ERRORS IN THE MEASUREMENT OF CRIME - AN APPLICATION OF JORESKOG'S METHOD FOR THE ANALYSIS OF GENERAL COVARIANCE STRUCTURES (FROM QUANTITATIVE STUDIES IN CRIMINOLOGY, 1978, BY CHARLES WELLFORD - SEE NCJ-52553

• NCJ Number: 52554
• Author(s): G BRIDGES
• Date Published: 1978
• Length: 21 pages
• Corporate Author: Sage Publications, Inc
• Publication Format: Document
• Publication Type: Statistics
• Language: English
• Country: United States of America
• Note: REVISION OF A PAPER PRESENTED AT THE ANNUAL MEETING OF THE AMERICAN SOCIETY OF CRIMINOLOGY IN ATLANTA, GEORGIA, 1977

Annotation

THE EXTENT AND EFFECTS OF ERROR IN VARIOUS CRIME MEASURES SPECIFICALLY IN SELF-REPORTED AND OFFICIAL CRIMINAL RECORDS, ARE EXAMINED.

Abstract

MEASURES OF HIDDEN AND OFFICIAL CRIMES ARE SUBJECT TO SYSTEMATIC ERRORS RESULTING IN DISTORTED INTERPRETATIONS OF THE RELATIONSHIP BETWEEN CRIME MEASURES IN THE GENERAL POPULATION AND EXPLANATORY VARIABLES SUCH AS SOCIOECONOMIC STATUS (SES). THE EFFECTS OF THESE ERRORS MAY BE ESTIMATED IN MULTIVARIATE MEASUREMENT MODELS. K.G. JORESKOG DEVELOPED AND DEMONSTRATED AN APPROACH TO ESTIMATING MEASUREMENT ERROR IN SYSTEMS OF VARIABLES THAT ALLEVIATES PROBLEMS FOUND IN PATH ANALYSIS (I.E., A METHOD TO DEDUCE THE PRESENCE OF SYSTEMATIC ERRORS FROM PATTERNS AMONG PATH COEFFICIENTS). JORESKOG'S METHOD, WHICH GENERALIZES THE PRINCIPLE OF MAXIMUM-LIKELIHOOD FACTOR ANALYSIS TO THE TREATMENT OF GENERAL COVARIANCE STRUCTURE, YIELDS LARGE-SAMPLE ESTIMATES OF MODEL PARAMETERS THAT ARE USUALLY NORMAL, EFFICIENT, AND CONSISTENT. A DESIGN FOR THE ANALYSIS OF DETERMINISTIC MODELS, DESCRIBED IN THE STANDARD FORM OF A SYSTEM OF LINEAR EQUATIONS, DIFFERENTIATES BETWEEN STRUCTURAL RELATIONSHIPS OF UNOBSERVED INDEPENDENT AND DEPENDENT VARIABLES, AND MEASUREMENT RELATIONSHIPS LINKING OBSERVED VARIABLES OR INDICATORS TO THESE UNOBSERVED VARIABLES. THE COMPONENTS OF SES ARE THEN INCORPORATED INTO THE MODEL TO EXAMINE THE RELATIONSHIP BETWEEN SES AND 'TRUE' HIDDEN AND OFFICIAL CRIMINALITY. THROUGHOUT THE ANALYSIS, LINKAGES AMONG THE INDICATORS OF HIDDEN AND OFFICIAL CRIMES ARE EXAMINED TO ASSESS THE TYPES AND MAGNITUDE OF MEASUREMENT ERRORS WHICH APPEAR IN THE SOLUTION. PERSONAL INTERVIEWS AND POLICE RECORDS PROVIDED DATA FOR A SYSTEMATIC SURVEY (CONDUCTED IN PHILADELPHIA, PA., IN 1971) OF 567 MALES BORN IN 1945. THE FOLLOWING SAMPLE MEASURES OF HIDDEN AND OFFICIAL CRIMINALITY EXPLORE THE STRUCTURAL AND MEASUREMENT RELATIONSHIP AMONG SELF-REPORTED AND OFFICIAL INDICES OF CRIME: (1) THE SUM OF ALL HIDDEN OFFENSES; (2) THE SUM OF HIDDEN INDEX OFFENSES; (3) THE SUM OF ALL HIDDEN OFFENSES COMMITTED BEFORE AGE 18; (4) THE RATIO OF HIDDEN INDEX OFFENSES COMMITTED BEFORE AGE 18 AND TOTAL HIDDEN OFFENSES; (5) THE SUM OF ALL POLICE CONTRACTS; (6) THE AGE AT FIRST POLICE CONTACT AND (7) THE LENGTH OF EXPOSURE TO POLICE CONTACT. IMPLICATIONS ARE THAT UNOBSERVED LEVELS OF HIDDEN OFFENSES MAY DISPROPORTIONATELY CONSIST OF NONINDEX OFFENSES BECAUSE OF THE STRONG INVERSE RELATIONSHIP WITH THE RATIO OF HIDDEN INDEX TO ALL HIDDEN OFFENSES, AND RELIANCE ON ANALYTICAL APPROACHES THAT OVERLOOK ERRORS IN CRIME AND DELINQUENCY STUDIES MAY RESULT IN PROBLEMS OF INFERENCE. TABULAR DATA, FIGURES, DETAILS OF THE ANALYSIS, NOTES, AND REFERENCES ARE PROVIDED. (JCP)