Introduction. Academic rigour is central to graduate assessment but how written rubrics actually translate into examiners’ judgements remains under-theorized. Objective. This paper investigates how standards of academic rigour are both articulated in written policy and enacted in practice when assessing Master of Arts dissertations. Materials and Methods. Drawing on a qualitative, multi-methods study conducted at the English Department, University of Batna 2, this research project employs a purposeful corpus comprising 120 Master's dissertations that were submitted between 1 May 2023 and 30 June 2025. Additionally, it incorporates the examiners' reports and semi-structured interviews with 12 supervisors and 13 examiners. A stratified sub-sample of 36 dissertations was analysed in depth. Data were examined through document analysis, thematic coding and cross-source triangulation to map written criteria against evaluative practice. Results. The results show that, although official rubrics supply clear procedural criteria, evaluators frequently rely on tacit interpretive standards so that policy and practice align only partially. Three interrelated mechanisms explain this divergence: methodological legibility (how clearly methodological choices make a thesis readable and defensible), supervisory socialisation (the informal norms supervisors transmit), and internal board composition (the mix of examiners’ expertise and expectations). Conclusion. We argue that improving fairness and consistency requires calibrated rubrics augmented with annotated exemplars, routine examiner-calibration workshops, and targeted supervisor development to increase analytic transparency. The study’s significance lies in offering an empirically grounded account of the policy – practice gap, providing concrete interventions for institutional assessment and quality-assurance, and setting an agenda for comparative and experimental research to evaluate the effectiveness of the proposed measures.