. Author manuscript; available in PMC: 2022 Jun 9.

Published in final edited form as: J Educ Psychol. 2021 Oct 11;114(2):215–238. doi: 10.1037/edu0000564

Table 5.

Model Fit Comparisons

Dimension of Written Composition	Figure	χ² (df), p value	CFI (TLI)	RMSEA (SRMR)	nBIC	Model comparison : Δχ² (Δdf, p value)
Writing Quality	Figure 4a	382.41 (200), < .001	.96 (.95)	.05 (.05)	45207.28
	Figure 4b ^*	367.99 (199), < .001	.96 (.95)	.05 (.05)	45195.55
	Figure 4c	367.99 (199), < .001	.96 (.95)	.05 (.05)	45195.55	4a vs. 4c: 12.42 (1, < .001)
Writing Productivity	Figure 4a	354.45 (200), < .001	.96 (.96)	.05 (.05)	50008.26
	Figure 4b	347.98 (199), < .001	.97 (.96)	.05 (.05)	50004.47
	Figure 4c ⁺	347.98 (199), < .001	.97 (.96)	.05 (.05)	50004.47	4a vs. 4b & 4c: 6.47 (1, .01)
Correctness in Writing	Figure 4a	392.70 (200), < .001	.96 (.95)	.05 (.05)	49673.95
	Figure 4b	380.95 (199), < .001	.96 (.95)	.05 (.05)	49664.88
	Figure 4c ⁺	380.95 (199), < .001	.96 (.95)	.05 (.05)	49664.88	4a vs. 4b & 4c: 9.75 (1, .002)

Note.

Heywood case

⁺

Statistically significant suppression effects; Bolded are final models.