Answer the following questions based on Section 5.2.1: Regression pitfalls - nonconstant variance.

Consider the following data:

X 0.8 0.6 0.8 0.2 0.5 1.0 0.3 0.1 0.2 0.5
Y 3.3 2.3 2.2 1.3 1.7 2.3 1.5 1.0 1.0 1.5
Use statistical software to fit a simple linear regression model with response variable, Y, and predictor variable, X (call this model 1). Then fit another simple linear regression model with response variable, loge(Y), and predictor variable, X (call this model 2). Use the fitted models to answer the following questions.
1. 

Construct a scatterplot with the model 1 residuals on the vertical axis and the model 1 fitted values on the horizontal axis. What is the best description of this plot?

2. 

If we apply a Breusch-Pagan/Cook-Weisberg test using the fitted values from model 1, we obtain a p-value of 0.031. True or false? Based on a significance level of 0.05, this suggests the constant variance assumption is violated.

3. 

Construct a scatterplot with the model 1 residuals on the vertical axis and the model 1 fitted values on the horizontal axis. What is the best description of this plot?

4. 

If we apply a Breusch-Pagan/Cook-Weisberg test using the fitted values from model 2, we obtain a p-value of 0.133. True or false? Based on a significance level of 0.05, this suggests the constant variance assumption is violated.

5. 

Which of the following tend to result from nonconstant variance of linear regression errors? (Select all that apply.)

Categories:

Tags:

Comments are closed