1. Does Achievement have different effects on Locus of Control for gi rls versus boys? Conduct a multigroup analysis using the model shown in Figure 18.24 for girls and boys. The data (a subset of the NELS data) are in the raw data file “ach locus sex data 2.sav”. The variables should be fairly self-explanatory. The SES variable is a composite of parent education, occupational status, and family income. The Ach8 latent variable is indexed by 8th grade test scores in Reading, Math, Science, and Social Studies. The Locus10 variable is a latent variable estimated by items asking students their agreement
with the statements shown in Table 18.4. The statements espouse an external locus of control, but “strongly agree” has a value of 1 and “strongly disagree” has a value of 4, so high scores represent an internal (more healthy) locus of control.
Estimate three multi-group models:
1. A model that makes no constraints across the groups.
2. A model with all factor loadings constrained to be equal across the two groups.
3. A model in which the path from Ach8 to Locus8 is constrained to be equal across groups.
Prepare a table of fits comparing these three models. Include χ
and probability; RMSEA; CFI; SRMR; and AIC, or whatever indexes are recommended by your instructor. Evaluate the fit of the initial model. Explain what you would conclude by comparing model 1 with 2 and by comparing model 2 and model 3. Which model fit the best? What does that mean? Interpret the results from the model that you believe is most supported. What do the results mean? Explain as you might to your Uncle Gus (a real world interpretation without jargon). Extra credit: The model shown is a subset of the panel model shown in Figure 18.11. Yet the estimate of the effect of Ach 8 on Locus10 is quite different from the effect shown in Figure 18.12. Why might that be?