Table 3.
Full and best-fitting bivariate Cholesky decomposition fit statistics for associations between objective and subjective sleep indicators
| Variables | Model | −2LL | df | AIC | ∆df | ∆χ 2 | P | rG | rC | rE |
|---|---|---|---|---|---|---|---|---|---|---|
| Objective sleep duration and sleep efficiency | ACE-ACE-ACE | 3,335.25 | 909 | 1,517.25 | – | – | – | 0.79 | 0.00 | 0.00 |
| AE-AE-ACE | 3,331.78 | 911 | 1,509.78 | 2 | 7.47 | 0.99 | 0.85 | – | 0.81 | |
| Objective sleep duration and SOL | ACE-ACE-ACE | 3,279.07 | 909 | 1,461.07 | – | – | – | 0.94 | 1.00 | 0.26 |
| AE-AE-ACE | 3,280.50 | 911 | 1,458.50 | 2 | 1.43 | 0.49 | 0.97 | – | 0.27 | |
| Objective sleep duration and sleep midpoint time variability | ACE-ACE-ACE | 839.09 | 909 | −978.91 | – | – | – | 0.00 | 0.61 | 0.08 |
| ACE-CE-CE | 840.53 | 911 | −981.47 | 2 | 1.44 | 0.49 | – | 0.58 | 0.07 | |
| Objective sleep midpoint time and sleep midpoint time variability | ACE-ACE-ACE | 437.05 | 908 | −1,378.95 | – | – | – | 1.00 | 0.35 | 0.28 |
| CE-CE-CE | 440.52 | 911 | −1,381.48 | 3 | 3.47 | 0.33 | – | 0.37 | 0.41 | |
| Objective sleep duration and parent- reported sleep duration | ACE-ACE-ACE | 1,973.29 | 1,019 | −64.71 | – | – | – | 0.02 | 0.89 | 0.09 |
| ACE-CE-ACE | 1,973.29 | 1,020 | −66.71 | 1 | 0.00 | 0.94 | – | 1.00 | 0.16 | |
| Objective sleep midpoint time and parent-reported sleep duration | ACE-ACE-ACE | 1,667.91 | 1,018 | −368.09 | – | – | – | 0.00 | 0.00 | 0.07 |
| ACE-E-ACE | 1,667.91 | 1,020 | −372.09 | 2 | 0.00 | 0.99 | – | – | 0.07 | |
| Objective sleep midpoint time variability and parent-reported sleep duration | ACE-ACE-ACE | 1,070.58 | 1,019 | −967.42 | – | – | – | 0.00 | 0.00 | 0.00 |
| ACE-E-ACE | 1,070.58 | 1,021 | −971.42 | 2 | 0.00 | 0.99 | – | – | 0.14 |
Sex and age were regressed out of variables prior to conducting models. Bolded models denote the best-fitting models. Models (e.g. ACE-ACE-ACE) are denoted to represent the genetic and environmental variance paths (retained) in the first phenotype, followed by the covariance paths shared (retained) between the first and second phenotype, and finally the genetic and environmental variance paths (retained) in the second phenotype after accounting for variance in the first phenotype. The −2LL is the −2 log likelihood model fit, and the AIC is the Akaike’s Information Criterion, which is an additional measure of model fit. ∆df shows the change in the degrees of freedom. ∆χ2 is the change in chi-squared values when dropping model parameters. p denotes the p-value level of significance for the chi-squared test. rG, genetic correlation between first and second phenotype in each model. rC, shared environmental correlation between first and second phenotype in each model. rE, nonshared environmental correlation between first and second phenotype in each model.