NCJ Number
57690
Journal
Sociological Methods and Research Volume: 7 Issue: 4 Dated: SPECIAL ISSUE (MAY 1979) Pages: 384-387
Date Published
1979
Length
12 pages
Annotation
THIS ARTICLE EXAMINES THE RELATIONSHIP BETWEEN THE TWO REGRESSION METHODS MOST OFTEN RECOMMENDED FOR ASSESSING GROUP EFFECTS IN NONEXPERIMENTAL DATA: COVARIANCE ANALYSIS AND CONTEXTUAL ANALYSIS.
Abstract
THE ESTIMATION OF GROUP EFFECTS IS APPROPRIATE TO SOCIAL SCIENCE BOTH BECAUSE THE GROUP EFFECTS ARE IMPORTANT IN THEIR OWN RIGHT--THE SOCIAL CONTEXT NO DOUBT AFFECTS HUMAN BEHAVIOR--AND BECAUSE SOCIAL SCIENTISTS MUST SOMETIMES RELY ON AGGREGATE DATA IN STUDYING INDIVIDUALS; THE ADVISABILITY OF SUCH A PRACTICE HINGES ON THE PRESENCE OF GROUP EFFECTS. COVARIANCE ANALYSIS AND CONTEXTUAL ANALYSIS BOTH ASSUME THAT INDIVIDUAL-LEVEL AND AGGREGATE-LEVEL DATA ARE AVAILABLE: THE ABSENCE OF INDIVIDUAL-LEVEL DATA OFTEN PRECLUDES THE SEPARATION OF INDIVIDUAL AND GROUP EFFECTS. THE EXAMINATION OF THE TWO METHODS OF ANALYSIS SHOWS THAT CONTEXTUAL EFFECTS ARE A SPECIAL CASE OF THE GROUP EFFECT OBTAINED IN COVARIANCE ANALYSIS. THIS FINDING IMPLIES THAT (1) THE GROUP EFFECT OF COVARIANCE ANALYSIS PROVIDES AN UPPER LIMIT FOR CONTEXTUAL EFFECTS; (2) COVARIANCE ANALYSIS IS MORE DIRECTLY APPLICABLE IN EXPLORATORY WORK, WHILE CONTEXTUAL ANALYSIS IS MORE DIRECTLY APPLICABLE IN CAUSAL ANALYSIS; AND (3) BOTH CONTEXTUAL AND COVARIANCE ANALYSIS ARE REQUIRED FOR A COMPLETE ACCOUNTING OF GROUP EFFECTS. IT IS CONCLUDED THAT THE TWO METHODS ARE BEST SUITED TO ANSWER DIFFERENT QUESTIONS. COVARIANCE ANALYSIS ASKS: HOW LARGE IS THE TOTAL IMPACT OF THE GROUPS? CONTEXTUAL ANALYSIS ASKS: WHAT CHARACTERISTICS OF THE GROUPS HAVE AN EFFECT? OFTEN, SOCIAL SCIENTISTS WHO STUDY GROUP EFFECTS ARE INTERESTED IN BOTH QUESTIONS; OFTEN, THEN, BOTH METHODS COULD BE USEFULLY EMPLOYED. FOOTNOTES AND REFERENCES ARE PROVIDED. (MHP)