ABSTRACT
One frequently encountered problem in scientific research is to assess whether two regression models, which describe the relationship of a same response variable Y on a same set of predict variables x1, · · · ,xp, for two groups or two treatments, etc., are the same or not. For example, the relationship between human systolic blood pressure (Y ) and age (x1) can be well described by a linear regression model for a certain range of age. Suppose that a linear regression model of Y on x1 is set up for each of the two gender groups, male and female. It is interesting to assess whether the two models are different and so two separate models are necessary for the two gender groups, or the two models are similar and therefore only one model is required to describe how blood pressure changes with age for both gender groups.
