Abstract
Multiple regression coefficients split by the levels of the dependent variable are examined. The decomposition of the coefficients can be defined by points on the ordinal scale or by levels in the numerical response using the Gifi system of binary variables. This approach permits consideration of specific values of the coefficients at each layer of the response variable. Numerical results illustrate how to identify levels of interpretable regression coefficients.
DOI
10.22237/jmasm/1351742580
Included in
Applied Statistics Commons, Social and Behavioral Sciences Commons, Statistical Theory Commons
