Skip to main content
. Author manuscript; available in PMC: 2023 Apr 1.
Published in final edited form as: Pediatr Obes. 2021 Nov 2;17(4):e12866. doi: 10.1111/ijpo.12866

TABLE 3.

Nested model comparisons. N = 294.

Model Free parameters CFI TLI RMSEA [90% CI] SRMR χ2 df p Δχ2 Δdf Δp
Inhibitory control (Go/No-Go task)
CLPM 44 0.91 0.77 0.12 [0.10, 0.15] 0.06 84.04 16 <0.01 – – –
vs. RI-CLPM 47 0.99 0.96 0.05 [0.00, 0.08] 0.05 21.08 13 0.07 27.89 3 <0.01
RI-CLPM
vs. RI-CLPM with autoregressive equality constraints 43 0.96 0.90 0.08 [0.05, 0.10] 0.07 47.59 17 <0.01 24.96 4 <0.01
vs. RI-CLPM with cross-lagged equality constraints† 43 0.99 0.98 0.04 [0.00, 0.07] 0.05 23.26 17 0.14 3.39 4 0.50
Working memory (Nebraska Barnyard task)
CLPM 44 0.88 0.70 0.15 [0.13, 0.18] 0.09 126.12 16 <0.01 — — —
vs. RI-CLPM 47 0.98 0.93 0.07 [0.04, 0.11] 0.06 39.67 13 <0.01 48.70 3 <0.01
RI-CLPM
vs. RI-CLPM with lag-2 autoregressive paths† 51 0.98 0.90 0.08 [0.05, 0.12] 0.05 27.86 9 <0.01 11.01 4 0.03
RI-CLPM with lag-2 autoregressive paths
vs. RI-CLPM with lag-2 autoregressive paths and autoregressive equality constraints 43 0.97 0.92 0.08 [0.05, 0.10] 0.06 46.62 17 <0.01 19.38 8 <0.01
vs. RI-CLPM with lag-2 autoregressive paths and cross-lagged equality constraints 47 0.98 0.90 0.08 [0.05, 0.11] 0.06 38.09 13 <0.01 10.34 4 0.04
Flexible shifting (Shape School task)
CLPM 44 0.92 0.80 0.11 [0.09, 0.14] 0.06 77.44 16 <0.01 — — —
vs. RI-CLPM† 47 0.99 0.98 0.03 [0.00, 0.07] 0.04 17.28 13 0.19 16.13 3 <0.01
RI-CLPM
vs. RI-CLPM with autoregressive equality constraints 43 0.99 0.97 0.04 [0.00, 0.07] 0.06 29.38 17 0.03 10.42 4 0.03
vs. RI-CLPM with cross-lagged equality constraints 43 0.93 0.90 0.07 [0.04, 0.11] 0.07 32.09 17 0.01 14.51 4 <0.01
†

Indicates the best-fitting, most parsimonious model. Lag-2 autoregressive paths describe non-consecutive autoregressive paths (i.e., from grades 1 to 3 and grades 2 to 4). Fit statistics include the χ2 statistic (recommended to be nonsignificant); the comparative fit index (CFI) and Tucker-Lewis Index (TLI; recommended to be 0.90 or higher)39; and the root mean squared error of approximation (RMSEA) and the standard root mean residual (SRMR, recommended to be 0.05 or lower).40 χ2 difference tests (Δχ2) were conducted using the Satorra–Bentler method.

Abbreviations: CI, confidence interval; CLPM, cross-lagged panel model; df, degrees of freedom; RI-CLPM, random intercept cross-lagged panel model.