ABSTRACT
Contrast coding offers a statistically more powerful alternative to the usual Omnibus F-tests of main and interaction effects programmed by default in popular statistical packages. Contrast coding can offer an additional advantage of providing a focused test of the specific relationship being hypothesized, rather than assuming and testing a one-size-fits-all relationship. We demonstrate the use of contrast coding first in research using ANOVA and extend it to applying contrast coding tools in regression tests as well. In response to critiques urging researchers to look beyond p-values, some researchers are using reliance on effect sizes as indicators of material significance. Therefore, we demonstrate how to determine effect sizes using contrast coding. We provide numerous examples to demonstrate how to apply contrast coding, how contrast coding results differ from traditional ANOVA or regression results, and when contrast coding offers advantages in analyzing research results and when it may not.
