AGGREGATE DATA ANALYSIS - AN OVERVIEW

• NCJ Number: 57689
• Journal: Sociological Methods and Research Volume: 7 Issue: 4 Dated: SPECIAL ISSUE (MAY 1979) Pages: 379-383
• Author(s): E F BORGATTA; D J JACKSON
• Date Published: 1979
• Length: 5 pages
• Corporate Author: Sage Publications, Inc
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

THIS ARTICLE PRESENTS AN OVERVIEW OF THE PROBLEMS SOCIAL SCIENTISTS ENCOUNTER IN ANALYZING AND INTERPRETING AGGREGATE DATA, AND INTRODUCES SOME SOLUTIONS.

Abstract

BECAUSE RESEARCHERS IN THE SOCIAL AND PSYCHOLOGICAL SCIENCES HAVE NOT ALWAYS HAD AMPLE MONEY OR RESOURCES FOR DATA COLLECTION, THEY HAVE HAD TO RELY ON ALTERNATIVES TO DIRECT STUDY. AVAILABILITY OF AGGREGATE DATA BASED ON CENSUSES AND REGISTRATIONS THUS ATTRACTED MANY USERS OVER THE HISTORY OF THE SOCIAL SCIENCES. THE QUESTIONS SURROUNDING THE USE OF THIS DATA, PLAGUED AS IT IS WITH THE METHODOLOGICAL ERRORS OF THE PAST, ARE NOW BECOMING SYSTEMATIZED. ONE SUCH ERROR HAS BEEN 'ECOLOGICAL FALLACY', WHEREIN THE INTERPRETATION OF INDIVIDUAL-LEVEL VARIABLES BASED ON ANALYSES OF DATA AGGREGATED BY GEOGRAPHICAL OR OTHER UNITS IS WARNED AGAINST. THE ASSUMPTION RESULTING FROM THIS CAUTIONARY ATTITUDE IS THAT EVERY INTERPRETATION AT THE INDIVIDUAL LEVEL OF AGGREGATE DATA IS TABOO. A MORE TEMPERED CONSIDERATION OF THE PROBLEM REVEALS THAT (1) WHILE ALWAYS SUSPECT, AGGREGATE DATA COULD SUGGEST FINDINGS THAT EXIST AT THE INDIVIDUAL LEVEL; (2) THE ANALYSIS OF AGGREGATE DATA COULD BE OF INTEREST IN ITSELF; (3) COMPARISON OF DIFFERENT LEVELS OF AGGREGATION AND INDIVIDUAL-LEVEL DATA COULD PROVIDE INTERESTING FINDINGS; AND (4) A PARTICULAR BRAND OF REDUCTIONISM CAN BE USED TO ATTRIBUTE SOME CHARACTERISTICS ASSOCIATED WITH GEOGRAPHICAL AND OTHER AGGREGATION UNITS TO INDIVIDUALS. OTHER ASPECTS OF ANALYZING AGGREGATE DATA INCLUDE CONVERTING AGGREGATE-LEVEL DATA INTO RATES, DEALING WITH THE 'REALITY OF GROUPS' FROM SPECULATIVE TERMS, DEALING WITH ISSUES OF HOW CHARACTERISTICS OF GROUPS ARE TO BE MEASURED INDEPENDENTLY OF THE CHARACTERISTICS OF THE MEMBERS, DEALING WITH ASSUMPTIONS OF LINEARITY IN THE MEASURES USED, AND DEALING WITH THE IMPLICIT ASSUMPTIONS ABOUT THE NATURE OF THE RELATIONSHIPS OBSERVED. TWO REFERENCES ARE PROVIDED. (MHP)