NCJ Number
60339
Journal
EVALUATION QUARTERLY Volume: 3 Issue: 3 Dated: (AUGUST 1979) Pages: 385-410
Date Published
1979
Length
26 pages
Annotation
AN INTRODUCTION TO THE USE OF ECONOMETRIC ESTIMATION PROCEDURES FOR ANALYZING DATA WITH BOTH CROSS-SECTIONAL AND LONGITUDINAL VARIATION IS PRESENTED AND COMPARED TO THE BOX-TIAO PROCEDURES USED BY DEUTSCH AND AULT.
Abstract
A MULTIPLE TIME SERIES DESIGN ALLOWS RESEARCHERS TO STUDY VARIATIONS OVER TIME AND BETWEEN DIFFERENT GROUPS. MOREOVER, THE MULTIPLE TIME SERIES DESIGN DOES NOT LIMIT RESEARCH EVALUATION TO THE PRESENCE OR ABSENCE OF A SINGLE TREATMENT METHOD AS IT CAN BE EXTENDED EASILY TO MAKE COMPARISONS BETWEEN DIFFERENT TREATMENT APPROACHES. DESPITE THE BENEFITS OF SUCH AN EVALUATION PROCEDURE, THIS METHOD HAS RARELY BEEN USED BECAUSE OF SUCH PROBLEMS AS THE FACT THAT APPROPRIATE DATA MAY NOT ALWAYS BE AVAILABLE, SOME RESEARCHERS MAY NOT BE AWARE THAT EXISTING DATA CAN BE USED FOR A MULTIPLE TIME SERIES ANALYSIS, AND OTHERS ARE UNAWARE OF THE BENEFITS POSSIBLE IN SUCH AN APPROACH. THE BENEFITS OF THIS ECONOMETTRIC APPROACH ARE DEMONSTRATED BY ANALYZING DATA COMPILED BY DEUTSCH AND ALT (1975) ON THE EFFECTIVENESS OF THE MASSACHUSETTS GUN CONTROL LAW. THE MOST REASONABLE GENERALIZED LEAST SQUARES (GLS) APPROACH USED IN THIS STUDY YIELDED RESULTS IDENTICAL TO THOSE PRODUCED BY DEUTSCH AND ALT WHO USED BOX-TIAO PROCEDURES. THE BOX-TIAO PROCEDURES SHOULD BE USED WHEN MORE THAN 50 LONGITUDINAL OBSERVATIONS ARE BEING STUDIED, AND THE GLS IS MOST APPROPRIATE WHEN EACH TIME SERIES IS SHORT. STATISTICAL TABLES, ENDNOTES, AND REFERENCES ARE INCLUDED. (KCP)