This paper asks which sub-groups of the population are affected by the payment of a small cash incentive to respond to a telephone survey. We find that an incentive improves response rates primarily amongst those individuals with the longest history of income support receipt. Importantly, these individuals are least likely to respond to the survey in the absence of an incentive. The incentive thus improves both average response rates and acts to equalize response rates across different socio-economic groups, potentially reducing non-response bias. Interestingly, the main channel through which the incentive appears to increase response rates is in improving the probability of making contact with individuals in the group with heavy exposure to the income support system.