When scale variables exist in relationships with one another, we can use one variable to predict scores on a related variable using linear regression. The procedure involves using sample data to create a regression equation that can be fitted to the sample. The fit is based upon the least squares method and allows us to predict other, novel outcomes.