Income and racial disparities in energy burden are a concern. Most program administrators have relied on survey data to assess whether their programs are serving the customer population equitably. But when survey-takers decline to provide information on income and race, the information those surveys provide may be compromised. If the likelihood of declining to provide that information is related to income and/or race, then the information that was provided is not representative of the sample as a whole. This is a genuine concern, as income and race are among the items that survey-takers are most likely to decline to provide. How do we determine the extent of the problem? Recent research indicates that Census demographic data on the areas where program participants live may provide an acceptable proxy for the characteristics of the participants. We, therefore, used Census data on income and race by zip code to assess the likelihood that declining to respond to questions on income and race are related to those characteristics. We conducted these analyses in two program administrator territories, in separate states, with very different racial and income profiles. In one territory, where 5% of survey-takers declined to identify their race and 10% declined to disclose their income, the likelihood of declining to identify either income or race was much higher in more-white and higher-income geographies. In the other territory, where 3% of survey-takers declined to identify their race and 24% declined to disclose their income, we found minimal differences. The latter territory was much more racially homogeneous, and tended to have higher incomes, than the former. These findings support the view that declining to provide income and race may bias information on those characteristics, although the extent to which that occurs may be related to the degree of racial and/or income homogeneity.