Skip to main content
. 2017 May 15;6(2):237–247. doi: 10.1556/2006.6.2017.026

Table 3.

HLM analysis predicting adolescents’ IA scores

IAT score
Fixed effects without robust standard errors Fixed effects with robust standard errors
bi SE T df pi bi SE T df pi
Initial level (initial level as outcome models)
Intercept (time) 29.39 0.94 31.37 32 .001 29.39 0.93 31.46 32 .001
Anxiety 9.33 1.28 7.28 32 .001 9.33 1.23 7.61 32 .001
Extraversion in classroom −6.84 5.37 −1.28 32 .212 −6.84 4.49 −1.52 32 .137
Anxiety × Extraversion in classroom −16.22 7.26 −2.23 32 .033 −16.22 5.16 −3.14 32 .004
Slope (initial level and slope as outcome models) according to age
Intercept (time) −4.68 1.17 −4.01 32 .001 −4.68 1.21 −3.89 32 .001
Anxiety −1.88 1.77 −1.06 32 .297 −1.88 1.71 −1.10 32 .280
Extraversion in classroom 10.37 7.11 1.46 32 .154 10.37 6.63 1.56 32 .127
Anxiety × Extraversion in classroom 5.70 10.96 .52 32 .606 5.70 9.01 .63 32 .531

Note. Reported are (a) the significances pi for increased fit by adding predictors of initial level and age-related changes of IAT scores and the unstandardized initial regression coefficients bi with the Level 1 predictors; (b) wave is centered at the age of 16. Anxiety was used as predictor at Level 2 (individual) and extraversion in classroom was used as predictor at Level 3 (classroom).