The Concept of Self: A Study of Black Identity and Self-Esteem

Appendixes

APPENDIXES APPENDIX A The Cross-SectionalStudy Figure A.l graphically depicts the model I proposed.' Employing structural equation modeling techniques (Joreskog & Sorbom, 1989), I arrived at overall fit measures which suggest that this model fits the data weL2Table A.l shows that the X 2 is 13.02, with 6 df. This value was statistically significant at the p = 0.04 level. Since X 2 is influenced by sample size (and I have a large sample), I gave more weight to other overall fit measures and the component fit measure. All of the other overall fit measures also suggest a good fit. For example, the goodness-of-fit index (GFI) was 1.00, the adjusted goodness-of-fit index (AGFI) was 0.99, and the comparative fit index (CFI) was 1.00. Looking at the factor loadings, I found that they were statisticallysignificant , fairly uniform in value (each of a substantial magnitude), and, thus, seemed to be reflective of their respective constructs (see Table A.2). Since I obtained a good fitting model at the overall level and component fit level, I was able to move to a further investigation of this model. As a first step, I examined the means for both self-esteem constructs to see whether the means were different across these two factors. I was addressing the question whether African Americans as a group have low self-esteem. There is the question ofwhat constitutes high or low self-esteem. I took the straightforward position that (using a 4-point scale) if the mean was above X2 X3 X4 X5 ' 6 FigureA.1. Conceptual Representation of a Model of Global Self-Esteem 181 APPENDIX A Table A.1. Goodness-of-Fitfor the Self-Esteem Model Summary Measures Estimates Goodness-of-Fit Index (GFI) Adjusted Goodness-of-Fit Index (AGFI) Norrned Fit Index Non-Norrned Fit Index Comparative Fit Index (CFI) Source: National Study of Black Americans (1979-80). Table A.2. Completely Standardized Coefficients of the Self-Esteem Model Parameters Estimates Source: National Study of Black Americans (1979-80). 2, then this would be indicative of high self-esteem. Any value lower than 2 would constitute low self-esteem. Keeping in mind that the range of the two self-esteem scales is 1 = "almostalways true" to 4 = "never true," I found the mean to be 3.59 for the positively worded scale and 3.48 for the negativelyworded scale. With the items coded so that a higher score indicated greater self-esteem, the findings indicated that Africans, as a collective, have above average self-esteem or self-worth. The reliabilities (as internal consistencies) of these scales were 0.64 for self-esteem, positively worded, and 0.68 for self-esteem, negatively worded. APPENDIX Tests Across the Social Structural Variables Education Education and Positivel. Worded Self-Esteem Model 1shows that X 2 is statistically nonsignificant (x2 = 9.04, with 6 degrees of freedom, p = 0.17) and the other measure of fit, the comparative fit index (CFI), is 1.00. Less promising findings exist for models 2 and 3, but both show a statistically significant x2,with statistically significant incremental x2s.The CFIs for both models, however, indicate an adequate fit. Education and Negativel. Worded Self-Esteem For all of the models, the X 2 is statistically significant, suggesting a poor fit. Moreover, only the CFI for model 1is considered high. A l lthe incremental x2swere large and significant. Overall, with the possible exception of model 1,there appeared to be a noticeable difference between model 2 and model 3 across education levels. The items had different reliabilities and explained a differential amount of variance across education levels. A cautious stance toward these findings would be to saythat the negatively worded self-esteem items had the same meaning across education groups, but do not seem to have equal reliabilities across these groups. (See Table B.1.) Age and Positivelg Worded Self-Esteem The hypothesis of equal factor loadings cannot be rejected based on the chisquare measure (x2/df (4) = 4.46, p = 0.35). Further, the CFI (1.00) also suggests an excellent fit. Therefore, it appeared that my proposed model fits the data across the different age groups and that the items in the positive self-esteem scale have the same meaning. Moving farther up the hierarchy, the picture became unclear. Model 2 poses the question of equal reliability. When I look at model 2...