<![CDATA[Differentiated instruction is a central obligation of Indonesia’s Merdeka Curriculum, yet reviews of the domestic evidence have concentrated on primary and lower secondary schooling and have left the upper secondary and vocational sectors almost unexamined. This review maps, appraises, and synthesises 28 Indonesian studies of differentiated instruction conducted in general upper secondary schools (SMA), vocational schools (SMK), and Islamic senior secondary schools (MA), published between 2021 and 2025. Reporting follows the PRISMA 2020 statement. Seven harvests conducted with Publish or Perish across Google Scholar and Scopus returned 2,169 records; after de-duplication, screening against context and scope criteria, and full-text eligibility assessment, 40 studies met all criteria, and the 28 whose full texts could be retrieved were synthesised narratively. Three questions guided the review: which dimensions of differentiation are genuinely operationalised and on what diagnostic basis, what learning outcomes are reported and with what degree of evidentiary certainty, and which enablers and barriers recur in implementation. Three findings stand out. First, differentiation in this corpus is heavily skewed toward process, is most often triggered by a visual, auditory and kinaesthetic learning-styles questionnaire rather than by diagnostic assessment of prerequisite knowledge, and treats product and learning environment as almost invisible; readiness, the dimension with the strongest empirical warrant, informs grouping in only five of the 28 studies. Second, every one of the 28 studies reports a positive result, yet none reports a standardised effect size or a confidence interval, and full-text appraisal exposed reporting anomalies in eight studies, including internally contradictory result tables, an inverted normality decision, and baseline non-equivalence in a study described as randomised that bear directly on how much weight these uniformly positive findings can carry. Third, barriers to implementation were reported far more specifically than enablers, and the clearest enabler identified, grounding grouping decisions in a diagnostic assessment rather than a preference questionnaire, was also the rarest. The Indonesian upper secondary evidence base is therefore abundant but thin: it documents enthusiasm for differentiation far more reliably than it estimates its effects]]>
