Testing the Question Behaviour Effect Hypothesis Versus the Non-reponse Bias Hypothesis
A. van Dongen1, R.A.C. Ruiter2, C. Abraham3, I. Veldhuizen1
1Sanquin Research, Amsterdam, the Netherlands
2Maastricht University, Maastricht, The Netherlands
3University of Exeter Medical School, Exeter, UK
Introduction We study whether previous findings in studies on blood donation behaviour that were labelled question behaviour effect (QBE) could actually reflect a non-response bias (NRB) instead. Higher response rates have also been associated with finding QBE. Methods Blood donors (N=1,395) were randomly allocated to one of six experimental questionnaire conditions or to a control group. The questionnaire conditions either included intention items to trigger a QBE, general health items, both items, and handwritten post-its to increase response rate or not. The outcome variable was the % of donors that donated after receiving a standard invitation. Results The questionnaire conditions did not influence the odds of donation compared to the control group (OR’s ranged from 0.70 to 1.01). Non-responders showed significantly lower odds of donating than controls (ORresponders: 1.05, 95%CI: .77-1.45; ORnon-responders: .63, 95%CI: .45-.88). Discussion Regardless of the content or response rate, receiving or filling in questionnaires does not seem to increase blood donation. However, non-responders donate less than responders. These findings support the NRB hypothesis over the QBE hypothesis.