Scribd
Upload
29 views

Lab #6 Assignment (Spring 2022)

The student is asked to analyze non-linear regressions between starting salary and mid-career salary data from 153 state schools to see which model best fits the data. They must run quadratic, log-linear, linear-log, and log-log regressions, interpret the coefficients, determine the best model using adjusted R-squared, and use the best model to predict mid-career salary for a starting salary of $50,000. They must report the regression equations, coefficient interpretations, best model selection, and predicted salary in an analytics report with appendix screenshots of the regression outputs.

Uploaded by

Kyle Anders
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
29 views

Lab #6 Assignment (Spring 2022)

The student is asked to analyze non-linear regressions between starting salary and mid-career salary data from 153 state schools to see which model best fits the data. They must run quadratic, log-linear, linear-log, and log-log regressions, interpret the coefficients, determine the best model using adjusted R-squared, and use the best model to predict mid-career salary for a starting salary of $50,000. They must report the regression equations, coefficient interpretations, best model selection, and predicted salary in an analytics report with appendix screenshots of the regression outputs.

Uploaded by

Kyle Anders
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Lab #6 Assignment: Non-linear Regression

Due Tuesday, April 12th by 11:59pm

You are an economic research analyst trying to understand how much impact a starting salary will have
on mid-career salary. You find data gathered by the Wall Street Journal on 153 state schools that
measures both the median starting salary of those schools’ graduates and the median mid-career salary
of those graduates. Assume that you have already run a linear regression and now want to analyze some
curved regressions because you feel that their coefficient interpretations may be more applicable to this
data. You will run different types of non-linear regressions to see which model is the best fit and see
what it says about the effect of starting salary on mid-career salary.

Include the following types of regression:

• Quadratic
• Lin-Log (Logarithmic)
• Log-Lin (Exponential)
• Log-Log

Report Deliverables

 Use the Analytics Report Outline from D2L


 Introduction (follow the directions in the outline)
 Data Analysis Section:
o For each model:
▪ The regression equation
▪ Interpretation of the coefficient
• For the quadratic: Interpret the marginal effect for someone with a
starting salary of $50,000 and say the shape (concave up or concave
down), including how you know.
• For the log-transformed models: This will be the normal coefficient
interpretations (see the slides for guidance).
o Determining the best model:
▪ A discussion of why you calculated a new Adjusted R2
▪ R2 and Standard Error interpretation for the best model
• Hint: Make sure to say the correct response variable for whichever
number you interpret.
▪ Use this model to predict the mid-career salary of someone with a starting
salary of $50,000 (show the equation you used).
 Conclusion: Close the document with a summary.
o Restate which model was the best and how do you know?
o Since we had 4 different coefficient interpretations (one for each model), restate which
one are we going to use?
 Appendix:
o Screenshots of the 4 different regression outputs (name in screenshot)
o Screenshot of Adjusted R2 comparison (name in screenshot)

Variable Descriptions

Starting Salary: Measured in dollars


Mid-Career Salary: Measured in dollars

Hint: Be careful with the units for when the variable has been logged.

You might also like