Digital School Census in 10 Weeks? How it was done in Sierra Leone


This page in:

Data being collected from students in Port Loko, Sierra Leone. (Photo: Kabira Namit/World Bank)
Les données concernent entre autres les élèves de Port Loko, Sierra Leone. (Photo: Kabira Namit/Banque mondiale)

Note: This blog is specifically about Sierra Leone’s successful transition to a digital school census but has broader implications for other countries who plan to adopt digital tools at a wider scale to collect data and monitor education and healthcare facilities in their countries. 

In April of last year, the new Minister of Finance of Sierra Leone approached the World Bank with a strong commitment to prioritize education and an intriguing request.

The government wanted to launch their flagship free education program but was facing a significant challenge.  With a complicated education landscape and limited resources, increased government financing needed to focus at the most underserved communities. However, data on schools, enrollment and teachers was scant and outdated. The specific locations of the schools in the country was unknown.  The government estimated that updating the data using the prevailing practice (having schools fill in paper-based forms, collect forms from schools, manually enter the data into a database, clean and analyze it) would take up to a year.
So, the minister asked if the World Bank could support the government in collecting accurate enrollment and infrastructure data for all the 11,000 schools in the country, including pictures, GPS co-ordinates, data on absenteeism and create a teacher database … and do it all in ten weeks.
It sounded like Mission Impossible. The only way to achieve it within the timeframe was to transition to a digital school census where enumerators collected data directly on tablets and updated the database in real time. 
Transition to digital data collection
And so the digital census took place over a period of ten weeks (and several sleepless nights) following five steps:

1 - Form design – The Annual School Census form for Sierra Leone was converted into an ODK (Open Data Kit) format. A number of field restrictions were added to the form to reduce errors in data entry. Making full use of the digital nature of the tool, the form was designed to collect GPS data, photographs of each school and measure student absenteeism.

2 - Procurement of tablets and solar chargers – 600 tablets were required to collect the data within the given timeframe. DFID (Department for International Development, UK Government) loaned approximately 200 tablets for the exercise. The remaining tablets were procured at a cost of US$ 110 dollars per unit. All bloatware was removed and all applications apart from a few that are required for data collection were locked so that enumerators couldn’t download movies and viruses or take selfies.

3 - Training of enumerators –  A cascade model was used for training enumerators. World Bank trained 60 Ministry of Education officers over three days. Then, these master trainers trained 600 enumerators over the subsequent week while the Bank provided monitoring and supervisory support.

Enumerators being trained in Freetown, Sierra Leone. (Photo: Kabira Namit / World Bank)
4 - Data collection - Schools were requested to compile their data in advance through radio announcements and direct phone calls. Enumerators were deployed with ambitious but realistic weekly targets. Each district was required to upload data at least once a week. Data collection was diligently monitored. Daily statistics were shared between the core monitoring team and weekly emails were sent out to senior management with updates on the data collection progress.

5 - Analysis and dissemination – A customized dashboard which was updated in real time was available to the core team. Data was analyzed and customized tables and graphs were ready for the government within a week of the data collection exercise. Currently, the Ministry is considering publishing an anonymized dataset on their website to support wider research and analysis of Sierra Leone’s education sector.

How much did it cost?
The total cost for the exercise (including procurement of tablets), hiring of enumerators, transport costs and data analysis for more than 11,000 schools was approximately US$200,000.
What’s the data useful for, anyway?
A lot. Accurate data enabled mapping of all the schools in the country. A full mapping of education-access by merging poverty, transport and census data supports evidence-based decision making for the construction of new schools.
Distribution of Primary Schools by Remoteness in Sierra Leone, Census 2017-18

Also, the data is currently being used for the distribution of textbooks and to make decisions regarding teacher deployment (calculating the optimal incentive level for teachers to live in remote areas). Other teams across the World Bank are using the data to map education investment behavior.

Should other countries follow suit?
It depends. Technical expertise can only take you so far. Political will and support from enthusiastic civil servants makes all the difference. It should definitely be considered as there are quite a few positive externalities of this exercise – including putting new technologies in the hands of young people.

It’s important to not get discouraged and aim high. In the words of Nelson Mandela, “It always seems impossible until it's done.”

Basil Ogbozor
February 06, 2019

This is amazing! Having adequate data is the beginning of any meaningful planning and implementation of interventions for improvement in any venture. The Sierra Leonian government should endeavour to update their education (school) data base built during this exercise at periodic intervals in order for it to remain relevant.
Another phase of intervention towards improving education in Sierra Leone is to look closely at teacher adequacy for the schools. Further and continuing training of teachers should be implemented as a proirity area for driving the new education initiative. Teacher incentivisation should form part of this exercise. Teachers' promotion should be based on the progress made by individual teachers to upgrade his/her expertise via self-effort improvement (MOOC) and students' performance at monitored external national exams.
Infrastructure and instructional materials for teaching-learning activities should follow. This will include empowering and tasking teachers and education administrators on developing indigenous insructional materials that help the student-learners to easily relate their theoretical iearning with their environment and he opportunities therein.
Another area that requires prority attention is student-learner motivation. The student-learners should be made to appreciate the value in education and encouraged to embrace the new opportunities that the current revolution in education in Sierra Leone presents to them. Merit-based scholarships should be instituted on a continuing basis to ensure that student-learner efforts are rewarded and encouraged. This will serve as a complement to any universal education program that the government may be able to implement either in phases or outrightly as time goes on.
It is also expected that this will be copied by other countries in the sub-Saharan African region and contextualised for the improvement of their educational systems.

February 06, 2019

This is great and I applaud the amazing results obtained! I would just say that the cost seems on the conservative side. Just on tablets (without solar chargers, programming the surveys in ODK and piloting it), 400 tablets at $110 per unit is $44K. That leaves $150k for doing a full-scale education census? Maybe it is true and salaries and mobilization costs for a team of 600 enumerators across the whole country are really low, but I think that this should include all the "in-kind" costs (both on the WB and government side) so as to not set wrong expectations.
From the post, I understand there was a whole technical team on the WB side who's salaries need to be taken into consideration. Flying and accommodations to set up the process and training people, the time of the civil servants and other various expenses are important costs too. There's usually the need of people behind to check and clean the data, as well as to set up reports (even if automatically generated). Not to mention 200 more tablets were loaned by DFID. By stating that the whole exercise cost only $200K, there is a risk that tomorrow country X goes to the WB and says "we have $150k and tablets, let's do a full-scale census as you did in SL". Maybe I am wrong, but I don't think that would be financially viable. Or is it?

February 07, 2019

This is excellent for so many reasons! (i) major kudos for successful implementation in such a short time frame (ii) thank you for breaking down the exact steps, it will help others (aka me) when we are trying to figure out digital data collection for our projects.
Will you be publishing any other products with more details on this process in the future? And final question, could you elaborate on removing bloatware?