|| Checking for direct PDF access through Ovid
To determine the quality of proxy health reports by telephone the 1984 Wisconsin Health Status Survey employed a repeated-measure design in an interview covering 22 recently occurring health and psychologic complaints. Comparisons on individuals in households containing two or more adults revealed a relatively weak correspondence between the respondent and proxy reports. While a previous analysis of these data found that certain characteristics thought to underlie reporting differences are not useful in explaining proxy underreporting, the current analysis focuses on spousal pairs and achieves greater success by applying somewhat more sophisticated methods. The authors examined the nature, persistence, and number of health complaints as factors in reporting bias. They find evidence that female proxies vary by symptom in their ability to report common complaints and also observed that proxy underreporting diminishes somewhat as the persistence of symptoms increases. Using multiplicative models, the authors show that the gross misclassification of complaints is concentrated in the respondent-proxy pairs with the shortest exposure to the symptom. Finally, an examination of the disagreements on all 22 health complaints simultaneously revealed that neither mutual misallocation by respondents and proxies nor a diminished health status of the individual reported on are important influences on reporting behavior.