Table 2.
Approaches | Chi-Square (df) | RMSEA | CFI | TLI | SRMR |
---|---|---|---|---|---|
Approach 1: Total score matrix | 437.161 (24) | .10 | .95 | .90 | .06 |
Approach 2: Within-student matrix | 346.517 (24) | .09 | .96 | .92 | .05 |
Approach 3: Between-classroom matrix | 95.708 (24) | .15 | .94 | .89 | .08 |
Approach 4: Multilevel-CFA model | 494.396 (48) | .08 | .95 | .91 | .05 (Within) .12 (Between) |
Approach 41: Multilevel-CFA model saturated at the within level. | 9.703(24) | .04 | .99 | .97 | .003 (Within) .01 (Between) |
Approach 4: Multilevel-CFA model saturated at the between level. | 423.205(24) | .10 | .96 | .83 | .05 (Within) .02 (Between) |
Note. Degrees of freedom appear in parentheses. RMSEA = root mean square error of approximation, CFI = comparative fit index, TLI = Tucker-Lewis Index; and SRMR = standardized root mean square residual.
The last two models in Table 2 fit Model 4 with a saturated model at the student or classroom level in order to isolate the lack of fit at the alternate level (Ryu & West, 2009). These models and their fit to the data are discussed later in the text.