A different approach to scaling up educational technology initiatives


This page in:

the way most projects 'scale up' just might yield inequitable results
the way most projects 'scale up'
just might yield inequitable results

Much is made of the necessity to 'scale up' in international development circles. Here at the World Bank, a quick search on our web site reveals publications and conferences with titles like Scaling Up Knowledge Sharing for Development, Global Scaling up Rural Sanitation Project, Scaling Up Local & Community Driven Development [pdf], Directions in hydropower: Scaling up for development, Scaling Up Affordable Health Insurance, Scaling up School Feeding -- the list goes on and on (and on). 'Scaling up', it would appear, is a goal (and a challenge) across pretty much all development sectors. How can you achieve 'scale'?

It can be deceptively easy to propose a solution to a problem when you don't really understand the problem (especially if you think you do!). The 'failure' of many projects to introduce new technologies in education can, to some degree, be traced back to this simple truism. If you are pointed in the wrong direction, technology can help you move in that direction more quickly. To paraphrase the technologist Bruce Schneier (who was himself paraphrasing someone else): If you think technology can solve your education problems, then you don't understand the problems and you don't understand the technology. The solution lies in process and systems -- and people. Technology can help in all of these areas -- but first we need to make sure we understand what it really is that we need to do.


When participating in planning discussions related to projects meant to introduce and support the use of educational technologies in schools in many middle and low income countries, I hear many complaints about 'pilot projects'. We have too many pilot projects, or so many people seem to believe, and they don't 'scale'. No more pilots is the 'logical' conclusion I then hear expressed from many policymakers, and from many donors. Personally, I tend to think we actually need more pilot projects. Many people automatically conflate (and thus confuse) small projects with 'pilot' projects. A pilot is meant to test something. We may well have too many small projects -- or, stated differently, we have lots of worthy small projects, but these small projects don't provide much insight into how such projects can be replicated or rolled out at a large scale across different environments. Related to the use of ICTs in education, what we really don't have are enough practical examples of how to pilot (activities, projects, approaches) at scale. If indeed 'scale' is an objective (and given the vast needs in so many places, one presumes it is) are there types of 'pilot' projects featuring the use of ICTs in education that might be more likely to scale?

This is an interesting question to consider. For one potential answer to this question -- or at least for insight into one potential approach that may help get us closer to an answer to this question -- a look at recent experiences in Uruguay might be useful.

In Uruguay, they weren't terribly interested in small 'pilot' projects when they decided to provide each student with her own free laptop. (As regular readers of this blog will know, Uruguay was the first country to provide all students in government schools with free laptops.) For a number of reasons, in Uruguay it was decided to meet this objective as quickly and as equitably as possible, which meant that they starting thinking about scale from the very start. (One can agree or disagree with this approach: I am just relating what occurred in this particular circumstance.) In order to achieve scale, the Uruguayans needed some models that would scale. What types of models might those be? Anticipating that that some of the most acute challenges might be found in the most rural and in the poorest communities, they decided to start there. As they were successful in those places, they would then gradually take the models that were working there and apply them in less rural, and in less poor, places, until at the end they rolled out the project in the most affluent communities in the urban capital of Montevideo.

In other words they decided to

Start 'down and out', and then move 'up and in'

This seems to me on its face to have been a largely sensible approach. I don't have data to support this -- there was no controlled experiment here -- but I do note that they were actually able to do this, and do it rather quickly!

Now, whether this approach might work in other places -- I don't know. One reason for this ignorance on my part is that the Uruguayan example is rather singular. I have worked with, evaluated, studied and/or advised scores of educational technology initiatives around the world in low and middle income countries. In my experience, there is a very typical pattern to how these things roll out:

1. There is a need to quickly demonstrate 'success', and so the choice is made to go where 'success' is most likely.
2. Where is success most likely? In urban schools, most commonly. Or sometimes, in semi- or peri-urban communities, provided they resemble urban schools in a number of key ways. Such as ...
3. The schools more or less 'work'. They have good teachers and strong principals. They are supported in various ways by local community groups -- and/or possibly by international NGOs and donors.
4. The schools have access to reliable (or pretty reliable) electricity, to mobile phone networks, and to reasonably reliable (and affordable) Internet connectivity.
5. The schools are accessible to vendors who are able to supply and support ICT hardware and software.
6. The schools can be reached easily by government education officials (for monitoring purposes, to provide additional support, etc.).
7. Classroom level instruction is in the country's dominant language (even better if it an 'international' language, which is what is spoken by most of the students at home).

