I'm not talking about outliers. I'm talking about our best estimates of average effects.