IMPROVING INTERVIEW METHOD AND QUESTIONNAIRE DESIGN - RESPONSE EFFECTS TO THREATENING QUESTIONS IN SURVEY RESEARCH

• NCJ Number: 69523
• Author(s): N M BRADBURN; S SUDMAN
• Date Published: 1979
• Length: 230 pages
• Sale Source: Jossey-Bass Publishers
• Language: English
• Country: United States of America
• Note: JOSSEY-BASS SOCIAL AND BEHAVIORAL SCIENCE SERIES NORC (NATIONAL OPINION RESEARCH CENTER) SERIES IN SOCIAL RESEARCH

Annotation

RESPONSE EFFECTS IN SURVEYS ARE IDENTIFIED AND MEASURED, AND PROCEDURES THAT YIELD THE MOST ACCURATE REPORTING ARE SUGGESTED.

Abstract

THE MATERIAL IS DESIGNED TO HELP USERS AND GATHER OF SURVEY DATA OVERCOME THE PROBLEM OF DISTORTIONS IN SURVEY DATA. FINDINGS ARE REPORTED FROM A SERIES OF EXPERIMENTAL STUDIES, INCLUDING TWO LARGE NATIONAL SAMPLES OF THE GENERAL POPULATION AND A LARGE SAMPLE IN CHICAGO CONDUCTED JOINTLY BY THE NATIONAL OPINION RESEARCH CENTER AND THE SURVEY RESEARCH LABORATORY. INFORMATION IS PROVIDED ON HOW UNDERREPORTING AND OVERREPORTING OF BEHAVIOR VARIES ACCORDING TO A QUESTION'S DEGREE OF THREAT AND A RESPONDENT'S CHARACTERISTICS. TESTED DATA ON HOW VARIATIONS IN QUESTION CONTENT, METHOD OF ADMINISTRATION, AND INTERVIEWER BEHAVIOR ENCOURAGE OR DISCOURAGE FULL REPORTING OF BEHAVIOR ARE ALSO PRESENTED. SEVERAL RESEARCH-PROVED PROCEDURES FOR INCREASING REPORTING AND MINIMIZING DISTORTION ARE DESCRIBED. FOUR INTERVIEWING TECHNIQUES--FACE-TOFACE, TELEPHONE, SELF-ADMINISTERED, AND RANDOMIZED RESPONSES--ARE COMPARED, AND THE TECHNIQUES WHICH PRODUCE THE MOST TRUTHFUL RESPONSES TO SLIGHTLY, MODERATELY, AND HIGHLY THREATENING QUESTIONS ARE INDICATED. OTHER TOPICS CONSIDERED ARE THE EFFECTS OF THE LENGTH, STRUCTURE, AND WORDING OF QUESTIONS; INTERVIEWERS' DELIBERATE OR INADVERTENT REWORDING OR ADDITIONS TO SURVEY QUESTIONS AND HOW THIS AFFECTS RESPONSES; AND THE DEGREE TO WHICH PRIOR EXPECTATIONS OF INTERVIEWERS DISTORT SURVEY DATA. ALSO CONSIDERED ARE THE EFFECTS OF RESPONDENT ANXIETIES, HOW RESPONSE RATE AND QUALITY ARE AFFECTED BY ASSURANCES OF CONFIDENTIALITY AND REQUIREMENTS OF INFORMED CONSENT, AND HOW MUCH THE PRESENCE OF OTHERS INHIBITS RESPONSES. THE MARLOW-CROWNE SCALE, WHICH HAS BEEN TOUTED AS A METHOD FOR IDENTIFYING PERSONS WHO DISTORT RESPONSES, IS SHOWN TO BE INEFFECTIVE FOR THAT PURPOSE. DATA IS PRESENTED WHICH WILL HELP RESEARCHERS IN QUANTIFYING SUCH RESPONSES AS 'PRETTY OFTEN' AND 'VERY OFTEN.' QUESTIONS USED IN THE EXPERIMENTS DESCRIBED ARE APPENDED, AND TABULAR DATA, SUBJECT INDEX, AND APPROXIMATELY 100 REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED)