<![CDATA[Pre-service special education teachers are expected to master a broad range of professional competencies including the ability to deliver subject-specific instruction, among them mathematics, to children with special needs (CSN). Yet how accurately these teachers in training appraise their own readiness for mathematics instruction remains poorly understood. This study examined discrepancies between self-assessment and peer-assessment in evaluating mathematics teaching skills among 61 final-year students enrolled in the Bachelor of Special Education program (S1 Pendidikan Luar Biasa) at Padang State University, Indonesia. Using a comparative within-subjects design, participants completed self-assessment instruments (18 items) after conducting a 45-minute mathematics lesson for small groups of CSN, while peers independently evaluated the same lessons using an observational rubric (14 items). A paired-samples t-test revealed statistically significant discrepancies (t = 2.907, p = 0.005; Cohen's d = 0.56): self-assessment scores (M = 4.41, SD = 0.61) exceeded peer-assessment scores (M = 4.13, SD = 0.37) by a practically meaningful margin. Notably, 70.5% of participants overestimated their own performance. The weak, non-significant correlation between modes (r = −.14, p = .277) indicates that self- and peer-assessment capture complementary rather than overlapping dimensions of teaching competence. These findings call for deliberate restructuring of assessment practices in special education teacher preparation specifically, integrating multiple evaluative perspectives, building assessment literacy, and treating discrepancies as a productive site for professional learning.[]]>
