Introduction—Comparing learner performance in southern Africa: A natural experiment

Abstract

Over the past two decades, the education policy discussion worldwide has shifted, from increasing educational attainment to improving ‘‘educational quality’’—that is, toward increasing student learning at each level of schooling (Hanushek and Woessman 2008; UNESCO 2005). An accelerated pace of international and national testing around the world has contributed to this shift. Despite widespread acceptance of the notion that improving student performance may have a high economic and social payoff, policy analysts in all countries have surprisingly little hard data on which to base educational strategies for raising achievement (UNESCO 2005). A major problem in drawing policy conclusions from analyses within one country or one state or region is that many key macro-educational policy variables, such as teacher recruitment, teacher training, and school supervision, are fairly uniform within such political units. One way to overcome the limits of single-country educational policy research is to undertake comparative studies of neighboring countries (or regions) with apparently similar socio-economic conditions, but significant differences in student performance and, possibly, educational policies. Examples include the United States and Canada, Finland and Norway, and Costa Rica and Panama; on the latter, see (Carnoy, Gove, and Marshall 2007). Some have used the term ‘‘natural experiment’’ for a situation where national social conditions are similar but policies and outcomes differ (Knight and Sabot 1990). In southern Africa, we have such a ready-made comparative case: South Africa and Botswana. The peoples inhabiting the region near the border are of the same language and culture, but students’ school performance, and, for historical reasons, the educational conditions and policies in the two countries, differ substantially. A second problem inherent in such research is the difficulty to identify the effects of school resources on student achievement. For one, much of the variation in student achievement across and even within schools is explained by the socio-economic background differences of students (Rothstein 2005). Further, the distribution of school resources is highly correlated with the family resources students bring to those schools (Barbarin and Richter 2001); students are not randomly assigned to schools, and neither are