Recent developments in mixed-methods research have paved the way for new methodological designs for educational evaluation. Over the last few years, the long-established paradigms of product- and control-oriented evaluation (deeply embedded in education policy-making) and formative evaluation (as an implicit premise for local quality assessment) have been questioned as foundations for evaluating educational reforms. In this paper, a meta-study of a theory-based evaluation, as a third alternative, is conducted. Based on a case study of the recent Swedish national curriculum reform (Lgr 11), the strengths and shortcomings of mixed-methods evaluation designs are examined against the pragmatist criteria of transactional realism and the seven levels of mixed-methods designs proposed by Biesta (2010).