Understanding teachers’ contribution to learning outcomes: Implications for teachers’ policy in Ethiopia

Estimating teachers’ relative contribution to raising learning outcomes at scale is important for informing policy on teachers and teacher development. Much of the research dedicated to assessing the contribution of teachers to their pupils’ progress in quantitative terms uses value-added modelling (VAM). Value-added modelling is typically estimated based on students’ measured learning gains over a course of a fixed period such as an academic year. Unfortunately, value-added estimates generally describe a black box of mechanisms and do not identify which specific teacher practices and/or interactions with pupils lie behind students’ progress. Therefore, it is worth asking whether and to what extent VAM can be used to understand teacher contribution to pupils’ progress, and how it can help us to inform teacher policy. In this article, we examine both questions in the context of Ethiopia. We argue that VAM offers important potential, but that it needs careful contextualization to adequately inform teacher policy. While data suited for value-added analysis are rarely available in low- and middle-income contexts, we are able to make use of such data from the Research on Improving Systems of Education (RISE) Ethiopia programme. Ethiopia is a particularly interesting case study. Important reforms to improve both equity and quality of basic education have been implemented, with a strong focus on strengthening teacher training. Nonetheless, learning outcomes have declined in recent years. Our findings provide additional evidence supporting the critics of unconditional VAM approaches. • In a context where several indicators of teaching quality have improved, pupils’ learning outcomes have nonetheless declined. • Outcomes based on VAMs should be interpreted with caution when used to assess teachers’ quality. • Our findings provide additional evidence supporting the critics of unconditional VAM approaches. • Important differences arise depending on whether unconditional or conditional VAM is used to assess teachers' contribution. • The considerable difference from one approach to the other should lead us to rethink the way we use VAMs.