Full model

Model

2LL

Parameters

df

AIC

BIC

Δχ
^{2}

Δdf

p

Saturated

25,044.31

70

9494

6056.31

−49,743.70

–

–

–

ADEs

25,155.82

28

9536

6083.82

−49,963.04

111.51

42

<.001

Nested models

Model

2LL

Parameters

df

AIC

BIC

Δχ
^{2}

Δdf

p

ADE scalar

25,160.31

25

9539

6082.31

−49,982.19

4.84

3

.21

ADE hom.

25,220.50

15

9549

6122.50

−50,000.77

64.68

13

<.001

AE scalar

25,165.81

20

9544

6077.81

−50,016.07

9.99

8

.27

AE scalar

25,208.87

14

9550

6108.87

−50,020.27

53.05

14

<.001

E scalar

26,053.90

9

9555

6943.90

−49,214.63

898.08

19

<.001

 Statistics for the ADEs model are in comparison with the saturated model. Statistics for all other models are in comparison to the ADEs model. Saturated model: model of the observed data, allowing means and variances to differ across twin order and zygosity. All other models constrain means and variances to be equal across twin order and zygosity

2LL fit statistic, which is 2*loglikelihood of the data, df degrees of freedom, AIC Akaike’s Information Criteria, BIC Bayesian Information Criteria, Δχ
^{2} change in 2LL between two models, distributed, χ^{2}, Δdf change in degrees of freedom between two models, A additive genetic influence, D nonadditive genetic influence, E nonshared environmental influence, s sibling interaction effects, hom. homogeneity model, which collapsed all estimates across sex