A total of N64 patients with k5 types of advanced cancer wer

A total of N=64 patients with k=5 types of advanced cancer were treated with ascorbate. The resulting survival times (in days) are displaced in table.

Stomach Bronchus Colon Ovary Breast

http://mypage.iu.edu/~mtrosset/StatInfeR/Data/cancer.dat

Let x1,...,x5 denote the samples displayed in table. Let y1,...,y5 denote the corresponding samples of the logarithms of the survival times:

>y1<-log(x1)

>y2<-log(x2)

>y3<-log(x3)

>y4<-log(x4)

>y5<-log(x5)

1. Construct two side-by-side boxplots, one of x1,...,x5,one of y1,...,y5. Which data, observed survival times or the transformed survival times, more nearly satisfy the ANOVA assumptions of normality and homoscedasticity? Explain?

2.Use ANOVA to investigate whether or not survival time differs with organ affected. Carefully state the null hypothesis that you are testing. Are you comparing population means of survival time or population means of log(survival time)? Use a significance level of =0.05 and organize your calculations in an ANOVA table. (preferable using R)

Solution