1. Why does the use of ANCOVA rely upon the assumption that the regression lines for the relationship between the dependent variable and the covariate for all levels of the factor being investigated are reasonably parallel?
2. The data given below are for the systolic blood pressure of ten adult males in an untreated control and ten more males given the experimental drug Systolsyn B.
(a) Draw a graph of the data for blood pressure in relation to age for both groups. Does an ANCOVA seem appropriate for examining whether there is a difference between the experimental group and the control? Why?
(b) Use a statistical package to run a preliminary analysis for parallelism on these data by specifying the interaction between blood pressure and age. Does this show a significant lack of parallelism?
(c) How might you proceed with this analysis?