Response bias in community surveys of mental health: Systematic bias or random noise?

Abstract

This paper examines the effects on a mental health survey of three forms of response bias— naysaying, perceived trait desirability and need for social approval. It is suggested that, although these three forms of response bias may be statistically associated with either the independent or the dependent variables, it is mmore meaningful to introduce them as controls on the relation between the independent and dependent variables. Data from a moderately poor, predominately black population suggest that these three forms of response bias are statistically associated with symptoms of depression and other forms of mental illness. Nonetheless, response bias generally does not act as a form of systematic bias that invalidates the pattern of relationships observed with traditional independent variables, but instead acts as random noise.