Statistical modeling relies on correctly partitioning variation into between-group and within-group sources. But what happens when principal component analysis (PCA)—often used for ...