The choice to start in such places is totally understandable. The thinking is: If it doesn't work there, it won't work anywhere. First, you need to see that something 'works' (a technology, an approach to using technology, an approach to training and supporting teachers to use technology, etc.). Once something is shown to 'work', it is easier to get the political support, and the funding, to expand your success to other schools across the country. (It doesn't hurt if the pilot schools can be easily visited by leading decision makers -- and it doesn't hurt at all if the children of such folks happen to go to such schools.)


If it (the technology, the model, the approach) 'works' in this sort of 'privileged' environment, success may be a product of a number of factors that that don't apply in other, less advantaged places.


If you want to go to scale with your educational technology initiative, first start down and out before you move up and in. Your learning curve will be steeper in the short run. The 'model' you end up with may have more modest goals when compared with what can be achieved in some of the most privileged and advantaged school and communities. But it just might work *everywhere*. Or, if not everywhere, at least it might work in a lot more places than if you had started 'up and in', and then tried to move 'down and out'.

Perhaps you can't start all the way down, and all the way out; perhaps such environments are simply too challenging to do much of anything. That is a political and policy choice that some places may (rightly or wrongly) choose to make. I fully concede that the Uruguayan example may be unique in ways (the country's level of affluence, its small size, its large urban population center, a social compact that highly values equity across society, the dominance of a single national language -- one can cite many such attributes pretty quickly) that set it apart from the situation in many other places. Fair enough. What works in Uruguay may not work in other places. 'Down and out' in Haiti, or in the Democratic Republic of Congo, or in Bihar state in India, or in Papua New Guinea, may represent a reality far beyond what 'down and out' might represent in a country like Uruguay.

But if you are truly interested in doing something impactful with the use of ICTs in the education sector at 'scale', and want to reach all (or almost all) schools, and children, and communities, there is abundant evidence that first starting where it is easiest doesn't tend to offer models that scale much beyond certain types of privileged environments. It does not necessarily follow that taking an opposite approach might be more successful -- but it would be interesting to see some places try to flip the conventional model for rolling out ICTs across an education system. Can't seem to find an approach to rolling out and supporting ICT use across an education system successfully at scale? Try starting down and out before you move up and in.

Note: The public domain image of a scale used at the topic of this blog post ("the way most projects 'scale up' just might yield inequitable results") comes from Wikimedia Commons and is in the public domain.


Michael Trucano

Global Lead for Innovation in Education, Sr. Education & Technology Policy Specialist

Nevine Gulamhusein
June 03, 2013

Michael, I can actualy agree with Bruce Schneier's comment about "The solution lies in process and systems -- and people". Once the process and systems are reviewed, risks are identified and can be contained with appropriate internal controls but the same time, because boundaries are indirectly removed, the same solution can be applied across borders which reduces the implementation time and increases compliance and due diligence. My work in the WBG, Grameen-Jameel and in the private sector has proven this time and again.

Michael Trucano
June 03, 2013

Thanks for your comment, Nevine.
I actually added the 'and people' part when paraphrasing him, but I suspect he would not be too upset by this amendment. So many of the conversations I am privvy to related to technology use in education in challenging environments seem to be, in part, about 'getting around' various human capacity restraints. (For those who don't work in international development agencies, 'human capacity constraints' is basically a rather opagque circumlocation for what normal people might call 'a lack of people with the necessary skills to do ___'.). Yes, where projects are designed well, you may be able to overcome certain human capacity constraints through the use of certain technologies -- but you'll probably run up against new capacity constraints as well that you hadn't considered. Ignoring -- or being ignorant of -- them won't make them go away, you'll need to plan/adapt/adjust/learn in this area as well.

