TABLE 3.

Adjusted gender-disparities before and after the reform on standardized test performance, subject-specific self-concept and interest, and school-related stress and health.

	b G9	SE	p	b G8	SE	p	Δ b	SE	p
Standardized test performance
Biology	19.29	6.41	0.002	19.98	5.80	0.001	–0.69	8.85	0.943
Mathematics	78.57	5.53	<0.001	68.44	4.76	<0.001	10.13	6.86	0.142
Physics	89.90	4.61	<0.001	90.77	5.19	<0.001	–0.86	5.62	0.872
Subject-specific self-concept
Biology	−−0.01	0.54	0.992	−−0.19	0.42	0.650	0.18	0.68	0.783
Mathematics	4.02	0.44	<0.001	4.45	0.42	<0.001	–0.42	0.59	0.468
Physics	6.55	0.43	<0.001	7.01	0.44	<0.001	–0.46	0.62	0.467
Subject-specific interest
Biology	−−0.51	0.57	0.369	−−0.96	0.56	0.088	0.45	0.74	0.550
Mathematics	1.94	0.53	<0.001	2.37	0.50	<0.001	–0.43	0.75	0.564
Physics	6.55	0.51	<0.001	5.43	0.59	<0.001	1.12	0.78	0.147
School-related stress
Difficulties to relax	−7.42	0.71	<0.001	−8.40	0.81	<0.001	0.98	0.90	0.271
Exhaustion	−5.08	0.66	<0.001	−−6.09	0.49	<0.001	1.01	0.72	0.161
Overload	−2.05	0.51	<0.001	−−3.68	0.70	<0.001	1.63	0.69	0.017
Malaise	2.82	0.97	0.004	1.48	1.10	0.176	1.34	0.87	0.123
Alignment issues	−3.29	0.67	<0.001	−−4.14	0.82	<0.001	0.86	0.77	0.265
Health
Overburdening	−3.63	0.70	<0.001	−−4.90	0.74	<0.001	1.28	0.75	0.089
Achievement-related fear	−6.17	0.69	<0.001	−−7.85	0.72	<0.001	1.68	0.65	0.010
Diverse symptoms	−12.36	2.66	<0.001	−−13.31	2.84	<0.001	0.95	0.95	0.586
Uneasiness	−−2.19	2.39	0.361	−−2.71	2.56	0.288	0.53	0.78	0.498
Depressive symptoms	−3.39	0.66	<0.001	−−4.31	0.62	<0.001	0.91	0.59	0.125
Gastrointestinal issues	−3.46	0.92	<0.001	−−3.86	0.93	<0.001	0.39	0.64	0.541

b _G9 = Gender differences before the reform; b _G8 = Gender differences after the reform. Positive values indicate higher values for boys. Δb = Difference of gender differences before (G9) minus after (G8) the reform. The metric of the latent variable was transformed to M = 500 and SD = 100 for standardized test performance and to M = 50 and SD = 10 for all other outcomes using pooled means and standard deviations. Statistically significant differences (p < 0.05) are printed in bold. Regression coefficients (b’s) are based on group mean differences in a multiple group model. Two-sided p-values are reported. In cases where we had a directional hypothesis based on prior literature (e.g., higher stress scores of girls), one-sided p-values should be calculated/interpreted, which can be calculated by dividing the reported two-sided p-value by 2. Covariates that were considered for adjustment can be found in the Instrument section. For achievement, explained variance of the latent variables ranged between 12 and 44% (M = 26%), for self-concept between 1 and 15% (M = 6%), for interest between 1 and 11% (M = 4%), for stress between 0 and 6% (M = 2%), and for health between 0 and 13% (M = 5%). Note that when excluding course level as a covariate, results remained comparable regarding the size of estimates. In addition, in these models all SEs in the adjusted models were smaller compared to the unadjusted models. Please be aware that when estimating results for stress and health, we applied exploratory SEMs, which led to slightly different measurement models (i.e., differences in factor loadings), when additional variables (e.g., covariates) were considered and this explains differences in SEs between the adjusted and unadjusted solution. Although the general loading pattern (see Supplementary Tables 3, 4) remained similar in adjusted and unadjusted models, SEs should not be directly compared across these two solutions, because they refer to slightly different latent variables.