Project 4: Multiple Linear Regression
 Due May 4, 2020 by 12:59am
 Points 25
 Submitting a file upload
 Available until May 12, 2020 at 12:59am
This assignment will allow you to work through the last part of the data analysis of Project 4 and get feedback that you can use to improve this section for the final project paper.
How to get started:
Multiple Regression: Start with a model that includes all seven of your predictor variables and then eliminate them one at a time to find the best overall model. Make sure to include the following information for the Multiple Regression. You can just type the answers OR you can write it out in full sentences. For your final paper you will need to write this out in full sentence/paragraph format.
You should already have the Project 4 data saved to your computer, but if you need it, here it is  Project 4: Data
Questions to answer/things to do:
 Write the claim you are testing and identify if it is the null or alternative. [Hint, for this project we DO think our response variable is affected by our predictor variables so it should be the alternative.]
 Analyze the data using Minitab and calculate the appropriate statistics:
 Based on your results explain why you are going to remove a variable from your model. Make sure to include the specific reason that you are making this decision. Example: For Model 1 teacher salary had the highest pvalue so it was removed from the Model. For Model 2 chronic truancy had the highest pvalue so it was removed from the Model... etc...
 Include appropriate Minitab output in paper or appendix for ALL SEVEN MODELS. Needed output from Minitab: Analysis of Variance, Model Summary, & Coefficients.
 Using the Stepwise Process repeat the steps above (removing one variable each time) until you have worked down to ONE predictor variable left in the model. Create a table with this information (see below) and also make sure to explain why you removed each predictor variable from the model. [Remember to include the Minitab output for each model in the paper or the Appendix!]
Model (Response = ?) Predictor variables & pvalue Rsq (adj) or Rsq Model 1 seven predictors & pvalue Rsq (adj) = Model 2 six predictors & pvalue Rsq (adj) = Model 3 five predictors & pvalue Rsq (adj) = Model 4 four predictors & pvalue Rsq (adj) = Model 5 three predictors & pvalue Rsq (adj) = Model 6 two predictors & pvalue Rsq (adj) = Model 7 one predictor & pvalue Rsq =  What Model provides the best fit and why is this model the best? [Hint, it won't necessarily be Model 7. There are two different things you need to look at when determining the best model. If you're not sure what these are, see the notes and video for "Multiple Regression Model of Best Fit."]
 Answer the following questions only for the Model of Best Fit. Make sure to include the Minitab output!
 There [IS, IS NOT] enough evidence to [REJECT, SUPPORT] the claim that [Insert claim for Model of Best Fit].
 A [very good, good, fair, poor] amount,[Rsquared adjusted %], of the variability in the [response variable] is explained by the [predictor variables].
 What relationship does each predictor (x) variable have with the response (y) variable when all other predictor variables are held constant?
 What is the realworld meaning of your results?
Rubric
Criteria  Ratings  Pts  

Claim? Is it null or alternative?
threshold:
pts


pts



Explanation for why variables were removed to get to the best fit model.
threshold:
pts


pts



Table with predictor variables and rsq/rsq adjusted values
threshold:
pts


pts



Final Model Information & Explanation
threshold:
pts


pts



Minitab Output for Models
threshold:
pts


pts



Minitab file was included with assignment
threshold:
pts


pts



Total Points:
25
out of 25
