Digital peer learning in poor schools

Enrolment levels in schools in India has gone up but learning outcomes remain pathetic. 60.2% of kids in 5th grade in India cannot divide a 3 digit number by a 1 digit number, 34% cant read at grade 2 levels. Improving the situation by teacher training, greater monitoring has failed. We would like to improve learning outcomes in schools.

Enable curiosity and engagement among kids though a structured peer learning method that involves role play. The method moves kids to the higher levels of Bloom’s taxonomy by building activities in increasing levels of complexity. The teacher becomes a facilitator.

The solution involves quite a bit of technology – very data driven, automated workflows, system assigned roles for kids. This will allow us to scale. Science and math content can be reused across the world – so that scales as well. 

60.2% of kids in 5th grade in private schools in India cannot divide a 3 digit number by a 1 digit number, 34% can’t read at grade 2 levels. The situation is a shade worse in public schools. These kids when they grow up will have a huge challenge earning a livelihood. The standard method of teaching in the schools is “chalk and talk”, “rote learning” – this enables passing an exam without actually learning much. Rough estimate – there are 230 million children in the Indian school system. There are at least 120 million children who are performing well below potential in India. The kids’ engagement levels in schools is very low.The teachers unfortunately have not been exposed to anything different – so the learning outcomes have been stagnant at very low levels. There has been no change in pedagogy inspite of all the changes that have been happening in the world. Scital wants to improve engagement levels among the kids, improve pedagogy to improve learning outcomes.

Describe your solution and how it works in simple terms. What is it? What does it do? What processes and technology does it use?

• We create activities that are curriculum aligned, designed to move kids to the higher order thinking on Bloom’s taxonomy
• We divide the kids into balanced groups and the group solves the problems. They use online and offline resources to get to a solution
• We randomly pick 3 children from different groups and assign them roles of “Presenter”, “Challenger”, “Summarizer”. The challenger reviews the solution of the presenter and asks questions. The summarizer summarizes the discussion
• This is followed by formative assessment
• Gaps measured through the assessment is bridged using peer mentoring where the peers are selected based on results of formative assessment. Assignment is totally aligned based on which kids has a gap in which competency to which kids has mastered that competency

The whole process is orchestrated using a tech platform. The method ensures high levels of participation by kids. The learning process is highly collaborative. The entire process is very outcome focussed and data driven.

India has over 380,000 private schools that cater to children from a low income background. A large chunk of the kids are first generation learners, in most cases the parents would not have had sufficient schooling. Bulk of the schools are in smaller towns and villages. Very often for the school owners, the school is merely a mechanism to monetize their real estate. The teachers are from the local community with insufficient exposure to pedagogy practices, digital. The system is trapped in a cycle where outdated chalk and talk methods continue with no changes over the years. The engagement levels are extremely low. This has resulted in very poor learning outcomes for the kids in this segment. PISA ranking of 182 out of 183 for India is a result of this situation.

Scital has experimented with the peer learning method that is a combination of activities, role play rendered on a digital + paper based platform with over 650 kids, 18 educational institutions and 50,000 hours+ testing for figuring out learning outcomes across different age groups.  The testing has helped Scital evolve its ideas, develop content and develop a spec for a tech platform that can be deployed in a low resource setting that is extremely outcome focused.  Scital has pent over 1 year with a single cohort of students, compared learning outcomes of students who have gone through the Scital method against a control group. We have seen a 27% higher academic scores in the bottom decile of students compared to a control group.

The solution does the following:

a) Move away from chalk and talk lecture delivery to a high engagement activity based peer learning

b) The model involves randomly selecting "Challengers" - who ask questions, "Presenters" who answer questions and "Summarizer" who summarize the discussions. This ensures broader participation levels compared to a traditional Indian classroom

c) Because the solution is partly digital, it builds digital literacy early on - something completely absent in this segment

d) The activities consciously integrate concepts and subjects to move the kids to higher levels on the Bloom's taxonomy. The teachers have a guide for the activities as well. The aim is systemic improvement

The challenge - Low learning outcome. The root cause - low engagement levels because of chalk and talk method of teaching, teachers not adequately trained on content, pedagogy. 

