The project “Comparative Studies into Teaching Approaches for Mathematical Modelling” (CoSTAMM) has so far involved a total of 453 participants in four studies, all focused on fostering students’ mathematical modelling competency through different learning environments, consisting of the same five-lesson modelling unit and two different teaching styles (“method-integrative” versus “teacher-directive”) which represent certain elements of quality teaching. All studies followed the same pre-post-test intervention design with the same instruments but differing in their educational contexts. Three studies were implemented at the tertiary level in South Africa, and one study at the secondary level in Germany. Against this background, this paper examines the question to what extent differential effects of the interventions may be explained by the implemented specific learning environment or by the specific educational context. For this purpose, all studies were integrated into a comparative linear mixed model and analysed at once. An additional qualitative evaluation of the participants’ responses to the test items was carried out to supply deeper insights. Overall, the modelling unit proved to be effective independently of the educational context, and more effective for the groups that followed the method-integrative teaching style. This paper thus provides empirical evidence that an educational intervention can effectively be transferred from one context to another while yielding simultaneously positive learning gains.