|
Abstract:
|
Web panel samples of patients with mental disorders face multiple problems, including coverage error, self-selection bias, and physicians cannot validate the purported diagnosis. To evaluate these web samples, we administered the same measures of disability to schizophrenic and bipolar patients via three modes: web, telephone, and in-person. Samples for the web and in-person modes were drawn from existing web panels, while the telephone sample was recruited via RDD methodology. Among schizophrenic patients, the web sample yielded significantly higher estimates of disability, and contained higher proportions of young, married, highly educated, White male respondents with full time employment. Mode differences among bipolar patients were much less pronounced. The significant differences in disability between modes were often eliminated after controlling for demographic differences.
|
