P5510 Homework Survival Analysis
1. Perform survival analysis using the three procedures (Life Tables, Kaplan Meier, and
Cox Regression) for the data of the Breast Cancer survival file in the SPSS folder. It is
also on my 5510 web page way down the list of files under the name “Breast Cancer
Survival”. Compare survival for the three PATHSCAT groups. In addition, in a 2nd Cox
Regression analysis include both AGE and HISTGRAD as covariates. Finally, perform a
final analysis in which the interaction of AGE and HISTGRAD is assessed.
Note that both PATHSCAT and HISTGRAD are categorical factors.
I believe PATHSCAT is for PATHological Size Category.
a. Submit the enough output from each analysis to show that you’ve done that analysis,
including the “Variables in the Equation” table if it’s available. Submit in the order: Life
Tables, KM, Cox for the first three analyses.
b. Submit a 1-3 paragraph summary of the results of each analysis.
c) Submit a summary of the differences between the multivariate Cox analysis and the
univariate Cox analyses.
2. Use the Turnover Data file on my 5510 web page for the following.
There are two variables in the file that related to previous work history. The first is called
NUMEMPNM. This is the number of difference employers in the past 5 years. The
other is LG10DOS1. It is the logarithm of the number of days on the most recent job.
Create an index combining these two variables called GOODHIST. It should be the
difference of the Z-scores of these two variables.
That is GOODHIST = (ZNUMEMPNM - ZLG10DOS1)/2.
a. Submit a histogram of GOODHIST to show that you’ve computed it correctly.
b. Determine whether GOODHIST contributes uniquely to prediction of turnover over
and above FR2 – whether applicant has friends at the company – and UNEMPL –
unemployment rate in the year of hire. Remember that the word, uniquely, has a specific
meaning in statistics. Refer to 513 notes on multiple regression analysis if you’ve
forgotten. (Note that the term, “uniquely”, does not refer to the interaction of two
variables.)
Submit enough SPSS, including the “Variables in the Equation” tables to show that
you’ve done the analyses correctly.
c. Submit a paragraph or two describing the results of the analyses.