Statistics 103 Problem Set 2

Statistics 103 Problem Set #2

Part I – Male Fruitflies

1. Examine the data.  Briefly discuss how the data is set up. 

    

2. Create a dummy variable called ‘ACTIVE’ that is =1 if the fruitfly has at least one partner and =0 if he has no sexual partners.  Create a dummy called ‘ACTIVE1’ if PARTNERS=1.  Create a dummy called ‘ACTIVE8’ if PARTNERS=8.  Why do we not need to create a variable called ‘ACTIVE0’?

    

3. Report the means, standard deviations, and medians of all variables.  Discuss the meaning of the mean of the TYPE, PARTNERS, and ACTIVE variables.

    

4. Report and compare the means, standard deviations, and medians for the LNGEVITY variable if the fruitfly has 0, 1, and 8 partners.  Does this suggest a relationship between sexual behavior and lifespan?

    

5. Suppose you are interested in finding the determinants of lifespan for male fruitflies.  You will test LNGEVITY = f(PARTNERS, THORAX, SLEEP).  What do you predict for the sign of the coefficients on the independent variables?

    

6. Run an OLS regression with LNGEVITY as your dependent variable and ACTIVE1, ACTIVE8, THORAX, and SLEEP as your independent variables and report the coefficients and t-stats for these variables as well as the constant in a table.  Also report the r-squared.  Discuss the results.  Were your predictions correct?  Interpret the coefficients?  What variables are significant at the .05 level?

    

7. Run another OLS regression with LNGEVITY as your dependent variable and THORAX, SLEEP, and TYPE as your dependent variables for those you are active only.  Prepare another table similar to the table needed for question 6 above.  Discuss these results.

    

8. What conclusions can you make about the sexual and sleeping behavior of male fruitflies and their lifespan?

Part II – CEOs

1. Examine the data.  Briefly discuss how the data is set up.  Are there any outliers in the data?  In other words, are there any observations that are considerably higher or considerably lower than the remainder of the data points?

    

   Create size Dummies for each SMALL, MEDIUM, and LARGE companies where SMALL are companies with up to $6 billion in sales, LARGE are companies with over $20 billion in sales, and MEDIUM are all those in between. 

    

2. Report the means, standard deviations, and medians of all variables.

1. Report the correlation coefficients for the variables.  Discuss your findings.

    

2. Suppose you are interested in what determines salary and/or total compensation.  Discuss which variables you should include on the right hand side of these regressions.  Be careful to discuss the concept of correlation vs. causation here.

    

3. Run an OLS regression with SALARY as your dependent variable and any appropriate independent variables of your choosing.  Report the appropriate coefficients, t-stats, and r-squared in a table.  Discuss the results.

    

4. Repeat question 6 for TOTCOMP.  Compare your results to those of question 6.

    

5. Using the variables you already have in place and/or any new ones you feel are appropriate, develop what you feel are the most appropriate OLS regressions for SALARY and for TOTCOMP.  Be sure to discuss your reasons why these are most appropriate and interesting.

    

6. What conclusions can you make about the determinants of CEO salary and compensation?  If you were to extend this study, what additional variables would you want to add and why?