An individual's inequality aversion (IA) is a central preference parameter that captures the welfare sacrifice from exposure to inequality. However, it is far from trivial how to best elicit IA estimates. Also, little is known about the behavioural determinants of IA and how they differ across domains such as income and health. Using representative surveys from England, this paper elicits comparable estimates of IA in the health and income domains using two alternative elicitation techniques: a direct trade-off and an indirect "imaginary-grandchild" approach that results from the choices between hypothetical lotteries. We make three distinct contributions to the literature. First, we show that IA systematically differs between income and health domains. Average estimates are around 0.8 for income IA and range from 0.8 to 1.5 for health IA. Second, we find that risk aversion and locus of control are central determinants of IA in both income and health domains. Finally, we present evidence suggesting that the distribution and comparison of IA vary depending on the elicitation method employed.