Testing the Validity of the Logistic Link Function in Lung Cancer Data

Presenter Information

Harrison Quick, Minnesota State University, Mankato

Location

CSU 284

Start Date

21-4-2008 8:00 AM

End Date

21-4-2008 10:00 AM

Student's Major

Mathematics and Statistics

Student's College

Science, Engineering and Technology

Mentor's Name

Mezbahur Rahman

Mentor's Department

Mathematics and Statistics

Mentor's College

Science, Engineering and Technology

Description

In statistics, logistic regression models are often used when the variable of interest is a dichotomous (success/failure) or categorical value. For example, one could use a logistic regression model to determine whether a person is at risk for lung cancer using predictor variables such as age, smoking status, and family history of lung cancer. Unfortunately, there is no guarantee that the logistic link function chosen is correct for the given set of data. Because of this, my project was to verify the validity of the logistic link function in Dr. Mezbahur Rahman and Dr. Larry Pearson's lung cancer study using the Kolmogorov-Smimov test statistic introduced by Qin and Zhang (1997). Also, I compared Qin and Zhang's results from bootstrap estimation to results found using a permutation test.