By Ben Laurance
Good intentions are not enough. In tackling the multitude of challenges faced in developing countries, only with a forensic, disciplined approach can dramatic impacts be achieved with limited resources. And crucially, technology and the facility to collect, analyse and act upon large amounts of data can play a pivotal role in reaching ambitious goals for development and social advancement.
In May this year [2019] Safeena Husain, the founder of Educate Girls, was the key speaker at an event organised by the Wheeler Institute for Business and Development at LBS. In conversation with Rajesh Chandy, Professor of Marketing and Academic Director of the Wheeler Institute, her story – outlining the ways in which her organisation aims to bring 1.6 million Indian girls who are currently not at school back into the classroom – vividly highlighted how the use of data and technology can transform the chances of reaching a target that seems audacious bordering on fanciful.
Every day in India, more than 4 million girls aged between seven and fourteen fail to show up in the classrooms where they should be.
Why? In almost every case, the reason is simply their gender. As Husain put it, for many families, “it’s rooted very much in a mindset where you believe that a goat is an asset and a girl is a liability.” Too often, sending a girl to school is seen as a waste of time. Even worse, she might develop a mind of her own and start answering back. And if she starts at school then subsequently drops out, then so what?
The arguments for educating girls are manifold. Take just a few:
- An educated girl is less likely to be married off while still in her childhood,;
- She is likely to earn between 10 percent and 20 percent more than one who is not;
- Educating girls “positively impacts nine of the 17 sustainable development goals,” Husain pointed out. “Climate scientists have recently rated girls’ education as number six of actions to reverse global warming. This is phenomenal because at number six it is rated higher than solar panels and electric cars. You wouldn’t think about that intuitively, but as fertility rates go down when girls are educated, that actually reduces carbon emissions significantly.”
- Immunisation rates go up 40 percent if a woman is educated;
- And critically, an educated girl is more than twice as likely to ensure that her children are educated. Said Husain: Educating a girl “is something we just have to do once… So, as long as we make it happen for her, she is going to make sure that the education outcomes are achieved for her own children and for generations to come… With girls education and investments, we have an opportunity to close the gender and literacy gap today in the world.”
Isn’t the solution simple? Build more schools, train more teachers and provide midday meals? Clearly, such measures are going to help. But they are not enough. In 2010, India’s Right to Education Act, offering free and compulsory education came into force. Yet research in 2017 showed that many children still lacked basic skills in literacy and numeracy: girls were faring worse than boys.
Educate Girls has looked at the problem in a different dimension: rather than simply provide more resources, see what is actually effective; concentrate on outcomes rather than inputs. Reflecting this approach, it launched the world’s first Development Impact Bond through which funding was linked to the outcomes of its operations, not its activities.
Educate Girls makes heavy use of technology and data – both for identifying where needs are greatest and what strategies yield the greatest impact.
First is the issue of finding out how many girls are out of school and where they live. “We go door-to-door and find every out-of-school girl,” said Husain. “All of our field staff have smartphones with an Android app. This app has digital maps; it has survey questions; it has prompts… All the data we collect is in real time. We can build maps very quickly of where the out-of-school girls are… how many there are in each area so we can make planning much more focused based on the data that is coming out.”
“This survey is at 100 percent saturation. If a district we are going to has 1,000 villages, we go to every single household in that district. It means we have phenomenal data.”
“Once we have mapped out where out-of-school girls actually are, then we can start neighbourhood meetings, individual counselling of parents – really talk to them to bring their girls back into the school system. We handle the whole enrolment process… We work with school management committees to make sure there is the right infrastructure such as separate toilets for girls.” In 2007-08, when Husain was setting up Educate Girls, only about 40 percent of schools had a separate girls’ toilet.
And how does all this data help? “I have data covering close to 4 million households,” said Husain. “Now, we can use advanced analytics to predict, and our predictions show that 40 percent of out-of-school girls in India are concentrated in 5 percent of villages.”
This is an immensely powerful piece of information. India has 650,000 villages. When Educate Girls targets a village, it identifies an individual or individuals who will become advocates of the project – “positive deviants” in Husain’s words, “Young, educated passionate individuals who don’t want to accept the status quo and want to be gender champions.” It is they who will challenge the ingrained mindset that is preventing so many girls from becoming educated. They will become part of what the organisation calls Team Balika.
