Skip to main content
. 2007 Dec;97(12):2281–2287. doi: 10.2105/AJPH.2006.103580

TABLE 3—

Random Effects Variances From the Mixed Models for the Effect of Covariates on the Odds of Experiencing Financial Stress Among Ever Smokers (N = 5699): Waves 1–4, the Household and Labour Dynamics in Australia Survey , 2002–2005

Model 1, variance (SE)a P Model 2, variance (SE)b P Model 3, variance (SE)c P
Collection district-level intercept (SE) 0.147 (0.024) < .001 0.115 (0.023) < .001 0.119 (0.023) < .001
Participant-level intercept (SE) 1.917 (0.051) < .001 1.970 (0.055) < .001 2.534 (0.194) < .001
Measurement occasion–level variance (SE) 0.476 (0.006) < .001 0.523 (0.007) < .001 0.472 (0.008) < .001
Measurement occasion–level slopes
Time 0.137 (0.016) < .001
Occupationd 0.119 (0.023) < .005
    Blue collar −0.029 (0.262)
    White collar 1.109 (0.332)
    Professional 0.275 (0.269)

aModel 1 included all occasion-level covariates, and the intercept (i.e., adjusted proportion of smokers experiencing financial stress or mean of material well-being) varied by participant and collection district.

bModel 2 added participant-level covariates to model 1.

cModel 3 built on model 2, because the effect of occasion-level covariates varied by participant (i.e., we allowed for level-2 random slopes). The full model included the slopes for which there was moderate evidence (P < .01) that they varied by participant.

dBlue-collar workers included people in trades, production and transport, and laborers; white-collar workers were clerical, service, and sales people; professionals were defined as managers, administrators, and associated professionals.