ERIC Number: ED111828
Record Type: Non-Journal
Publication Date: 1974-Aug
Reference Count: N/A
Reverse Commonality and Some Illustrative Applications.
Mayeske, George W.; Beaton, Albert E., Jr.
The results of an algorithm which is designed to take a set of commonality coefficients, either real or manipulated, and, if possible, produce one or more sets of regressor correlations that are consistent with them are examined. A number of different ways of resolving the higher order commonality values into their lower orders were tried and the number and nature of solutions generated from them were examined for their meaningfulness and variability. In general it was found that this could be a meaningful exercise because it allows an analyst to test his assumptions about the nature of the confounding to see if a solution can be obtained. If one cannot be obtained then his assumptions must be revised. But if one can be obtained, then he can examine them to see which kinds of variables have the greatest sensitivity to the assumptions. The resulting output also allows one to gauge the variability of regression coefficients that will satisfy the same set of commonality values and the effectiveness of the regression system. This technique called "Reverse Commonality" is best suited for an interactive computing arrangement so that an analyst can use it rapidly in a sequential manner. Since, for a large number of variables the sheer volume of commonality values become unmanageable and, since a number of different ways are available for resolving the confounding, an algorithm is needed which uniquely resolves the higher order values according to one's assumptions. (Author/DEP)
Publication Type: Reports - Research
Education Level: N/A
Authoring Institution: Office of Education (DHEW), Washington, DC. Office of Planning, Budgeting, and Evaluation.