Question

S a first step in developing a new drug to treat cancer in humans, an initial study of the drug is undertaken in rats. Three hundred rats with cancer are studied, and each is assigned to one of three treatments. One hundred rats are randomly assigned to a high dose of the new drug, 100 are randomly assigned to a low dose, and 100 are randomly assigned to a control group (no drug). After six months, each rat is examined and classified as having developed no tumors, one tumor, or two or more tumors.

Which of the following would be an appropriate alternative hypothesis in this study for a chi-square test for homogeneity of tumor response across treatments?

Answer 1

The appropriate alternative hypothesis for a chi-square test for homogeneity of tumor response across treatments would be:

- There is a significant association between the treatment received and the number of tumors developed by rats.

In other words, the null hypothesis would be that there is no association between the treatment received and the number of tumors developed by rats. The alternative hypothesis would state that there is a significant association between these variables. This hypothesis can be tested using a chi-square test, which would compare the observed frequencies of tumor response in each treatment group to the expected frequencies if there were no association between treatment and tumor response.