A note on confidence intervals with extended least squares parameter estimates