Bonus Question for Homework 1

Earn up to 20 points back on your Homework 1 score! This is not mandatory. I am therefore giving this assignment 0 points in Canvas, but I will manually adjust your homework 1 score if you turn in and receive points for this bonus question. Due November 1.

I made a mistake in my solution to problem 5 in Homework 1. See the crossed out lines in problem 5 for my original explanation: hw1soln.pdf Download hw1soln.pdf

Since I said in class that Multiple R^2 describes the fraction of variance in Y that is explained by the model, it might seem strange or even contradictory to read a negative value for Multiple R^2.  It turns out that there is nothing wrong with a negative R^2 value for certain classes of models. Your job is to provide an explanation for how negative multiple R^2 values occur and when they can and cannot occur.

A good answer will provide an easy to follow explanation, give a good example model which can be simulated in R and shown to have very negative Multiple R^2, illustrate results with a plot, and describe the classes of models that can and cannot have a negative R^2. Writing quality will be a factor in deciding the bonus points. Follow the general rubric provided in the file rubrichw1.pdf Download rubrichw1.pdf

EDIT -- Be sure that the writeup is processed and displayed by Canvas.