I spent years reading articles, text, etc about the use of residuals to determine model violation, but I have a hard time telling if they actually have occurred and how much the violation matters. I am particularly concerned with non-constant variance and non-linearity for my linear model.

This is the residuals against predicted and qq plot for normality, the former seems to me to not suggest heteroskedascity, the qq plot suggests outlier although I am not sure how serious this is. The white test (not shown) suggested either heteroskedascity or a misspecified model.

enter image description here

I ran partial regression plots as reccomended. I assume this does not suggest non-linearity

enter image description here

  • 1
    $\begingroup$ I find this post to be particularly helpful in interpreting QQ-plots: stats.stackexchange.com/a/101290/176202 Your QQ-plot does not show any obvious outliers to me. An outlier would be a deviating point, but it looks more like the residuals are left skewed. As for the other plots, it's hard to judge model specification without knowing the model and what kind of data these are. $\endgroup$ – Frans Rodenburg Apr 18 at 0:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.