East Carolina University
Department of Psychology


    Not satisfied with the fit between your data and your polynomial model?  Try a bumblebee regression instead.  Just put some pollen on each of your data points and let the bumblebee loose.  She will fit your data perfectly.  Alternatively, use Proc Bumblebee in SAS.


Now let us cross-validate our model by applying it to a second random sample from the same population.

 


Hmm, why does the model not generalize well?  Maybe I need to refit the data from a third sample from the same population.

 

Return to Wuensch's Stats Lessons Page