Content area
Full Text
UTILITY OF HEALTH DATA FROM SOCIAL SURVEYS: IS THERE A GOLD STANDARD FOR MEASURING MORBIDITY?*
Most sociological and epidemiological studies of health status in adulthood rely on reports of morbidity from respondents to social surveys. This study compares self-reported morbidity with indicators of morbidity from physicians' evaluations and examines the predictive validity of each indicator on self-assessed health and mortality in adulthood. Special attention is given to differences in the measures between white and African American adults. Adults from a large national survey received a detailed medical examination by a physician; they also were asked about the presence of 36 health conditions. Results indicate that self-reported morbidity is equal or superior to physician-evaluated morbidity in a prognostic sense. Both types of morbidity predict self-assessed health for white respondents, but physician-evaluated morbidity is not related to either self-assessed health or mortality for African American respondents.
Reliance on survey research methods by social scientists in the past half century, coupled with the widespread availability of data archives, means that much empirical knowledge in the social sciences depends on survey methods. In many situations, survey methods are the data collection strategy of choice, especially when coupled with appropriate sampling procedures. For studying quality-of-life indicators, attitudes, or characteristics of the self, survey methods may provide the most useful data because respondents are the best source of information for phenomena that typically "have no objective external referent" (Angel and Gronfein 1988:464). On the other hand, for researching topics such as church attendance or drug or alcohol consumption, survey methods may not be optimal because what respondents report may not accurately reflect objective realities (Bradburn, Rips, and Shevell 1987; Hadaway, Marler, and Chaves 1993). While relatively few investigators have examined how respondent self-reports of factual information correlate with other information sources, evidence is growing to show that respondent self-reports are influenced by contextual clues and respondent motivation (Nadeau and Niemi 1995; Presser and Traugott 1992).
The accuracy of survey respondents' reports of objective conditions affects virtually all fields of sociology. This paper examines this phenomenon with health data from physicians and survey respondents. Our analysis focuses on morbidity, a phenomenon that has an objective external referent. Some symptoms of disease may be difficult to corroborate from other sources, but...