ERIC Number: ED031096
Record Type: RIE
Publication Date: 1969-Feb-7
Reference Count: 0
A Beta Index to Confirm Causal Directions in a Closed System of Five Variables.
Nigro, George A.
The beta coefficient of an intermediate variable in a causal direction remains relatively constant as other system variables are introduced and controlled in stepped regression, whereas that in the acausal direction changes noticeably. Normalized random numbers (200x5) were generated and substituted in interdependent equations to produce five scores for each of 200 pseudo-people. Stepped-regression analysis was then applied on all possible three-variable paths. Beta differentials on a given intermediate variable were computed and compared for opposite directions. The resulting index persisted in yielding values between 0 and +1 in the one direction--the closer to 0, the stronger the direction--as other system variables were stepped in. But because some remaining three-variable paths imitated the known causal paths, further study was conducted to find support for the known causal paths, and to eliminate the imitations. Conclusions were that the beta index, found wanting initially, had greater utility as a confirming device, by the addition of the two modifications, in that all but the originally designed causal paths were eliminated as lacking causal-direction consistency. The index procedure may well be confined to the computer presently. (Author/GO)
Publication Type: N/A
Education Level: N/A
Sponsor: Office of Education (DHEW), Washington, DC.
Authoring Institution: N/A
Note: Paper presented at American Educational Research Association, Los Angeles, Calif., Feb. 7, 1969