Skip to main content

Table 3 Point estimate (and standard error) for the effect of early depressive symptoms on subsequent standardized NAPLAN numeracy scores, and point estimates for the variance components at levels 3, 2 and 1, from available case analysis (ACA) and 8 MI approaches applied to the CATS data analysis

From: Evaluation of approaches for multiple imputation of three-level data

Method

Regression coefficient estimate (SE)

Level 3 variance component

Level 2 variance component

Level 1 variance component

ACA

−0.022 (0.007)

0.042

0.239

0.232

JM-1L-DI-wide

−0.019 (0.007)

0.043

0.243

0.231

FCS-1L-DI-wide

−0.019 (0.008)

0.043

0.246

0.230

JM-2L-wide

−0.020 (0.007)

0.041

0.246

0.228

FCS-2L-wide

−0.022 (0.007)

0.042

0.245

0.229

JM-2L-DI

−0.020 (0.008)

0.042

0.237

0.228

FCS-2L-DI

–

–

–

–

FCS-3L in ml.lmer

−0.021 (0.007)

0.033

0.238

0.232

FCS-3L in Blimp

−0.021 (0.007)

0.040

0.238

0.228

  1. ACA available case analysis, DI dummy indicators, FCS fully conditional specification, JM joint modelling, NAPLAN National Assessment Program - Literacy and Numeracy, RE random effects, SE standard error