Reem N Bsaiso
May 31, 2013

Michael, you touch upon a core concept – for example: R&D is valuable, but more so is nationalization and application to make good use on a large scale. Scaling is a must to make use of all the resources poured into the pilot, but equally important is to have an impeccable pilot because cost effectiveness versus quality is of the essence in scaled up models. From my experience, scaling up should happen gradually and only when the model has been tested and tried against ‘redundant expenditures and actions’, i.e. losing the unnecessary parts . Sometimes a pilot has to be repeated as is, by multi-partners in different settings before scaling up, but it’s worth the wait.
Once you’ve designed a regional or global scale model, you can go on two ways, find a buy-in from a global or regional partners, or financial partner like a country to support the multi-player application or sell the model one country at a time, different donors, (i.e. dividing the funds over donors or countries), and slowly scale up each model to get the final scaled up effect. On example of a scaled up model I came up with, following 14 country projects, was scaling up in training 200,000 students to 1 million (tested and tried) so that the next model caters for 100 million Arab youth in the MENA over 5 – 6 years. Once aspect I noticed about this training model (initiated by the WB – as you know) is that as you scale up, the cost per student drops down from 1 US$ to 0.26 US$ to negligible. Why ? because once the model is adopted by existing systems, and becomes part of a larger national system, the same resources are used and instead of increasing cost, they add cost effectiveness to resources uses.
It’s important for the model – before scaling up – to have sound assessment and evaluation so that the cost effectiveness, the return on the US$ - i.e. investment return has been verified.

Reem N Bsaiso
May 31, 2013

Given that we live in a fast-advancing technology-based economies, it’s imperative to follow scaled up models to have everyone on board. Otherwise a new digital gaps will appear or widen.
On technology versus education: true…the change is really in the mind-set. The skill I’d go after is: ICT-based knowledge economy and entrepreneurial skills training. Why ? because ICT is a tool, yes a powerful one but remains a tool. Professor Sugata Mitra’s hole in the wall is a great example of mind-over-matter ..if you will. Why “Entrepreneurship” because there is more supply than demand. In the MENA, at least 100 million new jobs are required in the next decade.
As a policy maker in merging ICT in Education, and despite the success of the 2002/2003 GoJ (MoP)/UNDP initiative, results backed up. Meaning, we did graduate knowledge economy enabled students into a stream and an environment that’s more filled up with inhibitors than enablers to knowledge economy – that’s why my recent paper (WB/EIB-CMI) talks about knowledge economy country-level models and KE transfer channels, absorption, enablers, inhibitors…etc . I used the WB KE model as a starting point. To have a sturdy socio-economic regime, coupled with innovation, education and ICT-infrastructure, will make use and build upon the KE-enabled graduates.

Reem N Bsaiso
May 31, 2013

Too many pilot projects ? not a problem, just gather them up, learn from them and design ONE model out of those to scale up. Knowledge sharing is one problem in developing countries. when it comes to country level there does not have to be “ONE Winning idea” but combined ideas that are results-based. Once governments realized this, it will be easy to build on existing resources that has been reshaped to serve the outcome ? some indicators are: economic growth (higher knowledge-based GDP) and job creation on the outcome side. …higher standards of living.
To attempt to answer your question: are there types of 'pilot' projects featuring the use of ICTs in education that might be more likely to scale? You are well aware of one model of ICT-based teachers’ training. I applied this model on the level of 1000 teachers 200,000 students, but when I saw that MENA countries have teachers numbers that can reach to 500,000 (in Saudi Arabia) up to 1.2 Million in Egypt I realized that 20 countries x 200,000 students in nothing. Only one country accepted to scale up the model from 1000 to 10,000 teachers in one go, i.e. millions of students. It worked. But time is our enemy as you know Mike…..
Here is where time and effort is more precious that money.
As for the ‘amount’ of technology to use, it’s more of how and when to use it. Optimal results have been reached with min. technology. Students with the right skills set can create their own. So, once the right mind-set is in place, a laptop per student would pay back. Blended-learning (i.e. face-to-face) plus use of technology is of the essence specially in the first phases when one is introducing technology to render e-readiness for e-learning.
Where to start ? I visited many countries. In ideal situations we’d go country-wide scale heeding geographical distribution and reaching remote areas. But if resources are scares, many counties selected few governorates and few areas.
School-based systems are also very helpful. Dr. Bob Kozma placed one such model (our last WB-2010 Jordan’s Policy).
I like your words on ‘down and out’ first before up and in…actually, one of the ways for me to go forward was going backwards deep into the root of the matter to solve problems with the model before reaching sturdy grounds.

Ken Kaplan
November 18, 2019

I greatly enjoyed your articles. I am in the process of improving the technology to a number of schools in Kibera, Nairobi, Africa and at a school in the central highlands of Kenya. I have made a point of explaining to them that we need to first create a" beachhead" with the educators and insure they have the proficiency to not only utilize the tech as an instructional tool in the classroom, but be able to direct the students to become computer literate. then we can make the strides to infuse it into the students daily life as an asset to student centered learning and as a research tool. I look forward to reading more of your articles in the future.

Ken Kaplan