C B Phillips1, R Yates, N J Glasgow, K Ciszek, R Attewell. 1. Academic Unit of General Practice and Community Health, Australian National University Medical School, Australian Capital Territory. Christine.Phillips@calvary-act.com.au
Abstract
OBJECTIVE: Supplementary questionnaires provide additional information from subgroups, but may have an attritional effect on response rates. We examined the effects of different instruction methods on response rates to a two-part questionnaire. METHODS: The ACT Kindergarten Health Survey comprises a health questionnaire for all school-entry children. A supplementary questionnaire targets children with respiratory symptoms. We cluster-randomised 109 schools in the ACT (4,494 children) to two instruction groups. Group 1 (instruction burden) had instructions to complete the supplementary questionnaire if certain questions in the primary questionnaire were answered. Group 2 (response burden) had instructions to complete both questionnaires irrespective of answers to the primary questionnaire. RESULTS: Instructing all respondents to complete both questionnaires regardless of eligibility resulted in a statistically significantly lower primary questionnaire response rate (82% vs. 87%), but a statistically significantly higher response rate to the supplementary questionnaire (99% vs. 91%). The net effect was a small overall gain (82% vs. 79%) for the response burden group. CONCLUSION: Increasing the response burden had a minor impact on response rate to the primary questionnaire, but increased the response rate for most items to the supplementary questionnaire. IMPLICATIONS: Large surveys may be broken into primary and supplementary questionnaires, if strategies to maximise response rates are used. Questionnaires may need to be modified to take into account the likely attrition on response to either questionnaire resulting from instruction and response burdens.
RCT Entities:
OBJECTIVE: Supplementary questionnaires provide additional information from subgroups, but may have an attritional effect on response rates. We examined the effects of different instruction methods on response rates to a two-part questionnaire. METHODS: The ACT Kindergarten Health Survey comprises a health questionnaire for all school-entry children. A supplementary questionnaire targets children with respiratory symptoms. We cluster-randomised 109 schools in the ACT (4,494 children) to two instruction groups. Group 1 (instruction burden) had instructions to complete the supplementary questionnaire if certain questions in the primary questionnaire were answered. Group 2 (response burden) had instructions to complete both questionnaires irrespective of answers to the primary questionnaire. RESULTS: Instructing all respondents to complete both questionnaires regardless of eligibility resulted in a statistically significantly lower primary questionnaire response rate (82% vs. 87%), but a statistically significantly higher response rate to the supplementary questionnaire (99% vs. 91%). The net effect was a small overall gain (82% vs. 79%) for the response burden group. CONCLUSION: Increasing the response burden had a minor impact on response rate to the primary questionnaire, but increased the response rate for most items to the supplementary questionnaire. IMPLICATIONS: Large surveys may be broken into primary and supplementary questionnaires, if strategies to maximise response rates are used. Questionnaires may need to be modified to take into account the likely attrition on response to either questionnaire resulting from instruction and response burdens.
Authors: Michelle Boyd; Toby J Lasserson; Michael C McKean; Peter G Gibson; Francine M Ducharme; Michelle Haby Journal: Cochrane Database Syst Rev Date: 2009-04-15