“We currently have 13,000 villages and 13,000 Team Balika volunteers working with us,” said Husain. “Over the past 11 years, we have brought about 380,000 out-of-school girls back into the school system.”
That is fine as far as it goes. But, achieving its target of bringing 40 percent India’s out-of-school girls back into education within five years will involve a massive expansion of the organisation’s activities. That’s where the analytics come in, said Husain: “The trick is working out which 5 percent of the villages we should focus on and then we can bring 1.6 million girls back into the school system – which means I need to scale from 13,000 to 35,000 villages and find 35,000 volunteers.”
“We are getting larger and larger data sets. We are making predictions. They help us target geographies much faster. It helps us see correlations. There are about 263 indicators in our algorithm, and the majority are economic and social marginalisation… We can map for the whole country and say where we should be going.” That wealth of information can be shared with the government and other agencies involved in welfare and human rights issues, helping them to identify areas of greatest need.
At the core of the use of data is the aim of achieving the greatest impact with given resources. Educate Girls’ activities were looked at by four people – Ben Brockman, Andrew Fraker, Jeff McManus and Neil Buddy Shah – from IDinsight, a non-profit that uses data to help organisations tackling poverty. Their findings were published in the Stanford Social Innovation Review under the title Can Machine Learning Double Your Social Impact? The study makes some important points.
It said: “Making accurate predictions is only half the battle. To truly drive social impact with machine learning, social sector leaders must be willing and able to change how their organization operates based on predictions. Governments and non-profits are often accustomed to one-size-fits-all programs. Shifting to targeted approaches means it must be legally and politically feasible to prioritize action for some people or communities over others…”
“This is often easier for non-profits. For example, Educate Girls can decide to concentrate its resources in clusters of villages where it expects there are a large number of out-of-school girls, rather than work in all villages within a given administrative district. By contrast, governments interested in using machine learning may find that moving from a universal approach to a targeted one is more politically difficult and some cases impossible.”
And turning prediction into action also means taking into account how an organisation actually functions. Educate Girls works by going into groups of villages near one another. A single village in a remote area may be in dire need of help, but to start working there might not be the best use of resources if there are no other villages nearby in need of intervention.
In predicting which areas could most usefully be targeted, the IDinsight team “delivered the predictions to Educate Girls in a way that fits its existing operational model, in geographically compact clusters of villages… [to maximise] the number of out-of-school girls reached without disrupting Educate Girls’ operational model. Over the next five years, we estimate that Educate Girls would be able to reach around 1,000,000 out-of-school girls with its current expected budget and previous approach. However, we expect that by using the predictions generated by machine learning, it will be able to reach around 600,000 additional girls for roughly the same cost.”
It appears indisputable that data collection and analysis can pinpoint the areas into which Educate Girls should go. But, what of the effectiveness of its efforts once it arrives in a village?
Said Husain: “We now have learning data for a quarter of a million children.” That means it can quickly see which schools are doing best in terms of results and which are doing badly. Then it can ask what explanations there might be. She added: “Is it classrooms, is it facilities, is it the number of teachers?”
For example, Educate Girls found that in some areas, some teachers appeared to be getting great results while others weren’t. Because results could be analysed and compared at a very granular level, an important finding emerged: the pupils of a teacher who taught in the village where he or she lived achieved far better results than pupils whose teachers came from outside. Said Husain: “Because they were from the same village, the teachers didn’t just run the classes. At exam time, they went to the student’s home. They said: ‘Are you doing okay? Can I help you? Are you stuck somewhere? Can I help you revise?’”
Also, through classroom observation, and by seeing where pupils perform best, Educate Girls can help teachers up their game. Again, data gives important insights.
The achievements of Educate Girls are undeniable. The use of technology and data to target resources and evaluate outcomes has been key in the organisation’s success. If Husain has a regret it is only that Educate Girls didn’t see the potential earlier. She said: “We could have automated stuff rather faster. If there had been the opportunity, I would have invested much more heavily in the technology.”
Summary
- Every day, more than 4 million school-age girls in India are absent from the classrooms where they should be. Educate Girls, a non-profit organisation, has set itself the target of getting 1.6 million of those girls back into education within five years. It is an immense challenge. But, technology and data analysis are helping it target its resources to achieve this audacious ambition.
