Skip to main content
. 2021 May 12;12:599199. doi: 10.3389/fpsyg.2021.599199

TABLE 3.

Unstandardized regression coefficients, standard deviations, and Tymms’ Δ [with 95% confidence intervals] of the multilevel analyses predicting students’ math self-concept.

Model 1
Model 2
Girls
Boys
Girls
Boys
Variables B SD Δ B SD Δ B SD Δ B SD Δ
Within-level
Gender stereotype −0.26 [−0.35, −0.16] 1.13 [1.06, 1.19] −0.43 [−0.59, −0.27] 0.21 [0.11, 0.30] 1.13 [1.06, 1.19] 0.35 [0.19, 0.51] −0.15 [−0.22, −0.07] 1.13 [1.06, 1.19] −0.24 [−0.37, −0.11] 0.13 [0.06, 0.21] 1.13 [1.06, 1.19] 0.24 [0.10, 0.37]
Math grade 0.28 [0.25, 0.31] 2.93 [2.79, 3.06] 1.19 [1.07, 1.31] 0.24 [0.21, 0.27] 2.93 [2.79, 3.06] 1.12 [0.98, 1.26]
Age 0.09 [−0.01, 0.19] 0.65 [0.56, 0.73] 0.09 [−0.01, 0.18] −0.06 [−0.19, 0.08] 0.65 [0.56, 0.73] −0.06 [−0.19, 0.08]
Between-level
Gender stereotype −0.63 [−0.99, −0.28] 0.40 [0.33, 0.47] −0.38 [−0.59, −0.17] 0.28 [−0.22, 0.79] 0.40 [0.33, 0.47] 0.17 [−0.14, 0.48] −0.58 [−0.95, −0.21] 0.40 [0.33, 0.46] −0.34 [−0.56, −0.13] −0.03 [−0.52, 0.46] 0.40 [0.33, 0.46] −0.02 [−0.33, 0.29]
Math grade 0.08 [−0.06, 0.22] 0.90 [0.67, 1.08] 0.11 [−0.08, 0.29] −0.04 [−0.27, 0.19] 0.90 [0.67, 1.08] −0.06 [−0.39, 0.27]
Age −0.05 [−0.10, 0.01] 1.91 [1.70, 2.10] −0.13 [−0.29, 0.03] −0.07 [−0.16, 0.03] 1.91 [1.70, 2.10] −0.20 [−0.48, 0.08]
Contextual effect
Gender stereotype −0.38 [−0.76, −0.00] 0.40 [0.33, 0.47] −0.23 [−0.46, −0.00] 0.07 [−0.44, 0.59] 0.40 [0.33, 0.47] 0.05 [−0.27, 0.36] −0.44 [−0.82, −0.05] 0.40 [0.33, 0.46] −0.26 [−0.48, −0.03] −0.16 [−0.67, 0.34] 0.40 [0.33, 0.46] −0.10 [−0.42, 0.22]

The table shows the pooled results over m = 50 imputed data sets. Tymms’ (2004) Δ was calculated using the formula Δ = (2 × B × SDpredictor)/σ, where B is the unstandardized regression coefficient, SDpredictor is the standard deviation of the predictor variable (for the contextual effect: at the between-level), and σ is the total standard deviation of math self-concept (Model 1: σ = 1.35 for girls, σ = 1.32 for boys; Model 2: σ = 1.37 for girls, σ = 1.27 for boys). Both models also included 10 dummy variables at the between-level, indicating students’ affiliation to 1 of 11 schools (not depicted). Bold values are significantly different from zero (p < 0.05). N = 1,424 (n = 800 girls, n = 607 boys, n = 17 not specified).