- Introduction
- Staff
- Methodology
- Publications and Data Sets
Research Department
Methodology
Suppose that we want to improve public services such as local roads or rural dispensaries. Or suppose that we want to increase the efficiency of local administration. Is this goal more easily achieved by an information campaign in the local media, professional training of local bureaucrats, or a combination of both? Alternatively, suppose that we want to reduce sexual harassment of high school students in a given school district. Is it more easily achieved by increasing the number of female principals or by passing a statutory rape law?
These examples illustrate comparing the likely impact of various policies is a critical component of good policy decision making. IERPE believes that randomized evaluations can significantly contribute to development in Africa by helping to answer these types of questions.
Randomized Evaluation
Economic and social reforms tend to be highly intertwined. For example, decentralization is a process that is often concurrent with economic reforms like privatization or structural adjustment reforms, or with political reforms like increasing democratization. As a result, when trying to assess the impact of decentralization, evaluators are faced with the question of whether the differences they observe – for example, before and after the decentralization, or between decentralized and non-decentralized countries – are due to decentralization or are just correlated with it. In other words, it is difficult to identify which portion of the observed changes should be attributed to decentralization and which portion should be attributed to the other reforms.
Identification of the effect of public policies
What is missing in most policy evaluations is a good “comparison group”: a group of policy recipients who are not affected by the policy and that could serve as a base for comparing outcomes between groups who benefit from the policy change and groups who don’t. For the comparison group to be convincing, the best approach is to choose “treatment” and “comparison” groups randomly from the pool of available groups. This approach, commonly called a “prospective, randomized evaluation” is becoming more common in social sciences and it is the norm in medical research. Medical trials involve treatment and comparison groups that are randomly chosen from a sample of individuals. They allow researchers to measure the effectiveness of a medication free from confounding factors like severity of illness or motivation to get treated. In the same way, prospective evaluations in policy allow to check the effectiveness of policies free from confounding factors like other simultaneous policies, differential poverty levels or differential motivation to participate in government programs across groups.
Projects
In the next few years, IERPE aims to apply this methodology in policy-oriented research in the following fields: the politics of public goods provision, decentralization, state capacity building, education and health policies. The IERPE believes that policy research should be integrated into the existing policy decision making process to be fully effective. In that regard, IERPE has developed partnerships with government agencies, political parties, development agencies and local researchers.
IERPE Conference "The Political Barometer of Benin" Results of a nationwide public opinion survey on democracy...