ECON321 Lecture Notes - Lecture 9: Econometrics, Regression Analysis
5 Dec 2014
School
Department
Course
Professor
Document Summary
We have included a 0, an intercept term in our simple regression. However, we will sometimes have relationships for which our intercept should be zero. The problem with regressions through the origin, is that, unless 0 actually is zero, our estimator of 1 will be biased. Multiple regression analysis involves models with more than one independent variable. We almost always use multiple regression analysis rather than simple regression analysis in applied research. Our zero conditional mean assumption in this example is then: General format for multiple linear regression model with k independent variables: Yi= 0 1xi1 2xi2 k xik ui, i=1,,n where 0 is the intercept, and 1,, k are the slope parameters. Again, ui are the error terms and the xs are the independent variables. Specifically, 1 is the partial effect of x1 on y. The interpretation is: controlling for all other characteristics, holding all else (x2 ,xk) fixed, 1 is the effect of x1 on y.
