Active learning has become widely recognized as a desired strategy for teaching in engineering education. Many of the active-learning instructional materials have been developed in the US. Instructors in non-English speaking countries wanting to implement these materials therefore face the choice between (a) using the materials in the original English versions, and consequently teaching in a language other than their students' native language; and (b) using translations into the local language. Two sets of implementations of a particular set of instructional materials for introductory physics form the context for our investigation, one in Mexico and the other in Germany. At each institution, we have collected and analyzed data from students in locallanguage courses and courses taught in English. Results from both institutions indicate that, on average, students using the materials in English perform better than those using a translation. In our effort to explain this result, we have identified several variables that affect student performance. These include instructor factors such as commitment to active learning, as well as student factors such as reasoning skills, English and native language proficiency, mathematical skills, and prior academic performance. In our report, we analyze and discuss the influence of each of these factors.