ECON20003 Lecture Notes - Lecture 12: Regression Analysis, Pearson Product-Moment Correlation Coefficient, Tachykinin Receptor 1
Document Summary
Simple regression refers to equations with only one explanatory variable (x: multiple regression is equations with more than one explanatory variable, degrees of freedom for t-tests and interval estimates changes from n-2 to n-k-1. Interpretation changes to recognise that other variables are being held constant: change in y when x1 changes by 1 unit and x2 xn are held constant. If variables are non-linear (x2) interpretation changes accordingly. 95% interval estimate where degrees of freedom (n-k-1) is 1076: Maximum sampling error is given by (cid:1872)(cid:4672)2, (cid:2869)(cid:4673)(cid:1871)(cid:1857)( (cid:2869)) and helps find an interval estimate: to check for heteroscedasticity examine the residuals (cid:1857)=(cid:1851) (cid:1851) , use white heteroscedasticity robust standard errors to allow for possible existence of. If white standard errors are slightly larger then use them instead of normal heteroscedasticity. Goodness of fit: measure of goodness of fit of regression lines is r2, r2 measures the proportion of variation in the dependent variable explained by variation in the explanatory variables.