Measurement Error and Self-Reported Delinquency - An Examination of Interviewer Bias (From Measurement Issue in Criminal Justice, P 59-70, 1983, Gordon P Waldo, ed. - See NCJ-92338)

• NCJ Number: 92341
• Author(s): F Esbensen
• Date Published: 1983
• Length: 12 pages
• Grant Number(s): MH-27552; 78-JN-AX-0003
• Sponsoring Agency: Chicago Law Enforcement Study Group; US Securities and Exchange Cmssn
• Sale Source: Sage Publications, Inc
• Language: English
• Country: United States of America

Annotation

The characteristics of interviewers can bias data gathered from self-report surveys of juvenile delinquency.

Abstract

Careful training and monitoring of interviewers may reduce the amount of bias, but these precautions do not eliminate interviewer bias. Knowledge of interviewer variables can increase the explained variance from 8 percent to 24 percent, depending on the dependent variable. Study data came from the fifth year interviews of a longitudinal study involving a national probability sample of youths. The study examined the responses of 1,488 youths and the characteristics of 69 interviewers. It hypothesized that interviewers' sex, age, social class, race, and attitudes might influence the responses received. Females and older interviewers tended to record more socially acceptable responses than did males and younger interviewers. Greater social class distance resulted in a greater reporting of minor theft and assault, but this effect was negligible for the two felony scales used. Differences in the races of interviewers and respondents also affected the findings. Mixed results were obtained regarding the hypothesis that interviewers' prior expectation affected results. Larger effects might have been found in the first year of interviewing, since both interviewers and respondents in this study were familiar with the study process and the questions. Data tables, notes, and 34 references are provided.