ABSTRACT
Whether observed differences between groups exist independent of other third variables can be assessed with a procedure called analysis of covariance (ANCOVA). But ANCOVA is useful even in well conducted experiments in which the random assignment procedure nearly guarantees that the groups will be equal on all variables other than the variable being manipulated. The ANCOVA, through the statistical control process, can accomplish this if you had the foresight to measure those things likely to be related to the outcome variable. The ANCOVA summary table also provides a test of the significance of political discussion, controlling for political group. ANCOVA has all the same assumptions as ANOVA and multiple regression, but it also has one additional assumption called homogeneity of regression. Statistically removing the relationship between income and taxes paid would be statistically removing the mechanism that produces the differences in how much taxes are paid in the first place.
