Table 4 Bivariate twin model fit statistics for face orienting and face preference

Model	No. of parameters	−2LL	d.f.	AIC	Comparison model	Δχ²	Δd.f.	P value
Fully sat.	32	1,151.37	1,040	−928.63	NA	NA	NA	NA
ACE
ACE	15	1,167.82	1,057	−946.18	Fully sat.	16.45	17	0.492
ACE-nested models
AE	12	1,169.33	1,060	−950.67	ACE	1.51	3	0.680
CE	12	1,175.34	1,060	−944.66	ACE	7.52	3	0.057
E	9	1,211.82	1,063	−914.18	ACE	44.00	6	<0.001
AE-nested models
Unique path of 0	11	1,172.28	1,061	−949.72	AE	2.95	1	0.086
Shared path of 0	11	1,179.41	1,061	−942.59	AE	10.08	1	0.001

The best-fitting model was selected on the basis of non-significance (meaning that there was no decrement in fit compared with the saturated or the genetic model, indexed by the χ² distribution) and the AIC fit statistic (which incorporates information about both explained variance and parsimoniousness).
The fully sat. model is the fully saturated model of the observed data, which models the means and variances for both variables, and the phenotypic and CTCT correlations between the two variables, separately for each twin in a pair and across zygosity.
In bold: the best-fitting model was non-significant with the lowest AIC.
−2LL, fit statistic, which is minus two times the log-likelihood of the data.
d.f., degrees of freedom.
AIC, fit statistic—lower values denote better model fits.
Δχ², difference in −2LL statistic between two models, distributed χ².
Δd.f., difference in degrees of freedom between two models.

Quick links

Search