In this SLP assignment, you're going to experiment with simple hypothesis testing procedures that will build on your previous SPSS skills, involving calculation of confidence intervals, T-tests, and crosstabs.

You'll be using data relating to both corporate performance and the CEO's golf game. Part 1 of the assignment uses the same data employed in the study described by Bryant, and you have a chance to replicate the findings described therein, and perhaps improve on them yourself. Part 2 (optional) of the assignment involves a more complicated data assembled around the same question; in this part, you have a chance to look at some other aspects of the problem and explore the data to find interesting connections.

Part 1

The data described in the Bryant study can be found in the attached Excel Spreadsheet .

As always, the first step is to form a basic understanding of the data and the relationship to it through appropriate descriptive statistics and frequency tabulations. By now, either through practice or through reading, you should have a pretty good idea of what an appropriate description of the data might be; prepare an appropriate description as an introduction to your report.

The Bryant article reports its conclusion based on a subset of 44 cases from the 51 in the data set. This involves identifying 11 low performing companies, 22 middle performing companies, and 11 high-performing companies; calculating the mean golf handicap of the CEOs running these companies for each of the three groups; and in deciding whether the differences between these means are sufficiently great that we can believe they're real rather than a matter of chance. The article does not tell you which seven of the cases were removed before these calculations were made; all it says is that "Following accepted statistical techniques, Mr. Crystal removed seven chiefs from the analysis because they distorted the trend lines." You'll have to decide which seven were removed so that the numbers come out as reported. This turns out to be more complicated than it appears, and you may have to try a number of times before you actually replicate the numbers reported in the article (and even then you may not be able to replicate them to more than one decimal place). Let's just say that some slightly odd criteria were perhaps used to make this conclusion.

First, you'll need to divide the 51 cases into the three groups. For this, use stockrate as your measure and order-rank the companies from low to high; you can use frequencies to do this. Then recode into different variables to assign code numbers to the bottom 13, the middle 25, and the high 13 (if you are not familar with this procedure, please see Presentation 11 or Dr. Gupta's book p. 2-30 ff.) . This will give you your preliminary break on the groups, but you still need to decide which cases to remove to get down to the right 11-22-11 break.

There are a couple of choices in SPSS as to how to carry out this procedure. The easiest is to use ANALYZE:COMPARE MEANS: MEANS , using handicap as your dependent variable and your group break variable as independent. This will give you the means you seek; if they're not what you want, go back and reconfigure your groups, recalculate your means, and keep doing this until you have the values that are reported in the article, or close to them.  Or you can use DATA:SPLIT FILE:COMPARE GROUPS using your grouping variable to obtain basic descriptive statistics including means.

When you have successfully replicated the article's data analysis, report your findings, what you did to create them, and your evaluation of the procedures employed in the article.

The next step is to decide whether these differences are real or not. The article makes the claim that they are; see what you find. A good first step would be to calculate confidence intervals for the means of the three groups.  Following the procedures described in Presentation 6 , prepare graphs of confidence intervals for the means of handicap for each of the groups.  Interpret the results.

Then you should conduct independent samples t-tests to decide whether the differences between the groups are supported. This will involve three tests: group 1 vs. group 2, group 2 vs. group 3, and group 1 vs. group 3. In module five, we'll examine a technique for making these analyses simultaneously; for now, do it this way.  Following the procedures described in Presentation 6 , test the hypotheses of difference between the groups using independent samples t-tests.  Remember to enter the appropriate group numbers in the t-test procedure. Interpret the results.

The file also contains some data that were not part of the original data set, relating to location of the company, and its revenues and profits in 2004 and 2006; where these are not reported, the company had dropped out of the Fortune 500 through change or acquisition, and thus the numbers were not readily available. Company headquarters has been recoded into a variable called region , with the values (East, South, Midwest, and West). There has also been added a variable called survivor , which is coded as 1 if the company is still in independent existence in 2006 and 0 if it is not. Use a contingency table, following the procedures described in Presentation 10, to test a hypothesis about the relationship between region and survivor . Conduct an appropriate analysis, interpret, and report your findings.

Conduct one other analysis of your choice on these data - explore something that interests you, using any statistic of your choice.  Explain what you did, and why you did it.  Present your findings and interpret your results.

As usual, transfer your output to Word ^® and format it in an appropriate manner.  Conclude your report with a couple of paragraphs or so of analysis, in which you provide an overall interpretation of your data analysis and draw conclusions of appropriate scope about its value and applicability.

Price: $30.96

Solution: The downloadable solution consists of 13 pages, 1796 words and 12 charts.
Deliverable: Word Document