Update: I have included a modified version of this summaryR() command as part of my package tonymisc, which extends mtable() to report robust standard errors. The tonymisc package is available on CRAN through the install.packages() command.

If you have the right R commands at your disposal, it is simple to correct for heteroskedasticity using the robust correction that is commonly-used among economists. I recorded a video tutorial to describe the simplest (and most flexible) way I know to get R to compute robust standard errors.

The key is to use a "summary-style" command that has an option to correct for heteroskedasticity. The command I like to use is called summaryR(). Here is the script file with the summaryR() command.

I found this function on an R-help discussion board where several people were answering someone's question about extending the summary.lm() command.

I deserve none of the credit for writing this (credit goes to John Fox), but I consider it my duty to point out how nice this function is. I demonstrate how to use the function in this video

Here are is the script file I used in the video:

Here's a link to the data set.

Hi! The summaryR() command sounded like a dream to me but when I try it I get "Error:could not find function summaryR". I have the car package and I have installed your designed package.

ReplyDeleteBest regards David!

Hey David,

DeleteI ran into the same problem. Did you manage to find a solution to it?

Brandon

Hi Tony, I added return(sumry) in the function after print(sumry) which outputs a summary.lm object which is helpful for constructing output tables (using coef) and so on.

ReplyDeleteThanks for this!

Thx a bunch for the summaryR-function means alot for me to have found this easy to follow instruction - been searching like crazy for a couple of weeks.

ReplyDeleteBest regards Jesper