Multigroup Ethnic Identity

Implica�ons of Ethnic Iden�ty Explora�on and Ethnic Iden�ty Affirma�on and Belonging for Intergroup A�tudes Among Adolescents Results of the Factor Analysis of the Two Scales of the Multigroup Ethnic Identity Measure Multi-group Factor Analysis of Exploration and Affirmation and Belonging Confirmatory factor analyses of the ethnic identity exploration items (at time 1) and the ethnic identity affirmation and belonging items (at time 2) were performed on each of the three ethnic groups separately. These constructs were tested separately (prior to testing the hypothesized and alternative models) in order to establish that the factor structure obtained from previous exploratory methods (e.g. Phinney et al., 1997) not only fit the data for each of the three sub-samples but that the three groups were relatively equivalent in the structure of these two constructs. These CFA's use assessments that are keyed to the relationships specified in our hypothesized model. The factor structure consisted of the five exploration items loading together on an exploration factor, while the four affirmation and belonging items loaded together on a separate affirmation and belonging factor. For all three groups the factors were allowed to correlate. Each of the three groups required a minimum number of correlated errors to be added to the model in order for the model fit to reach acceptable levels. After model fit was reached for each individual group, the groups were than compared to each other in three complete invariance multi-group models. Complete invariance means that the groups are compared on every aspect of the factor structures, loadings and variances (e.g. error variances, factor variances). For simplicity, the three groups were compared pair-wise, and in each paired multi-sample analysis the loadings and variances were constrained to be equal between the two groups. When using this strategy, the two groups can be considered equal if, after the models are constrained to be equal, the multi-group model still fits. This procedure was followed three times, in order to compare each group to each of the other groups. The results of this procedure are presented in the sections that follow. Confirmatory factor analysis in the Euro-American sub-sample. The structure of the two ethnic identity factors, exploration and affirmation and belonging, fit well for the Euro-American sub-sample, with χ2(26) = 35.09, p = .110, CFI = .982 and RMSEA = .051. Even though the hypothesized factor model fit the data the Lagrange multiplier test indicated two large correlated errors. The paths appear superfluous in this factor but serve to dramatically improve the fit of the full hypothesized model (shown in Figure 4 below). The first correlated error was between two of the exploration items (items 2 and 3) and the addition of this path significantly improved the chi-square, χ2(1) = 9.881, p < .05. The addition of a correlation between errors associated with the second and third items of the affirmation and belonging subscale was also a significant improvement, χ2(1) = 6.713, p < .05. The final model fit was nearly perfect for the EuroAmerican sub-sample, with χ2(24) = 19.108, p = .746, CFI = 1.000 and RMSEA < .001. For the Euro-American sub-sample, all of the loadings were significant for both factors and the correlation between the two factors was significant and moderately large (r = .352). The exploration factor accounted for a range of 12% to 69% of the variability in the exploration items and the affirmation and belonging factor accounted for a range of 68% to 72% of the variability in the affirmation and belonging items. Confirmatory factor analysis in the Asian American sub-sample. As hypothesized, the exploration and affirmation and