Ray Baxter of Medical Systems Associates (MSA) wanted to identify the determinants of patient satisfaction in nursing homes. Towards this end, he administered a 12 page survey to a stratified random sample of 122 patients in 16 nursing homes in Wisconsin. Patient satisfaction with their nursing home was captured by the Environmental Rating Index (ERI), a composite of patients' answers to 14 questions related to issues such as room size, physical layout, staff courtesy, medical care, cleanliness of facilities, food preparation etc. Table 17-4 shows the cross-classification results for ERI with each of three different variables: a. nursing home affiliation, b. whether or not the patient had made any new friends, and c. the number of friends made.

1. Perform the chi-square test of association for ERI with each of the three different variables (a, b and c), and test whether each association is significant.

1. Based on the chi-square statistics obtained in each case, can one figure out which of the three associations is the strongest?

Assume that all sample sizes are twice that listed in Table 17-4 e.g. the sample sizes will be 38, 82 and 122 for the first cross-classification table instead of 19, 41 and 61 respectively and so on.

434 HYPOTHESIS TESTING: BASIC CONCEPTS AND TESTS OF ASSOCIATIONS elements in the evaluation measure was represented by a sepa- The focus of the initial analysis was on the determinants rate index, based on combinations of responses to various ques- of the Environmental Rating Index (ERD). Cross-classification tions, as follows: analyses with chi-square tests of significance were run for 1. An Environmental Rating Index (ERI) was based on combinations of many variables with the ERI. Only three of answers to 14 questions involving satisfaction with such the variables showed any statistical significance. (These vari- aspects of the nursing home as room size, physical layout. ables, and the strength of the relationships, are summarized in staff courtesy, medical care, cleanliness of facilities, food Table 17.4.) preparation, and so forth. None of the other variables, such as the modernity of the 2. A Psychological Adjustment Index was based on a series home, size of the home, reasons for being in the nursing home. of attitudinal questions involving such issues as perceived selection process, or patient mobility, were found to be signifi- self-usefulness, self-perceived level of activity, perceived canily associated with the ERI. As Ray Baxter reviewed these lifestyle change, self-perceived reaction of others to results, he was wondering what conclusions he could draw, and nursing-home patients, perceived difficulty in adjusting whether other analyses would be required to examine the basic to nursing-home life (upon arrival), desire to relocate, and "product orientation" hypothesis with the ERI. so on. Question for Discussion 3. A Physical Well-Being Validity Index was based on a 1. How can you help Ray Baxter develop some implications comparison of patients' self-perceived level of health based on the analyses? with that indicated by medical records. Table 17.4 Cross-Classification Results: Environmental Rating Index versus Selected Variables' Environmental rating index Variable Low Medium High Nursing home religious affiliation Church-supported 15.8% 34.1% 45.9% Nonsectarian 84.2 65.9 54.1 100.0 100.0 100.0 Sample size (19) (41) (61) "Have you made any new friends here?" Yes 84.2 71.8 96.7 No 15.8 28.2 3.3 100.0 100.0 100.0 Sample size (18) (39) (60) Number of friends "Just a couple" 31.2 21.4 5.1 "Just a few" 31.2 14.3 20.3 "Quite a few" 37.6 64.3 74.6 100.0 100.0 100.0 Sample size (16) (28) (59) "All variables are associated with the ERI at a level of .05 or greater using the y' test of significance