Gender disparities in educational and career pathways emerge
early in development and persist throughout the life course, with girls
and boys exhibiting divergent patterns in subject performance, educational
track selection, and occupational aspirations. These differences are
shaped substantially by parental gender biases that influence children’s
perceptions, attitudes, and educational decisions from early schooling
onward [1][2].
Despite growing recognition of this phenomenon, literature reveals significant
measurement limitations: most existing instruments lack rigorous
psychometric validation, fail to integrate assessment across multiple educational
domains (i.e. subjects, school tracks, and professional roles),
and focus disproportionately on STEM-related biases while neglecting
complementary stereotypes in humanities and social sciences[3].
This study addresses these gaps by developing and validating a psychometrically
sound instrument to measure school-related gender bias,
administered to two nationally representative samples: 2,215 parents of
adolescents aged 12–16 and 1,916 young adults aged 18–25. Data were
collected through a structured online questionnaire administered between
July and August 2025 as part of the EduCAR Youth project. Both samples
were drawn using quota sampling stratified by geographical area and
municipality size to ensure national representativeness, and are balanced
by gender and age.
We employed Item Response Theory (IRT) to construct the measurement
scale, using a two-parameter logistic (2PL) model to convert pattern of
responses in a latent trait score indicating the degree of presence of gender
bias [4] [5]. Exploratory factor analysis revealed a clear unidimensional
structure in both samples, with the first factor explaining approximately
70% of total variance. The instrument consists of 11 items covering
three domains: academic subjects (i.e., mathematics, literature, social
sciences), school tracks (i.e., classical lyceum, scientific lyceum, technical
institutes), and professional roles (i.e., engineering/IT professions, leadership
positions, care-oriented professions). The scale performed consistently
across both samples, demonstrating excellent internal consistency
(ω = 0.95), high item discrimination (parents: 1.94–3.35; young adults:
1.92–3.13), satisfactory fit statistics, and a TIF > 15 peaking at higher
levels of the latent trait. Differential Item Functioning analysis supported
partial measurement invariance, with 9 out of 11 items functioning equivalently
across samples.
Young adults (M = 0.45, SD = 0.70) showed significantly higher gender
bias than parents (M = 0.18, SD = 0.59; Kruskal–Wallis Test, p < .001).
In the parents’ sample, gender bias was significantly higher among fathers,
parents of boys, younger parents, residents of Central and Northern
Italy, and those with either high or low occupational status. Higher bias
was also associated with the use of private tutoring, remedial lessons,
and local educational support services, as well as expectations that children
would attend technical rather than academic secondary tracks. In
the young adults’ sample, higher bias levels were found among men,
residents outside the Islands, workers compared with students, those not
planning to have children, and individuals with lower levels of education.
The validated 11-item scale offers a brief, psychometrically robust tool
for screening gender bias across educational domains in both research
and applied settings. Future research should examine longitudinal relationships
between bias and educational outcomes, test cross-cultural
generalizability, and incorporate implicit measurement approaches [3].
