MOE VILLAGE SELECTION
SEJEN CI works with the Ivory Coast’s Ministry of National Education and Technical Training to inform the choice of data-supported public investments.
Context
Côte d’Ivoire’s Ministry of National Education and Technical Training (MENET-FP) has estimated a need for around 1,400 Collèges de Proximité (CDPs) across the country in 2018. As part of efforts to improve the state of education and the Compact signed with Côte d’Ivoire, the Millennium Challenge Corporation (MCC) has committed to building 84 CDPs in 2 regions of the country: Gbêkê and San Pedro. One of the conditions of the Compact is that the choice of localities for these CDPs be based on data in an objective, transparent, factual and reproducible process, thus enabling impartial allocation and equitable impact. Getting young people into school – a VCT objective – is linked to economic potential – but also contributes to reducing the risks associated with HIV/AIDS. To this end, the Data Collaboratives for Local Impact (DCLI) program provided technical assistance to MENET-FP, enabling SEJEN CI to develop a methodology based on data available in the education sector and other relevant sectors.
Problems
Despite the existence of data on secondary schools in Côte d’Ivoire, the implementation of an analytical approach that could facilitate decision-making has been hampered by several difficulties linked mainly to the quality of data on existing schools. Some examples of the problems encountered initially :
- Inconsistent coding for the schools surveyed, resulting in the absence of a join key to cross-tabulate two databases from different sources containing diverse information on schools. For example, only 12% correspondence initially found between two (2) government agency databases;
- Differences in the spelling of school names or localities between different databases;
- Erroneous definition of localities, often confused with neighborhoods, and errors in geolocation (incorrect coordinates) or categorization of environment (rural/urban);
- Data supplied in paper format, which had to be digitized for the purposes of this analysis.
Solutions
On the basis of geospatial and descriptive information on existing schools and localities, obtained from several key structures such as MENET-FP, the Comité National de Télédétection et d’Information Géographique (CNTIG), the Bureau National d’Études Techniques et de Développement (BNETD) and the Cabinet d’Expertise en Systèmes d’Informations Géographiques (CESIG), a database was set up with a view to conducting analyses that led to the following solutions:
A multi-criteria geospatial analytical model to identify an objective list of locations ;
A transparent, reliable, robust and reproducible methodology for selecting locations ;
Training MENET-FP staff to use, update and optimize the tool ;
MENET-FP receives full documentation on the entire process.
Methodology
The methodology used was as follows: a Multi-Criteria Spatial Analysis aimed at identifying suitable areas for the construction of CDPs (based on relevant geographical elements including existing roads, elementary school and secondary schools) followed by a Site Allocation Analysis. The site selection process combined primary criteria (rural/urban aspect, anticipated 2020-2021 enrolment and school map eligibility) and secondary criteria (existing public services and demand for secondary schools). Other criteria were also considered in the selection of sites, such as availability of land reserves, ease of access, surrounding population density, saturation of existing secondary schools, and natural barriers to be avoided (national parks, classified forests, waterways, etc.). To arrive at an optimal solution, the following points were addressed:
Identification of the best criteria following working sessions with all stakeholders ;
Comparison of several models with different approaches to the criterion of distance to be covered by schoolchildren ;
Correction of geolocation errors through appropriate projection of geospatial coordinates ;
Correction of the missing join key problem using Machine Learning algorithms.
Results and impacts
The results led to the selection of a variable number of localities based on different approaches, depending on whether the area of influence of existing colleges was 5, 10 or 15 km. In the end, 81 localities out of 348 candidates were retained for Gbêkê, while 92 localities out of 314 candidates were retained for San-Pedro. These localities will eventually lead to the 84 CDPs that will be built thanks to the Compact with the MCC.
Key contributors
