 (SD 11.2) errors on paper and 7.1 (SD 6.2) errors on electronic versions of the LPAQ and 2.6 (SD 3.8) and 1.1 (SD 1.4) errors on paper and electronic versions of the OQ, respectively. Fewer mistakes were made on electronic versions of both questionnaires combined (8.5 ± 6.1) when compared with the paper versions (12.7 ± 13.1). Only ~2% of the sample completed all questionnaires without detectable errors. The most common errors were reporting activities or frequencies inconsistently between past year survey and the current age epoch, reporting more years than allowed by age epoch and missing information. CONCLUSIONS: Despite the implications of "self-report" questionnaires, we recommend researchers provide participants with additional instructions, either verbally or as written tip sheet or both, and follow-up after questionnaire completion to correct mistakes as needed.
Authors: Lisa Chasan-Taber; J Bianca Erickson; Jeanne W McBride; Philip C Nasca; Scott Chasan-Taber; Patty S Freedson Journal: Am J Epidemiol Date: 2002-02-01 Impact factor: 4.897