STA429/1007 Assignment 8

Quiz on Thursday Nov. 18th

This assignment is based upon lecture material on logistic regression with a binary dependent variable. Please bring log and list file to the quiz.

The file birdlung.dat has these variables:

1. Lung Cancer	1=Yes, 0=No
2. Sex	1=Female, 0=Male
3. Socioeconomic Status	1=High, 0=Low
4. Birdkeeping	1=Yes, 0=No
5. Age
6. Years smoked
7. Cigarettes per day

Do a logistic regression, with lung cancer as the dependent variable. For this assignment, do not include any interactions among variables, even though it's something you would investigate in a more thorough data analysis. For this assignment, there is one primary issue.

1. Controlling for all other variables, is birdkeeping significantly related to the chance of getting lung cancer? Perform a likelihood ratio test to answer the question.
   1. What is the -2 Log Likelihood for the reduced model?
   2. What is the -2 Log Likelihood for the full model?
   3. What is the value of the test statistic G? The answer is a number.
   4. What are the degrees of freedom for the test? The answer is a number.
   5. What is the p-value? The answer is a number.
   6. Just for comparison, find the corresponding Wald chisquare statistic, the df and the p-value.
   7. What do you conclude? Presence of a relationship is not enough. Say what happened.
   8. For a bird-keeping woman of low socioeconomic status who is average on all the other independent variables, what is the estimated probability of lung cancer? This answer should be based on the full model.
   9. For a non-bird-keeping woman of low socioeconomic status who is average on all the other independent variables, what is the estimated probability of lung cancer? This answer should be based on the full model.
2. Also, Be able to interpret the Wald tests for each variable controlling for all the others. This is part of the standard output.