A quotient of independent measurements as outcome - A statistical method to compare groups in biological experiments

The methodology in this paper was proposed to analyze biological data. The variable of interest, W, was a quotient of measurements; W: = X/(Y . Z). Th e peculiar problem was that X, Y, and Z could only be determined in three i ndependent units of observation which were destroyed in the course of each measurement. A test statistic for the comparison of two treatments in W is proposed which is based on robust measures of location and dispersion in or der to account for outliers in the data. Simulations showed that the test s tatistic proposed is approximately normally distributed, even for small sam ple sizes.