ERIC Number: ED216957
Record Type: RIE
Publication Date: 1982-Mar
Reference Count: 0
Historical Policy Capturing Through Use of Computerized Databases and Trend Analysis Techniques.
Bobner, Ronald F.; And Others
This paper suggests one way that trends in the literature might be captured by historians through the use of computerized information retrieval services and particular statistical techniques for analyzing trends. For illustrative purposes, the topic "desegregation" was investigated in Lockheed's DIALOG (1982) information retrieval system utilizing four databases: "ERIC,""Dissertation Abstracts,""Sociological Abstracts," and "Psychological Abstracts." The time period searched was from 1966 to 1980 in each instance. The authors suggest several questions that might be asked about the literature retrieved and present appropriate means of statistically testing these questions. The statistical tests are based on a methodology known as multiple linear regression, a technique which is very appropriate for detecting trends. For example, the first question that might be asked is, "does the literature vary significantly from year to year?" The statistician, in this case, would test the slope of the line on the graph to see whether or not it differs significantly from zero. Questions might also be asked about trends in subtopics. For example, a historical perspective of desegregation, which is a broad and general topic, can be obtained by searching subtopics such as bussing. Looking at more than one subtopic and comparing the slopes of the lines provides a new perspective. Comparisons of literature in different databases can also be made. Using statistical strategies similar to the ones already described, one could compare the output from "ERIC" to that of "Sociological Abstracts." Appendices contain a list of the databases accessible through Lockheed's DIALOG service and in-depth explanations of the statistical tests. (RM)
Publication Type: Reports - Descriptive; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Identifiers: Multiple Linear Regression
Note: Paper presented at the Annual Meeting of the American Educational Research Association (New York, NY, March, 1982). Research was partially supported by Bierce Library, University of Akron.