Three experiments on the World Wide Web asked subjects to rate the severity of common health disorders such as acne or arthritis. People who had a disorder ("Haves") tended to rate it as less severe than people who did not have it ("Not-haves"). Two explanations of this Have versus Not-have discrepancy were rejected. By one account, people change their reference point when they rate a disorder that they have. More precise reference points would, on this account, reduce the discrepancy, but, if anything, the discrepancy was larger. By another account, people who do not have the disorder focus on attributes that are most affected by it, and the discrepancy should decrease when people make ratings on several attributes. Again, if anything, the discrepancy increased when ratings were on separate attributes (combined by a weighted average). The discrepancy varied in size and direction across disorders. Subjects also thought that they would be less affected than others.