Methodological Challenges and Continuing Social Concerns A Reply to Serpell and Wertsch’s Comments

It is always interesting to discover how one is interpreted by one's peers in an institutionalized setting such as the pages of Human Development, and I am doubly grateful to the editor and the commentators for this opportunity to re-reflect on the issue of the developmental consequences of schooling. The fact that my commentators have generally picked different aspects of the overall presentation as the focus of their comments makes my job somewhat easier by allowing me to consider them more or less separately before commenting on a major shared concern. Let me begin with the methodological issue that provided the major focus of a great deal of my own work over the years: How, given that institutionalized public basic schooling (IPBS in Serpell and Hatano's termionology) is a specially organized form of experience that is not shared by all developing humans, can we reach firm conclusions about its general cognitive consequences? The logical prerequisites for reaching such judgments (given the logic of explanatory science) are that we compare people (generally, but not necessarily, children of different ages) who have, or have not, experienced IPBS for different amounts of time. In addition, such people should be chosen at random and the tests used as our proxies for cogni-tive consequences should be equally familiar to both groups. The first prerequisite , random assignment, could be achieved only under the most draconian social circumstances, so various proxies, 'natural experiments,' are used in their stead. The decision of external agencies such as AID to build roads along certain routes in Liberia in the 1950s and 1960s and place schools in towns along those routes, or the Mexican government's decisions to place schools in some but not in other communities in rural Yucatan provided such natural experiments. One had to be cautious and take into account possible contaminating factors (roads change more than the presence or absence of a school), but a rough approximation to random assignment could be achieved. The second prerequisite , finding dependent variables that were equally experienced by those who had been to school and those who had not, posed the problem that absorbed most of my concern. Historical analysis of the tasks used to