Econometrics uses statistical models to develop and test economic theories. Pioneered by Klein, Frisch, and Kuznets, all Nobel Prize winners. Applied econometrics develops new hypotheses from existing data
Standard error measures regression model prediction accuracy. Data should be organized in five-column table. Independent variable labeled as X, dependent as Y
Multinomial logistic regression models nominal outcomes using predictor variables. Example: studying occupational choices influenced by education and father's occupation. Data analysis example uses hsbdemo dataset with 200 students
Perfect multicollinearity occurs when predictors have exact linear relationships. Imperfect multicollinearity involves nearly exact linear relationships. Perfect collinearity typically arises from redundant variables or dummy variables. Wide datasets often require advanced techniques like Bayesian modeling
GLS estimates unknown parameters in linear regression with residual correlation. First described by Alexander Aitken in 1935. Requires knowledge of covariance matrix for residuals. Minimizes squared Mahalanobis length of residual vector
MSE measures average squared difference between estimated and true values. MSE is always positive due to randomness or lack of information. MSE incorporates both variance and bias of estimator. For unbiased estimators, MSE equals variance