Evaluating alternative simple regression models, not-for-profit. (Chapter Appendix) ^ Kathy Hanks, executive assistant to the president of Eastern University, is concerned about the overhead costs at her university. Cost pressures are severe, so controlling and reducing overhead is very important. Hanks believes overhead costs incurred are generally a function of the number of different academic programs (including different specializations, degrees, and majors) that the university has and the number of enrolled students. Both have grown significantly over the years. She collects the following data:

INSTRUCTIONS
Form groups of two or more students to complete the following requirements:
REQUIRED
1. Plot the relationship between overhead costs and each of the following variables: (a) number of academic programs and (b) number of enrolled students.
2. Compare and evaluate the two simple regression models estimated by Hanks. Use the comparison format employed in Exhibit 10-12.
3. What insights do the analyses provide about controlling and reducing overhead costs at the university?
4. What are two important issues that would suggest the predictions are unreliable?

Overhead Costs Number of (in thousands) Programs Enrolled Students ear S16,200 23,040 20,160 24,120 23,400 27,720 28,440 24,120 27,360 35,640 37,440 45,720 3,400 5,000 2,600 4,700 3,900 4,900 5,700 3,900 3,500 3,700 5,600 7,600 29 36 49 53 54 58 4 72 83 73 101 103 10 12 * Regression 1. Overhead costs a+(b X number of academic programs) Variable Coefficient S8,553.30 $288.76 Standard Error t-Value 2.14 5,08 Constant Independent variable I: number of $4,002.41 academic programs r 0.72; Durbin-Watson statistic 2.07 The adjusted r0.693 56.80 Regression 2. Overhead costs-(bX number of enrolled students) Variable Coefficient S7,190.10 S 4.53 Standard Error t-Value 1.18 3.52 Constant S6,081.45 Independent variable 1: number of 1.29 enrolled students r 0.55: Durbin-Watson statistic 0.82 The adjusted r0.509