Relevance & Research Question: Rewards are ubiquitous in online research. In online access panels the type and value of rewards may not be known to the client. It is generally believed that rewards boost response rates. However response rates in online research are simply poor, with or without rewards. The value to the online access panel supplier is economic not methodological. With increased price compression in online sampling it is timely to consider the methodological implications of manipulating reward levels downwards.
Methods & Data: 20 treatments were undertaken; 5 reward levels by 4 interview lengths. Two cases of zero rewards were considered – one, the reward level is simply absent, the other clearly stated 0 points would be given. Matched samples of 2000 per cell were invited to the study. The body of the questionnaire was a battery to measure altruism (Rushton et al), a series of statements relating to attitudes to survey taking and rewards along with perception questions of the rewards on offer.
Results: The results demonstrated only marginal gains in response resulted from substantial increases in rewards levels. E.g. doubling the incentive from $5 to $10 increased response by a mere 14%. This is partly due to the invisibility of rewards. Only 60% thought they knew how much reward they were getting and, of these, only 60% were correct in their estimate. At the same time sample profiles, in terms of distribution of levels of altruism, were unaffected by rewards or interview length. Whilst this holds out promise for reducing rewards spend without impacting response or sample psychographics it does leave as an open question the problem of expectation of rewards – which is being set at the recruitment stage. We may then “get away” with reducing rewards in the short term, it may have a catastrophic long term impact unless we find new ways of recruiting respondents. That itself may have much more far reaching consequences for sample compositions.
Added Value: This paper enables researchers to make informed choices about rewards; not only the impact on response rates but also sample composition over and above demographics.