The first systematic studies of self-ratings of health (SRH) in the US were made at Duke University beginning in the 1950s. Healthy elderly volunteers were administered repeated medical examinations, and also rated their own health. Data from the 1960s and 1970s showed some association between SRH and physician' ratings, but not a perfect correspondence. When there were differences between the ratings, SRH was more often in the direction of better health than the physician's rating. The older the study subject was, the more likely they were to give an optimistic rating of their health, considering their higher levels of chronic illness and disability. Some of the last studies from this series showed that, over time, the self-ratings were very good predictors of future physicians' ratings. In fact, they were better at predicting future physician ratings than physician ratings were at predicting future self-ratings.