R2 - R squared

Gist

For regression, it is the proportion of variance explained

Mathematics

Mistakes

1. Maximizing the R2 is not the goal

Uses

1. It is good for comparing models with and without certain predictor variables
2. It is good for comparing to a null model

References

1. Gary King. 1986. “How Not to Lie With Statistics: Avoiding Common Mistakes in Quantitative Political Science”. American Journal of Political Science, 30, Pp. 666–687.