belonging items only marginally fit the data for the Asian American sub-sample, with χ2(26) = 75.138, p < .001, CFI = .908 and RMSEA = .115. The Lagrange multiplier test suggested the inclusion of a correlated error between items one and three of the affirmation and belonging subscale and this was a significant improvement, χ2(1) = 13.625, p < .05. A correlation between the error terms for items one and five of the exploration subscale was also included, χ2(1) = 9.560, p < .05. In addition, error terms for items three and five of the exploration subscale were also correlated, χ2(1) = 9.286, p < .05. Finally, the addition of a correlation between the error terms associated with items two and four of the affirmation and belonging subscale significantly improved the model, χ2(1) = 8.479, p < .05. The final model for the Asian American sub-sample fit well, χ2(22) = 34.188, p = .047, CFI = .977 and RMSEA = .062. For the Asian American sub-sample, all of the loadings were significant for both factors and the correlation between the two factors was significant and large (r = .698). The exploration factor accounted for a range of 15% to 56% of the variability in the exploration items and the affirmation and belonging factor accounted for a range of 55% to 71% of the variability in the affirmation and belonging items. Confirmatory factor analysis in the Latino sub-sample. For the Latino sub-sample, the hypothesized model fit the data moderately well, with χ2(26) = 39.852, p = .040, CFI = .925 and RMSEA = .071. The addition of a correlated error between the second and third items on the exploration subscale significantly improved the model, χ2(1) = 8.196, p < .05, and the resultant model fit well, with χ2(25) = 31.656, p = .168, CFI = .964 and RMSEA = .050. For the Latino American sub-sample, all of the loadings were significant for the affirmation and belonging items, however two of the loadings on the exploration factor were not significant, namely the second and third items on the exploration subscale. The correlation between the two factors was moderate (r = .367) and significant. The exploration factor accounted for a range of 4% to 40% of the variability in the exploration items and the affirmation and belonging factor accounted for a range of 32% to 66% of the variability in the affirmation and belonging items. Complete invariance multi-group confirmatory factor analysis: Euro-Americans and Asian Americans. When the Euro-American and Asian American sub-samples were constrained in a complete invariance multi-group analysis, the model fit the data for the groups moderately well, with χ2(65) = 113.027, p < .001, CFI = .955 and RMSEA = .051. The Lagrange multiplier test suggested that the two groups should be allowed to differ in terms of the correlation between the two factors and releasing this constraint significantly improved the model fit, χ2(1) = 13.105, p < .05. After accounting for this difference between the groups, the multi-group model fit the data well, with χ2(64) = 99.922, p = .002, CFI = .966 and RMSEA = .045. Complete invariance multi-group confirmatory factor analysis: Asian Americans and Latinos. The multi-group comparison of the Asian American and Latino sub-samples showed that the two groups were very similar and the model fit was very good, with χ2(66) = 95.511, p = .010, CFI = .959 and RMSEA = .042. This degree of model fit shows that, even though the factor loadings of two of the ethnic identity exploration items in the Latino sub-sample were not significant, those loadings were not significantly different from the loadings found in the Asian American sample. Complete invariance multi-group confirmatory factor analysis: Latinos and EuroAmericans. When the Latino and Euro-American sub-samples were compared in a complete