When missing data matters: Imputation, TOPSIS, and the misrepresentation of Morocco’s education system

Abstract

Annual reports, such as the Global Justice Index Report for Education, depict Morocco’s education system as underperforming globally. Such a situation results in large-scale reforms that require many financial resources. However, these rankings rely on opaque data-processing pipelines and unknown imputation of missing data. This study addresses this methodological opacity by proposing a reproducible reconstruction workflow that makes the data sources, missing-data treatment, imputation evaluation, and ranking procedure explicit. Therefore, we evaluate eight imputation strategies: mean, median, mode, hot-deck, linear interpolation, multiple imputation, KNNI, and LSTM. Among the evaluated imputation methods, linear interpolation produced the most accurate and stable results, with MAE values ranging from 0.6519 to 1.4935 and RMSE values ranging from 1.0343 to 2.4458 as missingness increased from 20% to 50%. Then, we apply a TOPSIS-based multicriteria decision framework to rank countries. The results show that Morocco evolves from a bottom-tier position in the early 2000s towards a stable middle group by 2023, following an accelerating upward pattern. All materials required for reproducibility, including code and data, are provided at: https://github.com/mouddentarik/Education_Ranking .