# Chapter 7 - ECMT1020 Chapter 7 Multiple Linear Regression...

This preview shows pages 1–9. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ECMT1020 Chapter 7 Multiple Linear Regression Analysis III Dr Boris Choy for ECMT1020 Topics covered 1. Multicollinearity 2. More regression models (a brief introduction) 3. Multiple linear regression analysis using Excel/KaddStat Additional teaching materials Excel file: Chapter 7.xls References Black 15.4, 14.1‐14.4, 15.2‐3 Learning Objectives Understand multicollinearity Detect multicollinearity Know other regression models Perform regression analysis using Excel/KaddStat Interpret the output Draw conclusion based on the output Multicollinearity Multicollinearity In multiple regression analysis, multicollinearity exists if some of the predictor variables are highly correlated. Adverse effect of multicollinearity: [1] These highly correlated predictor variables do not provide new information to the regression analysis. [2] Estimation of regression coefficients is subject to a large standard error regression coefficients may fluctuate dramatically standard errors are overestimated estimated regression coefficients are far from their true values a smaller t value in the hypothesis test of individual predictor No evidence of linear relationship even though there is one. Negative coefficients may be observed when there is a positive linear relationship. Multicollinearity Multicollinearity does not mean that the model is a failure – but it can be improved. Economic variables are always correlated. But if they are highly correlated, then it will be difficult to unravel their separate effects to the response variable. We may fail to detect the an effect of a predictor variable on the response variable because of multicollinearity even though there is an effect. Multicollinearity . separately and estimate cannot but estimate can We model. regression linear simple a to reduces model The ) ( becomes then say nearity), multicolli (perfect correlated perfectly are and If model he Consider t 2 1 2 1 1 2 1 2 2 1 1 1 2 2 1 2 2 1 1 x y x x y x x x x x x y ， Detecting Multicollinearity A simple way to detect possible multicollinearity is to assess the correlation coefficient between each pair of predictor variables – inspect the correlation matrix....
View Full Document

## This note was uploaded on 08/20/2011 for the course ECMT 1020 at University of Sydney.

### Page1 / 32

Chapter 7 - ECMT1020 Chapter 7 Multiple Linear Regression...

This preview shows document pages 1 - 9. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online