Appendix B: Detailed Methology AARP Multicultural Survey
giving the care group into a positive and a negative dimension. The negative one, IMPACT2, consists of the attitudes that caring for older people pushes one farther apart from them (Q74) and makes one resentful of the time commitment to them (Q76). The positive factor, IMPACT3, is made up of the two positive attitudes towards the elderly (Q75 and Q77) that come from caring for them, making one feel closer to them and making one feel more optimistic about one’s own old age. The two questions were combined for each factor to create two small indices.
RELATIONSHIP STRESS is the term applied to the factor that indicates the amount of stress generated between the respondent and various relatives and friends due to the effort of caring for older people (Q78-Q83). These questions grouped together in the factor analysis and one index was computed from the responses to the questions.
The resources that one might use to help cope with caring for older family members divide into three dimensions based on the factor analysis. The SPIRITUAL dimension combines questions Q84 and Q85, which deal with faith or prayer and religious institutions as resources for helping one care for older people. The second factor, PROFESSIONAL/INSTITUTIONAL RESOURCES, consists of seeking help from social workers, doctors, government agencies, community and organizations that assist older people, and attorneys and accountants (Q86-Q91). The final factor that clusters together (Q92-Q97) consists of FAMILY and FRIENDS. Scales were created for each of these factors.
The cumulative result of the factor analyses was the creation of 13 indices, representing 69 questions. These scales were then used in a series of analyses to identify which independent variables were more likely associated with the scales.
Multiple Linear Regression
Multiple linear regression investigates the extent to which all independent variables (e.g., age, income, race, sex) considered together influence the dependent variable(s) (the various attitudes under investigation). Furthermore, regression analyses may indicate which of the independent variables have the greater affect on the various attitudes and opinions being examined. Regression also helps determine whether a statistically significant influence of an independent variable (such as race) on a dependent variable (such as stress from caring for people) indicates a real or superficial relationship by controlling for other independent variables (such as income and education).
For this study, multiple linear regression analyses were run against the scales created from the factor analyses. The independent variables for the regressions were region, age (arrayed exactly from 45 to 55), gender, urbanity, income (missing cases replaced with the mean of ten nearby cases), availability of flex time at work, race (as dichotomous variables for each of the four race/ethnic groups under consideration), education, and place of birth (foreign or not). These results were used to guide analysts to the appropriate bivariate tables. Also, Oneway Analyses of Variance (ANOVA) were performed on the means of the scales for race and gender to better exam the relationships among these groups. Based upon these results, regressions on the 13 scales were run on each of the racial/ethnic samples where race was removed as an independent