Randomized, Controlled Trial Comparing Household Response and Completion Rates Using 1 and 2 US Dollar Unconditional Incentives in A Mail Push-to-Web Survey

Abstract Household surveys are an important means of obtaining standardized data; unfortunately, response rates have been declining for decades. Evidence based on social exchange theory supports the use of small, unconditional monetary incentives (pre-incentives) to increase survey response and completion rates. However, the optimal pre-incentive amount is unclear. Our project team was contracted to conduct a large survey (census) of all households in Indiana at which an individual with a developmental disability who was on the waitlist for certain state-based services resided. In the absence of a clear answer from the literature about what pre-incentive amount to use, we preregistered a sub-study to obtain clarity. We conducted a randomized controlled trial comparing a $1 cash pre-incentive to a $2 cash pre-incentive in a large, mail push-to-web household survey in the midwestern US, examining both response rates and completion rates (>80 percent of the survey). We hypothesized that a $2 pre-incentive would result in significantly more responses and completed surveys than a $1 pre-incentive. Using 1:1 blinded allocation, 6,525 households were randomized to receive either a $1 or $2 cash pre-incentive. All other procedures were identical between study arms. Of the 960 valid responses, 467 were from Arm 1 ($1) and 493 were from Arm 2 ($2). Of those, there were 798 complete responses (>80 percent of the instrument). We did not find sufficient evidence to conclude that there was a difference in either response (OR = 1.07, p = .361, 95 percent CI [0.929, 1.223]) or completion (OR = 1.07, p = .362, 95 percent CI [0.924, 1.242]) rates depending on the amount of pre-incentive included in the invitation. Although we provide multiple caveats and encourage caution when interpreting these findings, we suggest that individuals planning to use a small pre-incentive to encourage survey responses should not by default assume that a $2 incentive is superior to a $1 incentive.