The measurement and interpretation of health in social surveys

a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications and drafts do not necessarily reflect the opinions or policies of its research sponsors. The RAND unrestricted draft series is intended to transmit preliminary results of RAND research. Unrestricted drafts have not been formally reviewed or edited. The views and conclusions expressed are tentative. A draft should not be cited or quoted without permission of the author, unless the preface grants such permission. Health status is hard to measure. It is widely recognized that health is multi-dimensional reflecting the combination of an array of factors that include physical, mental and social well-being, genotype and phenotype influences as well as expectations and information. A multitude of health indicators have been used in scientific studies drawing on data from both the developed and developing world. Understanding what those indicators measure is central if the results reported in the studies are to be interpreted in a meaningful way. Whether one is interested in summarizing the health of a population or understanding the links between health and other measures of well-being at the individual level, poor measurement will likely yield poor inferences. There is a large literature that discusses the validity and limitations of different health measures. Murray and Chen (1992) and Sadana (2001) provide excellent reviews and discussion. Some of the most insightful empirical studies have compared indicators of specific morbidities reported by respondents in health interview surveys with indicators based on health examinations of the same individuals conducted by trained health workers. Other very influential studies have examined the extent to which self-reported health predicts health problems later in life. A third class of studies contrasts prevalence rates based on health interviews with other sources. This paper attempts to provide insights into the meaning of a set of relatively general health indicators that are commonly collected in health interview surveys. These self-reports are contrasted with a battery of physical assessments conducted on the same respondents in the same survey, usually a day or two after the interview. The key complexity that arises from these comparisons is that "true" health status is seldom known. To side-step this complexity, we begin with indicators of health (or, more precisely, nutritional) status which can be measured without any ambiguity: height and weight. Contrasting measurements with self-reports for the same person, we find that self-reports are subject to systematic biases. For example, …