. 2016 Sep 19;27(6):1615–1633. doi: 10.1177/0962280216666111

Table 1.

Comparison of fit statistics across autoregressive and simplex models.

General type	Figure #	Covariance type	b path type	Model χ²	RMSEA (90% CI)	AIC	BIC
Models without parameter restrictions^a
Autoregressive	5A	N	L	187.943 df = 20 p < 0.001	0.162 (0.141, 0.184) p < 0.001	6633.276	6753.863
Simplex	5B	N	L	115.696 df = 18 p < 0.001	0.130 (0.108, 0.153) p < 0.001	6565.030	6693.153
Autoregressive	6A	N	C	123.004 df = 20 p < 0.001	0.127 (0.106, 0.149) p < 0.001	6568.337	6688.924
Simplex	6B	N	C	87.706 df = 18 p < 0.001	0.110 (0.088, 0.134) p < 0.001	6537.039	6665.162
Models with parameter restrictions^a
Simplex	5C	N	L	117.023 df = 21 p < 0.001	0.120 (0.099, 0.141) p < 0.001	6560.356	6677.174
Simplex	6C	N	C	91.916 df = 22 p < 0.001	0.100 (0.079, 0.121) p < 0.001	6533.249	6646.299
Simplex	5D	L	L	106.322 df = 19 p < 0.001	0.120 (0.098, 0.143) p < 0.001	6553.655	6678.010
Simplex	5E	C	L	47.051 df = 20 p = 0.001	0.065 (0.041, 0.089) p = 0.14	6492.384	6612.970
Simplex	6D	L	C	82.252 df = 20 p < 0.001	0.099 (0.077, 0.121) p < 0.001	6527.585	6648.171
Simplex	6E	C	C	41.383 df = 21 p = 0.005	0.055 (0.030, 0.080) p = 0.34	6484.717	6601.534

^χ

²: chi-square model fit statistic; RMSEA: root mean square error of approximation; CI: confidence interval; AIC: Akaike’s Information Criterion; BIC: Bayesian Information Criterion; N: no measurement error covariances; C: contemporaneous; L: lagged; df: degrees of freedom.

Parameter restrictions are treatment to mediator at 52 weeks and treatment to outcome at 52 weeks = 0 and mediator to outcome b paths equal over time. Error variances are set equal for the mediator measures and the outcome measures in the simplex models (all models except 5A and 6A).