Jan 1, 1981
This report examines three questions about response errors in sensitive topic surveys: extent of the problem; impact of the problem; and overcoming the problem. The basic findings are that response errors do exist, that they can have important effects on statistical estimates, and that strategies are available to overcome the estimation biases caused by response errors. The authors conclude that surveys can be useful in policy research on sensitive topics when designs and analyses make appropriate provisions for response errors. Design thinking, however, needs to shift away from a preoccupation with average response bias toward a recognition of the effects of variable response errors and the appropriate mix of design and analysis procedures to minimize their effects on policy conclusions.