invariance multi-group analysis, there was an additional similarity between the groups, namely the correlation between the error terms of the second and third items of the exploration factor was significant for both groups. Since the two groups shared this extra parameter, it was also constrained between the groups. The resultant multi-group model fit the groups fairly well, with χ2(69) = 106.101, p = .003, CFI = .947 and RMSEA = .047. The Lagrange multiplier test suggested that the two groups should be allowed to differ in terms of the error term associated with the second affirmation and belonging item, and releasing this constraint significantly improved the model fit, χ2(1) = 21.177, p < .05, After accounting for this difference between the groups, the multi-group model fit the data well, χ2(68) = 84.924, p = .080, CFI = .976 and RMSEA = .032. As in the case of the multi-group model for Asian Americans and Latinos, this shows that even though the factor loadings of two of the ethnic identity exploration items in the Latino sub-sample were not significant, these loadings were not significantly different from the loadings of the same items in the Euro-American sub-sample. Correlation Tables for Each Ethnic Group Table 1: Correlation matrix with means and standard deviations in Model 1 for Euro-Americans (n = 137) Exploration Item 1 Item 2 Item 3 Affirmation & Belonging Item 4 Item 5 Item 1 Item 2 Item 3 Item 4 Item 1 1.000 Item 2 0.252 1.000 Item 3 0.201 0.352 1.000 Item 4 0.543 0.251 0.269 1.000 Item 5 0.350 0.294 0.198 0.518 1.000 Item 1 0.224 0.069 0.272 0.244 0.155 1.000 Item 2 0.200 0.074 0.201 0.249 0.247 0.771 1.000 Item 3 0.185 0.050 0.190 0.222 0.225 0.727 0.651 1.000 Item 4 0.209 0.076 0.172 0.267 0.251 0.737 0.787 0.691 1.000 Asian 0.211 -0.203 -0.116 0.274 0.102 0.183 0.144 0.154 0.096 Latinos 0.009 -0.225 -0.121 0.132 -0.005 0.100 0.117 0.202 0.091 Euro 0.062 -0.107 0.035 0.165 0.017 0.326 0.297 0.379 0.272 Standard Dev. 1.750 1.852 1.623 1.602 1.864 1.390 1.473 1.376 1.460 Mean 4.336 3.374 3.464 4.245 4.659 5.491 5.194 5.536 5.145 Exploration Affirmation & Belonging Warmth * Ingroup warmth in boldface Table 2: Correlation matrix with means and standard deviations in Model 1 for Asian Americans (n = 144) Exploration Affirmation & Belonging Item 1 Item 2 Item 3 Item 4 Item 5 Item 1 Item 2 Item 3 Item 4 Item 1 1.000 Item 2 0.299 1.000 Item 3 0.267 0.183 1.000 Item 4 0.458 0.191 0.333 1.000 Item 5 0.263 0.316 0.053 0.382 1.000 Item 1 0.408 0.303 0.143 0.247 0.470 1.000 Item 2 0.388 0.450 0.205 0.228 0.399 0.661 1.000 Item 3 0.339 0.280 0.092 0.255 0.427 0.821 0.589 1.000 Item 4 0.469 0.427 0.149 0.299 0.420 0.688 0.628 0.717 1.000 Asian 0.173 0.306 0.111 0.252 0.273 0.497 0.404 0.430 0.430 Latinos 0.073 0.017 0.030 0.244 0.210 0.328 0.183 0.244 0.155 Euro 0.169 0.053 0.037 0.182 0.197 0.237 0.158 0.194 0.124 Standard Dev. 1.487 1.598 1.427 1.660 1.510 1.298 1.433 1.469 1.439 Mean 4.660 4.009 3.860 4.293 5.371 5.578 5.208 5.391 5.231 Exploration Affirmation & Belonging Warmth * Ingroup warmth in boldface Table 3: Correlation matrix with means and standard deviations in Model 1 for Latinos (n = 108) Exploration Item 1 Item 2 Item 3 Affirmation & Belonging Item 4 Item 5 Item 1 Item 2 Item 3 Item 4 Item 1 1.000 Item 2 0.071 1.000 Item 3 0.081 0.315 1.000 Item 4 0.385 0.175 0.197 1.000 Item 5 0.341 0.172 0.170 0.233 1.000 Item 1 0.218 0.085 0.047 0.134 0.304 1.000 Item 2 0.223 -0.031 0.143 0.020 0.145 0.358 1.000 Item 3 0.094 -0.073 0.063 -0.051 0.249 0.561 0.496 1.000 Item 4 0.203 -0.090 -0.005 0.046 0.312 0.545 0.396 0.621 1.000 Asian 0.165 -0.267 -0.117 0.016 0.104 0.325 0.097 0.185 0.331 Latinos 0.208 -0.056 -0.014 -0.021 0.168 0.596 0.280 0.340 0.497 Euro 0.103 -0.245 -0.092 -0.012 -0.041 0.238 0.025 0.104 0.186 Standard