Genetics Example

Genetics Example

Sample Size for Analysis of Variance with unequal n’s

Does a candidate gene influence impaired glucose tolerance in a specified population thought to be susceptible to diabetes? A study is planned to compare a measure of glucose tolerance between groups with different genotypes. The percentages of the A1A1, A1A2, and A2A2 genotypes are expected to be about 81%, 15%, and 4% respectively among normals, about 60%, 26%, 14% among patients with NIDDM, and about halfway between these in the population to be studied. A pilot study suggests a mean for the outcome measure of 5.1 in the A1A1 groups with a standard deviation of 2.0; results would be of interest if the group with the A1A2 genotype had an average response as low as 4.6.

Sample size calculations could be based on using the two-sample t-test to compare those with A1A1 to those with A1A2, or on analysis of variance comparing the three genotype groups, or on using a contrast to test for a linear dose response effect. nQuery Advisor® provides easy-to-use sample size tables allowing for unequal group sample sizes for all three options.

Select File … New, and in the Study Goal and Design Box, select Means, >Two Groups, and Test, highlight One-way analysis of variance and check the Unequal n’s box. Fill in the significance level, and the number of groups, and bring up the side-table (see Next Step) to fill in the information about means and sample sizes. You can specify sample sizes for each group directly and solve for power or specify the ratios of sample sizes and the power and solve for the total sample size.

If the study population can be enriched with a few additional individuals with the A2A2 and A1A2 genotypes (4 and 23 respectively in this example), 72 fewer individuals with the A1A1 genotype will be needed, and the overall sample size could be cut from 460 to 415 while maintaining the same power.

nQuery Advisor side tables for MGT0U:1

Sample size tables or plots showing the effect of different scenarios on required sample size can be printed directly or copied into a document.