- Professor Rajesh Chandy, Academic Director of the Wheeler Institute for Business and Development and a long-standing friend of Educate Girls lead the discussion with Safeena Husain. Much of Chandy’s current research is at the intersection of business, innovation, entrepreneurship, and development. His recent projects have covered the impact of business skills among micro-entrepreneurs in South Africa, novel financing approaches in Ghana, property rights in slums in Egypt, innovation among farmers in India, highways and private education expenditures in India, and using big data for development outcomes.
Latest thought leadership articles
That is fine as far as it goes. But, achieving its target of bringing 40 percent India’s out-of-school girls back into education within five years will involve a massive expansion of the organisation’s activities. That’s where the analytics come in, said Husain: “The trick is working out which 5 percent of the villages we should focus on and then we can bring 1.6 million girls back into the school system – which means I need to scale from 13,000 to 35,000 villages and find 35,000 volunteers.”
“It’s rooted in a mindset where you believe that a goat is an asset and a girl is a liability.”
“We are getting larger and larger data sets. We are making predictions. They help us target geographies much faster. It helps us see correlations. There are about 263 indicators in our algorithm, and the majority are economic and social marginalisation… We can map for the whole country and say where we should be going.” That wealth of information can be shared with the government and other agencies involved in welfare and human rights issues, helping them to identify areas of greatest need.
At the core of the use of data is the aim of achieving the greatest impact with given resources. Educate Girls’ activities were looked at by four people – Ben Brockman, Andrew Fraker, Jeff McManus and Neil Buddy Shah – from IDinsight, a non-profit that uses data to help organisations tackling poverty. Their findings were published in the Stanford Social Innovation Review under the title Can Machine Learning Double Your Social Impact? The study makes some important points.
It said: “Making accurate predictions is only half the battle. To truly drive social impact with machine learning, social sector leaders must be willing and able to change how their organization operates based on predictions. Governments and non-profits are often accustomed to one-size-fits-all programs. Shifting to targeted approaches means it must be legally and politically feasible to prioritize action for some people or communities over others…”
“This is often easier for non-profits. For example, Educate Girls can decide to concentrate its resources in clusters of villages where it expects there are a large number of out-of-school girls, rather than work in all villages within a given administrative district. By contrast, governments interested in using machine learning may find that moving from a universal approach to a targeted one is more politically difficult and some cases impossible.”
And turning prediction into action also means taking into account how an organisation actually functions. Educate Girls works by going into groups of villages near one another. A single village in a remote area may be in dire need of help, but to start working there might not be the best use of resources if there are no other villages nearby in need of intervention.
In predicting which areas could most usefully be targeted, the IDinsight team “delivered the predictions to Educate Girls in a way that fits its existing operational model, in geographically compact clusters of villages… [to maximise] the number of out-of-school girls reached without disrupting Educate Girls’ operational model. Over the next five years, we estimate that Educate Girls would be able to reach around 1,000,000 out-of-school girls with its current expected budget and previous approach. However, we expect that by using the predictions generated by machine learning, it will be able to reach around 600,000 additional girls for roughly the same cost.”
It appears indisputable that data collection and analysis can pinpoint the areas into which Educate Girls should go. But, what of the effectiveness of its efforts once it arrives in a village?
Said Husain: “We now have learning data for a quarter of a million children.” That means it can quickly see which schools are doing best in terms of results and which are doing badly. Then it can ask what explanations there might be. She added: “Is it classrooms, is it facilities, is it the number of teachers?”
For example, Educate Girls found that in some areas, some teachers appeared to be getting great results while others weren’t. Because results could be analysed and compared at a very granular level, an important finding emerged: the pupils of a teacher who taught in the village where he or she lived achieved far better results than pupils whose teachers came from outside. Said Husain: “Because they were from the same village, the teachers didn’t just run the classes. At exam time, they went to the student’s home. They said: ‘Are you doing okay? Can I help you? Are you stuck somewhere? Can I help you revise?’”
Also, through classroom observation, and by seeing where pupils perform best, Educate Girls can help teachers up their game. Again, data gives important insights.
The achievements of Educate Girls are undeniable. The use of technology and data to target resources and evaluate outcomes has been key in the organisation’s success. If Husain has a regret it is only that Educate Girls didn’t see the potential earlier. She said: “We could have automated stuff rather faster. If there had been the opportunity, I would have invested much more heavily in the technology.”
This article was originally published on London Business School’s website.
