I ran a regression and the intercept is statistically insignificant (the p-value is greater than 0.05). I tried to look in some textbooks as to how to handle this scenario but I am still unsure. One textbook I looked at 'Basic Econometrics'by Gujarati and Porter says that if the intercept is insignificant then we have a regression through the origin and that if we remove the intercept, in this case, the model will be more precise. On the other hand, 'Introductory Econometrics'by Chris Brooks says that even if the intercept is insignificant, we should not remove it from the model.

Which one of these textbooks is correct? Should I leave the insignificant intercept in the model or run a regression through the origin?

**Contents**hide

#### Best Answer

If the true data generating process implies that Y=0 whenever X=0, then exclude the intercept, otherwise keep the intercept, even it is not significant. One example of this: response variable (Y) is distance the can run, and the covariate (X) is volume of gasoline consumed by the car. When X = 0, Y = 0.

### Similar Posts:

- Solved – Regression: Insignificant Intercept
- Solved – Regression: Insignificant Intercept
- Solved – Ridge Regression: When should the intercept be included ? What is the purpose of the intercept term?
- Solved – Is it ok to remove the intercept in a linear regression model (OLS) if the results are really good?
- Solved – Constant in OLS model