Table 4.

Multilevel logistic model for binary high/low life satisfaction, MCMC estimates (posterior SD)

Fixed effects	Model 1e	Model 2e	Model 3e	Model 4e	Model 5e
Transformed FASa	1.12 (0.07)	1.13 (0.07)	1.13 (0.07)	0.01 (0.38)	3.40 (1.09)
Square of Transformed FASa	−0.85 (0.16)	−0.85 (0.17)	−0.85 (0.17)	−0.84 (0.17)	−0.85 (0.17)
Gini coefficientb		−0.28 (0.12)		−0.08 (0.13)	−0.15 (0.14)
GDP (PPP US$)c			0.30 (0.06)	0.36 (0.12)	0.32 (0.11)
Transformed FAS*Ginib interaction				0.35 (0.12)
Transformed FAS*GDP (PPP US$)c interaction					−0.23 (0.11)
Random effects parameters
Level 1 (child) varianced	1	1	1	1	1
Level 2 (school) variance	0.120 (0.015)	0.121 (0.014)	0.120 (0.015)	0.121 (0.015)	0.120 (0.015)
Level 3 (stratum) variance	0.005 (0.004)	0.004 (0.004)	0.004 (0.004)	0.005 (0.004)	0.004 (0.004)
Level 4 (country)
Variance (intercept)	0.165 (0.043)	0.157 (0.044)	0.125 (0.034)	0.131 (0.037)	0.129 (0.037)
Variance (slope)	0.098 (0.040)	0.099 (0.041)	0.097 (0.040)	0.065 (0.031)	0.087 (0.037)
Covariance (int,slope)	−0.016 (0.030)	0.016 (0.033)	0.010 (0.027)	0.029 (0.026)	0.029 (0.030)
	46,198.7	46,195.5	46,201.7	46,196.4	46,198.8
	695.9	700.2	695.3	694.7	697.6
DIC	46,894.6	46,895.7	46,897.0	46,891.1	46,896.4

Via Markov chain Monte Carlo (MCMC); estimates are based on a chain of length of 150,000 following a burn-in of 50,000 for models 1–3 and a chain length of 300,000 following a burn-in of 200,000 for models 4 and 5

^a FAS standardised using ridit transformation and centred

^b Gini coefficient has been multiplied by 10 to make effect sizes equivalent to those of logged GDP (PPP US$)

^c Log of GDP (PPP US$)

^d Variance at the child level is constrained to 1

^e Adjusting for sex, age and family structure

^f  is the expectation of the deviance and is a measure of how well the model fits the data, is the effective number of parameters and DIC is the Deviance Information Criterion; the larger this is, the worse the model fit