Dev. 1.513 1.541 1.461 1.491 1.545 1.202 1.347 1.261 1.350 Mean 4.361 3.343 3.815 4.298 5.460 5.889 5.358 5.917 5.526 Exploration Affirmation & Belonging Warmth * Ingroup warmth in boldface Table 4: Correlation matrix with means and standard deviations in Model 2 for Euro-Americans (n = 137) Exploration Item 1 Exploration Item 1 1.000 Item 2 Item 3 Affirmation & Belonging Item 4 Item 5 Item 1 Item 2 Item 3 Item 4 Item 2 0.443 1.000 Item 3 0.426 0.515 1.000 Item 4 0.603 0.246 0.395 1.000 Item 5 0.530 0.408 0.298 0.404 1.000 Item 1 0.199 0.347 0.334 0.210 0.285 1.000 Item 2 0.182 0.264 0.340 0.224 0.242 0.670 1.000 Item 3 0.242 0.235 0.305 0.430 0.440 0.438 0.319 1.000 Item 4 0.292 0.292 0.302 0.355 0.375 0.464 0.471 0.565 1.000 Asian 0.345 0.141 0.147 0.142 0.261 -0.033 -0.030 0.137 0.128 Latinos 0.309 0.027 0.084 0.198 0.119 -0.005 -0.054 0.179 0.109 Euro 0.263 0.139 0.145 0.158 0.222 0.132 0.079 0.246 0.248 Standard Dev. 1.674 1.741 1.592 1.642 1.673 1.502 1.566 1.278 1.477 Mean 4.648 3.912 3.840 4.656 4.873 5.380 4.808 5.482 4.887 Affirmation & Belonging Warmth * Ingroup warmth in boldface Table 5: Correlation matrix with means and standard deviations in Model 2 for Asian Americans (n = 144) Exploration Item 1 Item 2 Item 3 Affirmation & Belonging Item 4 Item 5 Item 1 Item 2 Item 3 Item 4 Item 1 1.000 Item 2 0.428 1.000 Item 3 0.406 0.505 1.000 Item 4 0.596 0.260 0.371 1.000 Item 5 0.484 0.469 0.322 0.353 1.000 Item 1 0.277 0.285 0.160 0.278 0.361 1.000 Item 2 0.342 0.279 0.221 0.318 0.251 0.609 1.000 Item 3 0.389 0.378 0.367 0.366 0.462 0.573 0.465 1.000 Item 4 0.313 0.373 0.302 0.427 0.374 0.536 0.608 0.646 1.000 Asian 0.242 0.261 0.266 0.261 0.249 0.289 0.276 0.295 0.453 Latinos 0.128 -0.052 0.155 0.230 0.045 0.128 0.063 0.137 0.202 Euro 0.196 0.062 0.141 0.278 0.169 0.158 0.116 0.061 0.256 Standard Dev. 1.435 1.669 1.520 1.598 1.413 1.408 1.525 1.472 1.621 Mean 4.938 4.271 4.194 4.688 5.467 5.502 5.021 5.382 5.048 Exploration Affirmation & Belonging Warmth * Ingroup warmth in boldface Table 6: Correlation matrix with means and standard deviations in Model 2 for Latinos (n = 108) Exploration Item 1 Exploration Item 2 Item 1 1.000 Item 2 0.281 1.000 Item 3 0.407 0.367 Item 3 1.000 Affirmation & Belonging Item 4 Item 5 Item 1 Item 2 Item 3 Item 4 Item 4 0.359 0.240 0.301 1.000 Item 5 0.291 0.291 0.186 0.360 1.000 Item 1 0.118 0.072 0.223 0.225 0.337 1.000 Item 2 0.037 0.218 0.369 0.117 0.192 0.383 1.000 Item 3 0.309 0.144 0.041 0.117 0.405 0.272 0.232 1.000 Item 4 0.280 0.213 0.258 0.245 0.411 0.220 0.328 0.646 1.000 Asian 0.291 0.038 0.146 0.182 0.254 0.099 -0.0003 0.210 0.174 Latinos 0.261 0.019 0.041 0.140 0.439 0.277 -0.040 0.200 0.234 Euro 0.270 -0.113 0.090 0.123 0.174 0.107 -0.231 0.026 0.047 Standard Dev. 1.389 1.483 1.400 1.403 1.435 1.426 1.378 1.091 1.192 Mean 4.657 4.023 4.180 4.778 5.294 5.774 5.005 6.028 5.681 Affirmation & Belonging Warmth * Ingroup warmth in boldface Items from the Multigroup Ethnic Identity Measure Ethnic Identity Exploration Items 1. I have spent time trying to find out more about my ethnic group, such as its history, traditions, and customs 2. I am active in organizations or social groups that include mostly members of my own ethnic group 3. I think a lot about how my life will be affected by my ethnic group membership 4. In order to learn more about my ethnic background, I have often talked to other people about my ethnic group 5. I participate in cultural practices of my own group, such as special food, music, or customs Ethnic Identity Affirmation and Belonging Items 1. I am happy that I am a member of the ethnic group I belong to 2. I have a strong sense of belonging to my own ethnic group 3. I have a lot of pride in my ethnic group 4. I feel a strong attachment towards my own ethnic group