1. Conduct the analyses outlined in this chapter. The data are in the file labeled “kabc cfamatrices.xls.” The first two worksheets include the matrices for males and females separately. The third worksheet contains the matrix with sex as a part of the matrix (for the MIMIC model). See the website (www.tzkeith.com) for initial setup for these models for Amos and Mplus.
2. Conduct the higher-order invariance tests as outlined, but not detailed, in this chapter.
Make sure your degrees of freedom match those shown in Table 20.8 below for each model:
3. Test the invariance of shorter self-concept and locus of control scales from NELS across sex. The proper composition of these scales is represented in Figure 20.18. A reminder of the item wording is shown in Table 20.9. The data are in the excel file “sc loc matrix 2.xls” with a separate tab for boys and girls (there is also a tab with the full matrix). Note that the four self-concept items have been reversed so that for all items a high score represents a positive self-concept or an internal locus of control. Note also that the figure shown below represents a “conceptual model” only; it does not include errors
or other crucial model details. (Your model should include those details, however!)
a) Test this initial model across groups (configural invariance). You do not need to estimate means & intercepts at this step (but I recommend that you do so).
b) Add a covariance between the errors of measurement for items BYS44Dr and BYS44Er.
Figure 20.18
Conceptual model for Exercise 3: invariance of self-concept and locus of control for boys
and girls.
Table 20.9
Self-Concept and Locus of Control Items for Exercise 3. Each item’s response choices
ranged from 1, strongly agree, to 4, strongly disagree. As shown, positively worded items were reversed
so that for all items a high score represents a high self-concept or a high (internal) locus of control.
c) Test metric invariance across groups.
d) Test intercept invariance (aka scalar or strong invariance; you
do
need to estimate means & intercepts at this step).
e) Based on the text output, does it appear that boys and girls differ in their overall latent mean level of self-concept? Locus of control? On what did you base this conclusion? If you concluded that there were differences, which sex scored higher, and by how many points?
f) Test the equivalence of the latent means for locus of control using a model constraint and fit statistics. Do boys and girls differ in their levels of locus of control? Why do you come to that conclusion?
g) Test the equivalence of the latent means for self-concept using a model constraint and fit statistics. Do boys and girls differ in their levels of self-concept? Why do you come to that conclusion? Briefly interpret your findings from questions f and g (e.g., who, if anyone, scored higher?).
h) Provide a table of fit statistics for the models liste d in steps a through g. Be sure to list the corrected RMSEA.
i) Would you be willing to accept configural invariance (step a)? Metric invariance?
Intercept invariance? Briefly explain why or why not.