Contrast

is a comparison of the effects (or parameter estimates) produced for the different levels of a treatment factor. If the factor levels correspond to different amounts x of a substance (for example a drug or a fertiliser), the POL function allows linear, quadratic, cubic or quartic polynomials of x to be fitted, or the REG function allows you to define your own regressions on x. Alternatively, the COMPARISON function allows comparisons to be defined between qualitative levels (ANOVA only).