The article discusses a new method called quantile regression for analyzing data with many small groups of related observations. Instead of just looking at the average relationship between variables, this method looks at different percentiles of the data to get a more accurate understanding of how variables are related. The researchers used a robust approach called generalized estimating equations to make their conclusions, which can be easily applied. They also suggested ways to check if the model fits the data well. The new method was tested in a toxicology study involving animals.