Reconceptualizing Educational Assessment Integrity for Sustainable Development: A Multilevel Validity Modeling Approach

Assessment integrity is essential for ensuring fairness, validity, and reliability in education and for advancing sustainable development goals in Ghana. This study employed a quantitative, cross-sectional, correlational design to examine the multidimensional nature of educational assessment integrity within sustainable development frameworks. Guided by Messick’s Unified Validity Theory, data were collected from a stratified multistage sample of 2,400 stakeholders, including students, teachers, administrators, and assessment officers across all 16 regions. Multilevel structural equation modeling (MSEM) was used for analysis. Descriptive results indicated high perceptions of assessment integrity (M = 4.00–4.30), with Consequences, Fairness, and Content Relevance rated highest. Reliability and validity indices were strong (Cronbach’s α = 0.82–0.90; AVE = 0.58; CR = 0.87), supported by robust factor loadings (0.75–0.83). Model fit indices demonstrated excellent fit (CFI = 0.94, TLI = 0.92, RMSEA = 0.045, SRMR = 0.035). Regression findings showed that teacher-related factors significantly predicted Structural Fidelity (β = 0.32, p < .001) and Content Relevance (β = 0.28, p < .001), while female respondents reported higher perceptions of ethical and social consequences (β = 0.15, p = .004). Institutional characteristics, particularly private school status, significantly influenced Generalizability (β = 0.25, p = .002). The model explained up to 14% of variance in key constructs. The findings confirm the multidimensionality of assessment integrity and underscore the influence of contextual and individual factors.