. Author manuscript; available in PMC: 2020 Nov 1.

Published in final edited form as: J Educ Psychol. 2019 Mar 7;111(8):1447–1461. doi: 10.1037/edu0000355

Table 6.

Fit indexes from path models from the longitudinal sample

Model	χ²	df	parameters	p	Models	Δχ²	p	RMSEA (90% confidence interval)	CFI	NFI
			Overall Sample
M1. Autoregressive paths	45.36	9	12	.0001	---	---	---	.15 (.11, .19)	.90	.87
M2. M1 + correlated error	13.94	8	13	.0832	M2 vs. M1	31.42	.0001	.06 (.00, .12)	.98	.96
M3. M2 + att & anxiety to math	13.58	6	15	.0346	M3 vs. M2	0.36	.8353	.08 (.02, .14)	.98	.96
M4. M2 + math to att & anxiety	6.3	6	15	.3905	M4 vs. M2	7.64	.0219	.02 (.00. .10)	1.0	.98
M5. M4 + att & anxiety to math	5.94	4	17	.2037	M5 vs. M4	0.36	.8353	.05 (.00, .13)	.99	.98

		Simultaneous Multi-sample: Boys and Girls
M6. M4 simultaneous	8.02	6	21	.2368	---	---	---	.06 (.00, 16)	.99	.96

Note: att = mathematics attitudes; anxiety = mathematics anxiety for evaluations. RMSEA = root mean square error approximation (lower values mean a better fit); CFI = Comparative Fit Index (higher values mean a better fit); NFI = Bentler-Bonnet Normed Fit Index (higher values mean a better fit). The df for the Δχ² is the difference in df across the two models noted in the Models column.