Clinical Biostatistics: Mouse livers exercise, answer 1

1. What assumptions would we have to make about the data for these tests. Which one is clearly violated here?

Suggested answer

The assumptions of two sample t method are:

The observations and groups are independent. We should not have, for example, a group of 10 observations where there are 2 subjects with 5 observations on each. We should not have links between observations in the two groups, such as a matched study where each subject in one group is matched, e.g. by age and sex, with a subject in the other group.
The outcome variable, diameter of canaliculus, follows a Normal distribution in each population.
The variances of the outcome variable are the same in each population.

The assumptions of the large sample z method for two means are

The observations and groups are independent, as for the two sample t method.
The samples are large enough for the standard errors to be well estimated. A rule of thumb is at least 50 in each group.

The assumption which is clearly violated is that the observations are independent. We have 280 measurements from only three mice. The measurements on the same mouse may be more similar than they are to measurements from another mouse.

Back to question.

Back to Clinical Biostatistics index.

To Martin Bland's home page.

This page maintained by Martin Bland.
Last updated: 31 July, 2006.