day7lm - 2/24/12 1 PADP 8130: Linear Models Specifca(on...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon

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

View Full DocumentRight Arrow Icon

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

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 2/24/12 1 PADP 8130: Linear Models Specifca(on Angela Fer¡g, Ph.D. Specifca¡on is about deciding which regressors to put in the regression. • OmiHed variable bias • Including irrelevant variables • Specifca¡on tests – OmiHed Variable Test – Chow Test – RESET 2/24/12 2 What is the consequence of omiSng an important regressor? y = X 1 β 1 + X 2 β 2 + ε y = X 1 β 1 + ε b 1 = (X 1 ' X 1 )-1 X 1 ' y = (X 1 ' X 1 )-1 X 1 ' ( X 1 β 1 + X 2 β 2 + ε ) = β 1 + (X 1 ' X 1 )-1 X 1 ' X 2 β 2 + (X 1 ' X 1 )-1 X 1 ' ε E ( b 1 ) = β 1 + (X 1 ' X 1 )-1 X 1 ' X 2 β 2 True model: Es-mated model: Omi2ed Variable Bias! Direc:on of Bias E ( b 1 ) = β 1 + (X 1 ' X 1 )-1 X 1 ' X 2 β 2 If X 1 ' X 2 > 0, then b 1 will be biased upwards. If X 1 ' X 2 < 0, then b 1 will be biased down. 2/24/12 3 What about variance? • The variance is smaller when you omit an important variable. • So, you’ll es:mate the wrong answer really precisely. Var ( b 1 ) = σ 2 ( X 1 ' X 1 ) − 1 < σ 2 ( X ' X ) − 1 because ( X 1 ' X 1 ) − 1 < ( X ' X ) − 1 True model Es:mated model What is the consequence of including irrelevant variables in the regression? y = X 1 β 1 + ε y = X 1 β 1 + X 2 + ε y = X 1 β 1 + X 2 β 2 + ε y = X β + ε b = (X ' X)-1 X ' y = (X ' X)-1 X ' ( X β + ε ) E ( b ) = β = β 1 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ True model: Es-mated model: Unbiased! 2/24/12 4 But, what about variance? • The variance is larger when you include an irrelevant variable. • So, you’ll es:mate the right answer imprecisely. Var ( b ) = σ 2 ( X ' X ) − 1 > σ 2 ( X 1 ' X 1 ) − 1 because ( X 1 ' X 1 ) − 1 < ( X ' X ) − 1 Es:mated model True model Specifca:on Tests OmiHed Variable Test ¡ put in a rich set oF...
View Full Document

This note was uploaded on 03/28/2012 for the course PADP 8130 taught by Professor Fertig during the Spring '12 term at LSU.

Page1 / 7

day7lm - 2/24/12 1 PADP 8130: Linear Models Specifca(on...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online