Results of a randomised controlled trial comparing two different incentives to improve survey response rates

Background: Efforts to improve institutional research culture require that we understand current research cultures as experienced by our research communities. Low response rates in qualitative surveys of researcher experience may reduce their usefulness. Various incentives to improve response rates have been proposed, but it is not known whether the chance of a single, large incentive is more effective than a greater chance of a smaller incentive.
Methods: We embedded a randomised controlled trial within the Edinburgh Research Culture Survey, comparing the effect on number of responses received of (1) a single reward of £1000 or (2) a reward of £50 for 1 in every 50 respondents, selected by random draw. Ethical approval was obtained from the University of Edinburgh College of Arts, Humanities, and Social Sciences Research Ethics Committee.
Results: The incentive offered had no effect on the number of responses received (odds ratio 0.94 [95% CI 0.83 to 1.06]), but, having completed the survey, more participants entered the random draw in the £1000 incentive group (odds ratio 1.49 [1.16 to 1.92]). Staff were more likely to complete the survey, while students were more likely to enter the random draws. Most respondents provided answers to almost all of the survey items (IQR 171 to 172 or 175 items), and this did not differ between groups.
Conclusions: In this test of different incentives to increase survey participation, we observed no difference between a moderate chance of a modest reward and a small chance of a high reward. Those conducting such surveys can select their incentive system based on other factors such as cost.