Multiple linear regression + QQplots problem Includes pics

Click For Summary
SUMMARY

The discussion centers on the challenges of performing multiple linear regression when the residuals are not normally distributed, as assessed through QQ plots. The user observed that while 95% of the data aligns with the normal line, 5% deviates significantly, raising concerns about the validity of the regression results. Recommendations include exploring robust regression techniques available in the MASS package in R to address the non-normality of residuals and improve the reliability of coefficient estimates.

PREREQUISITES
  • Understanding of multiple linear regression principles
  • Familiarity with QQ plots for assessing normality
  • Knowledge of R programming and the MASS package
  • Concept of robust regression techniques
NEXT STEPS
  • Research robust regression methods in the MASS package in R
  • Learn how to interpret QQ plots for residual analysis
  • Explore data transformation techniques to achieve normality
  • Investigate alternative regression models for non-normal data
USEFUL FOR

Data analysts, statisticians, and researchers involved in regression analysis who need to ensure the validity of their models despite non-normal residuals.

emelie_earl
Messages
3
Reaction score
0
I want to do multiple linear regression, but one of the requirements is the residuals to be normally distributed, and I can check that with QQplots but then the QQ plot shows it is about 95% of data fit into the normal line, but 5% is way off!

can I still proceed ?*or do I have to find a way to transform the data ?*


5.jpg


5_residuals.jpg
 
Physics news on Phys.org
Your plots show serious non-normality in the error structure, with (as you've noted) problems in the tails, and since least squares is incredibly non-robust you're correct to be concerned.
1) Have you noticed any strange behavior in your estimates (coefficients with signs opposite what you might expect)?
2) Have you tried a robust regression? The MASS package in R provides several good options.
 
statdad said:
Your plots show serious non-normality in the error structure, with (as you've noted) problems in the tails, and since least squares is incredibly non-robust you're correct to be concerned.
1) Have you noticed any strange behavior in your estimates (coefficients with signs opposite what you might expect)?
2) Have you tried a robust regression? The MASS package in R provides several good options.


Thank you!
I will try Robust regression.
 

Similar threads

Undergrad Linear regression, feature scaling, and regression coefficients
  • · Replies 8 ·
Replies
8
Views
3K
Undergrad Expected coefficient change from simple to multiple linear regression
  • · Replies 4 ·
Replies
4
Views
2K
Undergrad Coefficient sign flip in linear regression with correlated predictors
  • · Replies 13 ·
Replies
13
Views
5K
Undergrad Normality of errors and residuals in ordinary linear regression
  • · Replies 3 ·
Replies
3
Views
2K
Undergrad How (un)reasonable is graphic linear regression in a monolog graphic?
  • · Replies 6 ·
Replies
6
Views
2K
Undergrad Linear regression and random variables
  • · Replies 30 ·
2
Replies
30
Views
5K
Undergrad What Are the Key Differences Between OLS and GLM Models in Statistical Analysis?
  • · Replies 3 ·
Replies
3
Views
3K
Undergrad Can Alternative Least Squares Methods Be Used in Linear Regression?
  • · Replies 6 ·
Replies
6
Views
3K
Undergrad What is Nonlinear Least Squares in Regression Analysis?
  • · Replies 4 ·
Replies
4
Views
2K
Undergrad How to propagate errors through a regression & non-linear model?
  • · Replies 8 ·
Replies
8
Views
3K