Consistency in self-reports of HIV testing: longitudinal findings from the National AIDS Behavioral Surveys.

This paper assesses consistency in self-reports of human immunodeficiency virus testing using two waves of longitudinal data from a large, national probability survey, the National AIDS Behavioral Survey. Of those reporting at Wave I that they had been tested for reasons other than blood donation, 18 percent reported at Wave 2 that they had never been tested. Of those reporting at Wave I that they had been tested when they donated blood, 29 percent reported at Wave 2 that they had never been tested. Inconsistent responses may be due to poor recall and to high self-presentation bias, that is, a desire to provide socially acceptable answers. Poor recall may be exacerbated by passive conditions such as blood donation. The authors conclude with recommendations for reducing measurement error in surveys of testing behavior.