I. True/False/Uncertain - Briey explain. No credit without an explanation (8 marks each). 1. With 5fold cross validation1 20% of the sample data are never used to estimate the model, just to estimate the TMSE. 2. TMSE with crossvalidation is not always decreasing when adding more Xs, so it is a good estimator for GMSE. 3. Using the bootstrap procedure for model averaging} not every observation of the original sample will be selected at least once. 4. One can use the LRT to test the hypothesis H0 : B5 = O in a legit model. 5. The slope of the probit model is maximized when (I) = 0.5