ECO220Y1 Lecture Notes - Lecture 20: Regression Analysis, Homoscedasticity, Statistical Significance
27
ECO220Y1 Full Course Notes
Verified Note
27 documents
Document Summary
Multiple regression allows us to control for experience, discipline, etc to isolate effect of sex: control for lurking variables. Multiple coefficients because more than one x variable with k explanatory variables. Simple vs multiple regression yi = 0 + 1x1i + 2x2i + + kxki + i k = 1 in simple regression. I explains everything aside from x1, x2 xk that explains y. No simple formula to find coefficients: need matrix algebra or software, ols estimate solves min (yi - i)2, residuals: ei = yi - i, linearity. Each x linearly related to y: errors independent, homoscedasticity of errors, normally distributed errors, constant included, each x not correlated to error. Hypothesis testing: use t test with = n - k- 1, sbj obtained with software, test of statistical significance. 0: h0: j = j, h1: j j, h1: j 0.
