NCJ Number
58101
Journal
Journal of the American Statistical Association Volume: 73 Issue: 361 Dated: (MARCH 1978) Pages: 16-23
Date Published
1978
Length
8 pages
Annotation
AN INVESTIGATION OF VARIABLILITY OF THE NATIONAL CRIME SURVEY (NCS) STATISTICS ATTRIBUTABLE TO INTERVIEWERS IS UNDERTAKEN IN THIS STUDY OF EIGHT 'IMPACT AID CITIES' SELECTED BY THE U.S. DEPARTMENT OF JUSTICE.
Abstract
THIS COMPONENT OF RESPONSE VARIANCE REFLECTS THE PROPENSITY OF INTERVIEWERS, CODERS, AND EDITORS TO INTERPRET AND CARRY OUT SURVEY OPERATIONS IN DIFFERENT WAYS. THE RESULTS OF SIMILAR STUDIES HAVE LED TO CHANGES IN SURVEY OPERATIONS. THE SURVEY WAS CONDUCTED IN ATLANTA, BALTIMORE, CLEVELAND, DALLAS, DENVER, NEWARK, PORTLAND, AND ST. LOUIS. THIS INTERVIEWER VARIANCE STUDY (IVS) WAS BASED ON THE 1975 NCS SURVEY. IN ORDER TO ESTIMATE THE CORRELATED COMPONENT OF RESPONSE VARIANCE, THE METHOD OF INTERPENETRATED SUBSAMPLES DEVELOPED BY MAHALANOBIS (1946) WAS USED. THE MATHEMATICAL MODEL USED IN THIS STUDY WAS AN ADAPTATION OF A MODEL DEVELOPED BY HANSEN, HURWITZ, AND BERSHAD (1961). THE 1975 NCS STATISTICS FOR THE EIGHT CITIES ARE SUBJECT TO INTERVIEWER VARIABILITY; HOWEVER, THE EXTENT TO WHICH THIS WAS A FACTOR VARIED CONSIDERABLY AMONG THE CITIES AND ACCORDING TO THE NATURE OF THE STATISTICS. FURTHERMORE, USERS OF THIS SURVEY MAY DEVELOP MISCONCEPTIONS REGARDING THE QUALITY OF NCS DATA. FOR EXAMPLE, THE ESTIMATED TOTAL HOUSEHOLD VICTIMIZATION RATE FOR AN AVERAGE INTERVIEWER ASSIGNMENT AREA IN NEWARK WAS 0.26; FOR BALTIMORE IT WAS 0.52. CAUTION SHOULD BE EXERCISED REGARDING THE INTERPRETATION OF DIFFERENCES AMONG THE CITIES IN REPORTED VICTIMIZATION RATES. IN ADDITION TO THE NEED FOR FURTHER RESEARCH, IT WAS SUGGESTED THAT THE TRAINING AND OBSERVATION OF NCS INTERVIEWERS COULD POSSIBLY BE REVISED WITH A GREATER EMPHASIS PLACED ON CONCEPTS RELATED TO ITEMS FOR WHICH LARGE INTERVIEWER VARIABLILITY WAS REPORTED. REFERENCES AND STATISTICS ARE PROVIDED. (KCP)