Correlated Confounding Variables Are Not Easily Controlled for in Large Survey Research

William H. Press

Results in epidemiology and social science often require the removal of confounding effects from measurements of the pairwise correlation of variables in survey data. This is typic…