Peer effects, pupil-teacher ratios, and teacher incentives in Kenya
Project Status: Complete
 Printer-Friendly
Project Overview:

While the introduction of free primary education in many developing countries has been very successful at raising primary school enrollment, it has generally not been matched by increases in the number of government-salaried teachers, due to budget constraints. The resulting overcrowding of schools, as well as the influx of new students with little or no preparation, poses a new challenge to communities: how to ensure minimum quality of education.

This project asks what the main determinants of the quality of education are, and how governments or donors can improve learning in this new context. Specifically, the project asks:

  • Does lowering the pupil-teacher ratio improve learning?

  • Do more homogenous classes increase average learning? Do they hurt the students who are "tracked" in the lower-performing class?

  • Can hiring contract teachers locally at low pay help increase students' learning? Can contract teachers perform well despite their lack of experience and low pay?

  • Can empowering the local community to monitor teachers' performance increase teachers' effort and students' learning?


    • This project gathered data from a pilot program that provided schools with funding to hire locally a contract teacher to address overcrowding. The contract teachers were paid approximately a quarter of the salary of regular civil service teachers. The NGO implementing the program was able to offer funding to 140 out of 210 eligible schools in the area; the 140 schools were randomly selected, so the remaining 70 served as a comparison group. The schools that received funding ("program schools") reduced class size in grade 1 by creating an additional section for that grade, taught by the contract teacher.

    Of the schools that received funding, half (called "non-tracked schools") randomly assigned students between the section taught by the contract teacher and the section taught by the regular civil service teacher; in the other half ("tracked schools"), students were assigned to sections based on levels of preparedness, and the contract teacher was randomly assigned to one of the two sections (the lower- or higher-preparedness section). Finally, half of all funded schools (non-tracked and tracked) received additional training on local school committee oversight in order to monitor the locally-hired contract teacher.

    By comparing various measures of education quality (student test scores and attendance, teacher attendance and effort) across schools, it is possible to isolate the effects of class size, local monitoring of teachers, and tracking by initial level of preparedness.

    Policy Relevance:
    As we get closer to the goal of universal primary education and as governments and donors shift their attention from increasing school participation to improving the quality of education, the need for evidence on what affects education quality becomes crucial. This study provides evidence on ways to improve learning in the context of crowded primary schools in poor rural settings. We find that tracking by initial preparedness and locally-hired contract teachers are promising ways of improving educational outcomes.

    Timeline:
    Baseline surveys conducted Spring 2005, endline surveys conducted Fall 2006, long-term follow-up Fall 2007.
    Sample:

    210 primary schools in Bungoma, Butere, and Mumias districts in Western Province, Kenya

    Main Results:

    Hiring an extra teacher on a local contract led to a large improvement in educational outcomes in the program schools on average. However, the impact depended heavily on how the program was implemented. There were significantly larger gains in tracked schools. In particular, students in tracked schools benefited at all levels: students assigned to the lower-preparedness track performed better on a standardized test than students in non-tracked schools after 18 months. This suggests that being in a homogenous class seems to matter more than the average level of peers.

    The smallest impact on student test scores was observed when the class size reduction took place without either tracking or reinforced community oversight. This was due to a negative effect of the program on regular teachers, who were observed to teach less often in the program schools than in the comparison group. This negative effect, however, can be reduced through either community oversight or tracking.

    Researchers:

    Esther Duflo, Pascaline Dupas, Michael Kremer, Christel Vermeersch.
    Papers:

    Peer Effects and the Impact of Tracking: Evidence from a Randomized Evaluation in Kenya
    Esther Duflo, Pascaline Dupas, Michael Kremer.
    August 2009
    Partners:

    Kenyan Ministry of Education, ICS Africa (a Dutch NGO)
    Themes:

    Education
    Funding:

    World Bank and Government of the Netherlands
    News:

    E60-275
    30 Memorial Drive
    Cambridge, MA
    02142     		Contact Us

    Search 
    Site Map