BEYOND A STRAIGHT LINE FIT - PROBATION PROJECTION TECHNIQUES WHICH USE READILY AVAILABLE DATA (FROM AMERICAN CORRECTIONAL ASSOCIATION - PROCEEDINGS OF 106TH ANNUAL CONGRESS OF CORRECTIONS, 1976 - SEE NCJ-49145)

• NCJ Number: 49167
• Author(s): T G CRAGO; C S HROMAS
• Date Published: 1976
• Length: 11 pages
• Corporate Author: American Correctional Association
• Publication Format: Document
• Publication Type: Statistics
• Language: English
• Country: United States of America
• Note: *This document is currently unavailable from NCJRS.

Annotation

COLORADO'S USE OF MULTIPLE LINER REGRESSION ANALYSIS OF EXISTING DATA TO PROJECT PRISON POPULATION AND RATES OF COMMITMENT IS DESCRIBED.

Abstract

PROJECTION RESEARCH IN COLORADO HAS DEPENDED ON MANUAL INFORMATION SYSTEMS. THE APPROACH TO THE PROJECTION PROBLEM WAS TO PREDICT QUARTERLY COMMITMENTS ON THE BASIS OF UNEMPLOYMENT DATA, STATE POPULATION DATA, AND SEASONAL VARIATION KNOWN OR ASSUMED TO BE CORRELATED WITH THE RATE OF COMMITMENT. TO PREDICT THE TOTAL INCARCERATED POPULATION, CHANGES IN THE STATE'S PAROLE REVOCATION RATE AND IN COLORADO'S SENTENCING LAW WERE ALSO TAKEN INTO CONSIDERATION. MULTIPLE LINEAR REGRESSION, A STANDARD MULTIVARIATE STATISTICAL TECHNIQUE, WAS USED TO PROJECT COMMITMENTS AND PRISON POPULATION. THE METHOD EXPLAINED STATISTICALLY 84 PERCENT OF THE TOTAL VARIANCE IN THE NUMBER OF COMMITMENTS OVER A 5-YEAR PERIOD, THE CORRELATION BETWEEN OBSERVED AND PREDICTED VALUES BEING APPROXIMATELY 0.92. THE METHOD EXPLAINED 75 PERCENT OF THE VARIATION IN TOTAL INCARCERATED POPULATION OVER A 5-YEAR PERIOD, WITH CORRELATION BETWEEN OBSERVED AND PREDICTED VALUES IN EXCESS OF 0.86. A LIST OF REFERENCES AND SUPPORTING GRAPHS AND DATA ARE PROVIDED. (LKM)