DESIGN AND CONDUCT OF QUASI-EXPERIMENTS AND TRUE EXPERIMENTS IN FIELD SETTINGS (FROM HANDBOOK OF INDUSTRIAL AND ORGANIZATIONAL PSYCHOLOGY, 1976, BY MARVIN D DUNNETTE - SEE NCJ-52907)

• NCJ Number: 52908
• Author(s): T D COOK; D T CAMPBELL
• Date Published: 1976
• Length: 104 pages
• Grant Number(s): GSOC-7103704
• Sponsoring Agency: Russell Sage Foundation; National Science Foundation
• Corporate Author: Rand McNally and Co
• Publication Format: Document
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

THE FOUR TYPES OF EXPERIMENTAL VALIDITY ARE EXPLAINED, A NUMBER OF QUASI-EXPERIMENTAL DESIGNS ARE OUTLINED, AND EXAMPLES OF TRUE EXPERIMENTS WHICH HAVE BEEN SUCCESSFUL IN FIELD SETTINGS ARE GIVEN.

Abstract

AFTER AN INTRODUCTION EXPLAINING THE DIFFERENCE BETWEEN TRUE EXPERIMENTAL DESIGNS, IN WHICH SUBJECTS ARE RANDOMLY ASSIGNED TO TREATMENT GROUPS, AND QUASI-EXPERIMENTAL DESIGNS, WHICH COMPENSATE FOR NONRANDOM ASSIGNMENT TO GROUPS, THE NEED TO CONSIDER TESTING VALIDITY IS EMPHASIZED. FOUR TYPES OF VALIDITY ARE DESCRIBED: STATISTICAL CONCLUSION VALIDITY, INTERNAL VALIDITY, CONSTRUCT VALIDITY, AND EXTERNAL VALIDITY. QUASI-EXPERIMENTAL DESIGNS ARE MOST OFTEN USED IN INDUSTRIAL AND ORGANIZATIONAL PSYCHOLOGY STUDIES BECAUSE OF THE GENERAL UNAVAILABILITY OF RANDOMLY ASSIGNED TREATMENT GROUPS. A NUMBER OF NONEQUIVALENT CONTROL GROUP DESIGNS, COHORT DESIGNS, REGRESSION-DISCONTINUITY DESIGNS, TIME-SERIES DESIGNS, AND CORRELATIONAL DESIGNS WHICH HAVE BEEN USED OR WHICH COULD BE USED IN FIELD STUDIES ARE OUTLINED. THE RESEARCHER IS WARNED AGAINST USING DESIGNS WITH ONLY ONE GROUP OR WITH ONLY ONE TESTING. THEN SEVEN GENERALLY INTERPRETABLE DESIGNS ARE DESCRIBED AND ILLUSTRATED BY CHARTS. COHORT DESIGNS ARE ONE POSSIBLE NONEQUIVALENT GROUPING. GRAPHS ILLUSTRATE SEVERAL REGRESSION-DISCONTINUITY DESIGNS AND TIME-SERIES DESIGNS WHILE THE DISCUSSION POINTS OUT THEIR USEFULNESS TO ACTUAL ORGANIZATIONAL SETTINGS. CAUSAL MODELING, PATH ANALYSIS, STRUCTURAL EQUATION TECHNIQUES, AND THE USE OF CORRELATIONAL DESIGNS ARE EXPLORED. THE PROBLEMS OF TRUE EXPERIMENTAL DESIGN ARE POSED AGAINST A DESCRIPTION OF SEVERAL SUCCESSFUL STUDIES USING EXPERIMENTAL DESIGN, AND RESEARCHERS ARE URGED TO LOOK FOR OPPORTUNITIES TO CONDUCT SUCH STUDIES. GRAPHS, TABLES, AND A BIBLIOGRAPHY ARE INCLUDED. (GLR)