Problem Set 1
Due Friday, January 28, 2000
Note: You will need a calculator for this exercise. Show the calculations when you hand in the answers. In your comments, state any assumptions that you use.
1. For this question use Table 3 of the attached article, “The selling of innocence: The gestalt of danger in the lives of youth prostitutes,” by Bernard Schissel and Kari Fedec, from the Canadian Journal of Criminology, January, 1999, pp. 33-56.
a. For each of non-aboriginal and aboriginal members of the sample, test the hypothesis that there is a relationship between severity of alcohol abuse and whether or not there was involvement in prostitution. Use the chi-square test. Comment on the results and describe the relationship in words. Why might your chi-square value differ from that shown in the table?
b. Test for a difference in proportions for the following two parts of the table:
i. The difference in teen pregnancy for non-aboriginals who were involved in prostitution and those who were not.
ii. The difference in teen pregnancy for non-aboriginals not involved in prostitution and aboriginals not involved in prostitution.
Comment on the results.
c. Construct 2x2 tables for each of the two groups in part b. and test for a relationship in each case using the chi-square statistic. Using the chi-square test, also test for a relationship between sexual assault and involvement in prostitution for aboriginal members of the sample. For each of these three tables, calculate phi, the contingency coefficient, and Cramer’s V. Comment on the results and compare the first two chi-square tests with the results in part b.
d. The size of the first difference of proportion in b. and the difference in proportion for involvement in sexual assault for aboriginals not involved and involved in prostitution seem to be at odds with the reported chi-square values in the table. From the data you have calculated, attempt to explain this discrepancy.
2. The attached tables from the Fall 1998 Survey of Student Attitudes and Experiences show the political parties best reflecting the beliefs of undergraduates surveyed at the University of Regina. The original frequency distributions are provided in the first two tables. In the cross-classification table at the bottom of the page, at each of the federal and provincial level, the Reform, Saskatchewan, and Conservative party have been merged, and are labelled “Reform.” Use the data in the last table to calculate lambda, first using federal political preference to predict provincial political preference, and then using provincial political preference to predict federal political preference. Describe the relationship between the two variables and comment on the usefulness of lambda in this example.