d) Model 4 (Adj. R = 0.8569) 15. Why is Adjusted R preferred over simple R for comparing these models? * 1 point a) Adjusted R always gives a higher value. b) Adjusted R penalizes adding irrelevant predictors, rewarding parsimony. c) Adjusted R ignores p-values. d) Adjusted R only works for simple regression. 16. Model 4 includes only one predictor (ESGScore) and yields an Adj. R of 0.8569. This reflects: * 1 point a) A very poor model, since only one variable is included. b) Strong evidence that a single ESG composite score explains much of ROA. c) Overfitting, since the model is too simple. d) The need to add more variables regardless of their significance. 17. Comparing Model 1 and Model 2, Model 2 drops EnvironmentalScore and ESGScore, reducing Adj. R slightly. This demonstrates: * 1 point a) That including more variables always improves explanatory power. b) That parsimony may be preferred if insignificant variables are excluded. c) That adjusted R should not be used for model comparison. d) That dropping variables always worsens the model. 18. Which model best balances parsimony and explanatory power? * 1 point a) Model 1, since it has the highest Adj. R even with more variables. b) Model 2, since it uses fewer variables with only a small drop in Adj. R. c) Model 3, since it retains significant variables with strong economic meaning. d) Model 4, since it is the simplest model with one significant predictor. Part 3: Data Cleaning Basics 19. Which of the following best describes the purpose of data cleaning before running regression analysis? * 1 point a) To improve the theoretical foundation of the model b) To ensure that errors, missing values, and inconsistencies do not bias results c) To make the regression output look more significant d) To automatically increase the adjusted R 20. Why is handling missing data important in dataset preparation? * 1 point a) Missing values are automatically ignored by statistical software without consequences b) Missing data can lead to biased estimates if not addressed properly c) Missing data always improves model parsimony d) Missing values can be left as they are in regression models 21. Which of the following is not a common method of handling missing data? * 1 point a) Deleting rows with missing values b) Imputing values (mean, median, regression-based) c) Replacing missing data with outliers d) Using "missing indicators" in the model 22. Suppose 10% of a variable's observations are missing. Which method is most appropriate when the missingness is completely random? * 1 point a) Dropping all rows with missing values b) Replacing missing values with the mean of the variable c) Assigning all missing values a zero d) Ignoring the problem since it is random 23. Why should outliers be carefully examined in regression datasets? * 1 point a) Outliers always improve explanatory power b) Outliers can heavily influence regression coefficients and distort relationships c) Outliers have no effect on OLS estimation d) Outliers automatically indicate data entry errors 24. Which diagnostic measure is commonly used to detect influential outliers in regression? * 1 point a) Correlation coefficient b) Variance Inflation Factor (VIF) c) Cook's Distance d) Adjusted R 25. When deciding how to treat outliers, the most important principle is: * 1 point a) Always delete them to improve model fit b) Replace them with the mean value c) Investigate whether they are errors, unusual but valid cases, or influential distortions