Student design competitions connect higher education and professional practice, yet their educational value depends on whether evaluation is transparent, interpretable, and fair. In the context of sustainable design education, opaque and heterogeneous judging practices may weaken the link between competition participation, capability development, and long-term talent cultivation. To address this problem, this study develops a five-dimensional evaluation framework based on design evaluation theory and data-driven modelling. The framework is composed of innovation and creativity, artistic aesthetics, applied technology, work normativity, and practical promotion. A total of 202 award-winning entries from six national competitions in China were re-evaluated by six expert judges under a double-blind procedure, producing 1,212 valid scoring records. Regression modelling was used to identify event-specific weighting patterns and to examine how dimension scores relate to overall evaluation outcomes. The results show clear variation across competitions and experts, indicating that evaluation is not simply the direct application of stated criteria, but a practice shaped by competition-specific priorities and judgement patterns. The framework provides an evidence-based basis for improving transparency and fairness in competition governance and for supporting more sustainable design education through clearer and more actionable evaluation.