Skip to main content

Table 5 Random effects and model diagnostics from hierarchical linear models for continuous birth weight in BC, Canada

From: The reduction of birth weight by fine particulate matter and its modification by maternal and neighbourhood-level factors: a multilevel analysis in British Columbia, Canada

Random effects & model diagnostics

Null model

Null + r.slope

Model-1 (level-1)

Model-2 (SES)

Model-3 (PM2.5)

Model-4 (PM2.5 interact)

Variance components

      

 L1 residual (sd)

560.5

442.9

435.2

435.3

435.3

435.1

 L2 intercept (sd)

78.8

67.6

67.0

45.9

39.7

37.9

 L2 slope (sd)

--

30.1

26.6

27.0

27.2

27.4

Intercept

3434.2

3448.4

3524.2

3521.3

3523.4

3522.6

AIC

597007

489148

480867

479403

478997

478753

L1-PCV (%)

Ref

37.5

39.7

39.7

39.7

39.7

L2-PCV (%)

Ref

26.4

27.7

66.0

74.5

76.8

ICC/VPCa

0.019

0.023a

0.023a

0.011a

0.008a

0.008a

L1 Moran’s Ib

0.122

0.108

0.105

0.043

0.022

0.018

L2ri Moran’s Ib

0.300

0.301

0.310

0.113

0.079

0.071

L2rs Moran’s Ib

--

0.018

0.018

0.018

0.017

0.017

  1. L1: level-1 = individual-level; L2: level-2 = DA-level; sd: standard deviation; AIC: Akaike Information Criterion; PCV: proportional change in variance; aICC: Intra-class correlation – is called the VPC (variance partition coefficient) when conditional on the random-slope variable, thus values in table represent intercepts for individuals with mean gestational age (~39 weeks); L2ri: level-2 random intercept; L2rs: level-2 random slope; ball results were significant p < 0.05 with 999 permutations using a queen criterion spatial weight matrix