We want to find a regression model that explains the average amount of sleep at night of individuals. Our baseline regression model has as dependent variable (the -variable). The explanatory variables (the -variables) are: , , , and . Note that: - measures the average minutes of sleep per night. - measures the minutes worked per week. - is the age of a person in years. - measures how many years a person has been married. - is the hourly wage in US cents. Using the data in Computer_Exercise_11.xlsx, please answer the following questions. Assume throughout that assumptions MLR.1-MLR.6 are satisfied. 1. Consider the rules of thumb for when a variable should be considered in natural logarithms and when it should not (see slides Chapter 6). Based on these rules, should you change any of the variables in the baseline regression model into logarithms? If yes, please do so and explain your decision. 2. Estimate the baseline model by OLS (using if necessary - the appropriate logarithmic transformation of the variables, as discussed in part 1. above). Interpret the coefficient estimates. Are all variables individually statistically significant? What about the overall significance of the regression? 3. Without re-estimating your baseline model in part 2., what would be the estimated impact (and the corresponding standard error) of the amount of time spent working on that persons sleep record, if were measured in hours instead of minutes? 4. Which one of the variables in your baseline model in part 2. has the most important/largest impact on . To answer this question, compute the beta/standardized coefficients. 5. Now add an additional regressor to your baseline model: 2. Estimate this model (using if necessary - the appropriate logarithmic transformation of the variables, as discussed in part 1. above) by OLS. What is the estimated impact of a persons marriage record on his/her amount of sleep? 6. Evaluate the (joint!!!!) hypothesis that the years of marriage have no effect on a persons sleep, against the two-sided alternative hypothesis. 7. Based on your estimates from part 5., is there a turnaround point (see slides Chapter 6) of the effect of on ? If yes, after how many years does this turnaround point happen? 8. If you found a turnaround point in part 7., report which percentage of your sample falls below and above this point. 9. Which model fits the data better - the baseline model, or the model including 2? Why?