Assignment Four: Test on Friday Oct 10th

This is an assignment that you should do, but don't bring your list file to the test . I am going to do the assignment too, and a good part of the test will be based on my list file, which will be provided.

In my public directory is a file called restaurant.dat. Get a copy with

cp /student/jbrunner/public/restaurant.dat .

Include the final period; it refers to your current directory. Here is a description of the data. The cases in the file are restaurants, and the variables are listed below. My variable names are given in CAPITAL LETTERS. Use my variable names. It will make the list file for the test look more familiar.

  1. ID: Identification number
  2. OUTLOOK: Rating (we don't know by whom) from 1 = very unfavorable to 7 = very favorable
  3. SALES: (Gross 1979 sales in $1000s
  4. NEWCAP: New capital invested in 1979, in $1000s
  5. VALUE: Estimated market value of business in 1979, in $1000s
  6. COSTGOOD: Cost of goods sold as a percentage of sales
  7. WAGES: Wages as a percentage of sales
  8. ADS: Advertising as a percentage of sales
  9. TYPEFOOD: 1 = Fast food, 2 = supper club, 3 = other
  10. SEATS: Number of seats in dining area
  11. OWNER: 1 = sole proprietorship, 2 = partnership, 3 = corporation
  12. FT.EMPL: Number of full time employees
  13. PT:EMPL: Number of part time employees
  14. SIZE: "Size" of restaurant. 1 = 1 to 9.5 full-time equivalent employees, 2 = 10 to 20 full-time equivalent employees, 3 = over 20 full-time equivalent employees, where full-time equivalent employees = (number of full time + .5(number of part time).

Make a command file that reads and labels the data (including variable labels, and value labels where appropriate). Run descriptives to obtain means, standard deviations, etc. of the quantitative variables. Run frequencies to get frequency distributions of the categorical variables (a variable may occur in both sets).

Perform a bunch of reasonable tests to answer questions you could ask of these data, LIMITED to t-tests, one-way anova, scatterplots, correlation, simple regression, and chi-square tests of independence. Don't hesitate to consider the creation of new variables. For some of the analyses, you may consider OUTLOOK to be quantitative.