ABSTRACT
All measurements, whether generated in the laboratory or in the field, are vulnerable to error. Yet all too often, public health policymaking and practice base decisions on such empirical data without taking their methodological limitations fully into account. For example, findings from studies that use small samples of convenience are often given the same weight as the results from larger, population-based samples. And the reporting bias that may afflict self-disclosure of information is not always explored and assessed. Of concern to scientists attending the 1996 National Institutes of Health (NIH) conference and contributing to this volume is the quality of self-reported measurements found in many health studies that address such factors as symptoms, exposures, compliance with treatment, and utilization of health care services.
