Consider a linear probability model (LPM) where the $Y_i$ is a dummy variable.

To test the presence of heteroskedasticity in our data, I executed the following code:

plot(y = LPM$residuals^2, x = LPM$fitted.values,
     ylab = "Squared residuals", xlab = "estimated probabilities")

But how am I supposed to interpret the graphic that results?

  • $\begingroup$ If you have a linear probability model, doesn't there have to be heteroscedasticity? $\endgroup$ May 18, 2021 at 15:23


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.