We illustrate how to use JMP with the rat tumor data found in
Statistical Methods for Survival Data Analysis (1980) Elisa T. Lee. Lifetime Learning Publications. Belmont, California.
This data gives the tumor-free time for rats with three diets after tumor cells were injected into their foot pads.
Import into JMP the rat tumor data:
(Instructions for importing a text file into JMP)
The JMP data table should look as follows:
To find out about these tests we go to the Tools menu and select the ? tool. Then click on top of the table. We get the following text:
We see that both test detect a difference among the three populations.
Now we exclude the low-fat diet and perform the test again:
We see that there is a difference between the two population according to the Log-Rank test which gives more weight to the larger survival times than the Wilcoxon test which picks not difference between the two populations.
To see the behavior of the hazard rates we do an Exponential Plot:
We see that the hazard rate increases much faster for the unsaturated diet. In fact, the exponential distribution seems to fit the low-fat and saturated diet (they have constant hazard rates) but not the unsaturated diet. As seen below the lognormal distribution seems to fit the unsaturated diet best:
In conclusion, the data seems to indicate the diet does have an effect on tumor-free time.