I wish to run a linear regression model, with a dependent variable Y and several explanatory variables.

The distribution of Y looks like this:

enter image description here

Clearly not normally distributed. The sample size is about 40 observations.

In this problem I wish to use SPSS, and I didn't find an option for robust regression. This leads me to bootstrapping.

Can I use linear regression with bootstrapping when I have skewed data like this, with an outlier? Do you have other suggestions for modeling this kind of data?

Any tips will be most appreciated ! Thank you !

| cite | improve this question | | | | |
  1. Linear regression does not require a normally distributed dependent variable, only the error term should be normally distributed ( but that is only important in small samples)

  2. A bootstrap with outliers means that your estimates of the sampling distributions are going to be bimodal. That is fine if you think the outliers are real, but you again rely a lot on those special observations and they better be right.

| cite | improve this answer | | | | |

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.