A comparison of criteria for stepwise regression, 1977
Smith, Shelia Marie
1970-1979
This paper compares two criteria for stepwise regression, a statistical procedure for selecting variables in regression analysis. Chapter I contains concepts that will be needed to introduce stepwise regression. Chapter II discusses the old criterion for stepwise regression. This criterion enters or removes variables in the regression equation on the basis of a test of whether certain partial correlation coefficients are zero. Chapter III explains the new criterion. This procedure enters or removes variables on the basis of whether the Mean Square Error is significantly reduced as a result of adding a new variable or deleting an old one. The two methods are compared with examples involving actual data. Findings of both criteria and summary remarks are given in Chapter IV.
text
application/pdf
1977-08-01
thesis
Master of Science (MS)
Atlanta University
Mathematics
Jones, Arthur
Clark Atlanta University
Georgia--Atlanta
http://hdl.handle.net/20.500.12322/cau.td:1977_smith_shelia_m