If you run a linear regression in R, then you have to check regression assumptions. In general, with linear regression you model a linear relationship between each predictor (independent variable) and the criterion (dependent variable). When assumption testing you have to check whether there is a linear relationsship between those variables (and not a u-shaped, inverted u-shaped, or curved relationship) because otherwise you would need a different modeling strategy (e.g., polynomial regression).

For many linear regression assumptions in R, there are several testing strategies. For the linearity assumption, there are at least two options:
1. Running bivariate scatterplots
2. Running the rainbow-test from the lmtest-package

свернуть