next up previous
Next: Plots Up: Regression Previous: Least Squares (LS) Estimation

Wilcoxon Estimation

Wilcoxon regression may be performed by unchecking the "LS Regression" button. An intercept is included by default. Using the Height-Weight data and performing a Wilcoxon regression we get
Rweb:> b  
                 Coef  Std. Err  t-ratio 
intercept  -338.33300 174.42100 -1.93975 
Height        7.22222   2.32444  3.10708 
Rweb:> diagnosticvar <- c(" R2 (ROBUST COEF.) ","THE FINAL FULL MODEL DISPERSION IS:
                          ","THE ESTIMATE OF SCALE IS:")  
Rweb:> diagnostic <- wilother(diagnosticvar)  
Rweb:> diagnostic  
                                         Value 
 R2 (ROBUST COEF.)                    0.494471 
THE FINAL FULL MODEL DISPERSION IS: 134.983000 
THE ESTIMATE OF SCALE IS:            19.723500
Hence the estimated Wilcoxon regression line is Weight = -338.3 + 7.2*Height and it accounts for 49.4% of the variation in weight.


2000-08-21