The solution design involved - collaborative, social, problem based and inquiry based learning that leads to substantially higher levels of engagement. The process and system ensures that teachers move from lecturers to facilitators, the problems are designed to move kids to higher order thinking on Bloom's taxonomy. Random assignment of roles of challenger, presenter, summarizer ensures broader levels of engagement. Activities solved in small groups ensures kids pay attention.

1. We selected smaller town, low income schools because they have significantly poorer learning outcomes

2. We opted to start with 6th grade kids because we observed that standard deviation in academic scores seems to rise dramatically at 6th grade

3. The two schools where we spent maximum time was Citizen School in Hoskote ( considered to be a part of rural bangalore) and Pearl School in Harapanahalli ( Small town, literacy levels below national average). 

4. The pilot with these two schools was about 95 children

5. We also experimented with 550+ young adults in smaller towns as a part of "Karnataka State Government Elevate 100 prize" on peer learning to improve employability.  But we settled on 6th grade because we felt that this will produce maximum impact over the longer term. 

Positive impact of peer learning is well researched. The downside is, in a group, there are non-participants who don’t benefit. Our first innovation – adding role play (presenter, challenger, summarizer picked randomly). This forces engagement among all kids.

Part II – Kids do the activities in a platform like google meet. We integrate participation data from this platform along with formative assessment to do the adaptive learning and feed into a workflow. For instance, in a class 20 kids do well in individual formative assessment, 5 don’t.  The options are a) The teacher can re-teach the topic b) The kid can be given a bunch of content to go through and redo the formative assessment c) There can be peer mentoring. The idea is data from participation + formative assessment will be used to develop what is optimal for a kid.  All adaptive learning platforms rely solely on formative assessment – we combine that with social participation and custom-built workflow.

 In our tests, we have seen a 27% improvement in academic scores of bottom decile and a 32% reduction in standard deviation.

 Currently the two methods that dominate the market are a) Chalk and talk teaching b) Kids self-learn using video content.  The impact from both is not great – completion rates of education videos is abysmally low. With demonstrated improvement using our method, schools will embrace collaborative learning. We will open up the platform for other enterprises to launch peer learning based courses.

Pre-pandemic + last academic year, we served about 750 students. This year is very uncertain - schools are shut down. The third wave is expected to impact the children - so it is not clear whether schools will operate and how will they operate this academic year.  Our initial plan was to serve about 3250 students in academic year 2021-22. Raise funds, improve organizational capability and scale to a point where we could serve 65,000 students in 4 years.  Given the situation, our hope is, serve 65000 students in 5 years instead of 4 directly. But in about 3-4 years time get the platform to a stage of maturity where independent entrepreneurs can use our platform and impact educational outcomes. Through the independent organizations, we hope to serve half a million students in 5 years.

There are two parts to the impact – Number of students impacted, improvement in academic scores (of the bottom decile of students and the class average). The first part – number of students is quite straightforward. The model requires the whole grade in a school to participate and we bill based on the class size.

For measuring improvement, the process carried out during the pilot was as follows:

• Take the academic scores of the assessments of 10th grade students of the 2019-20 cohort in a school – we took scores of all the tests conducted in the first 7 months of school.  This is the baseline.
• For the 2020-2021 cohort of 10th grade students – they went through the Scital program. They had the same syllabus, same teachers, same tests – but different pedagogy. We took data of all assessments for the first 7 months
• Compare the scores of the two cohorts to measure changes

The gaps that Scital has not been able to figure out yet – in a peer learning methodology, the social and communication skills of the kids showed visible improvement. Class participation during a lecture also improved – but we could not figure out a way to measure the improvement.  So for now, we plan to go with academic scores.

 3 contractors, 2 full time, 3 interns. But because of the pandemic, we have paused operations - low income schools unlikely to operate this academic year

In Phase 1, our focus is academic outcome, develop tech specs and model for scale. So the current team is academic heavy. Curriculum lead is a Teach For India fellow with experience in low income school. The interns are teach for india fellows as well with experience in curriculum development and teaching low income schools. Board of advisors - 3PhDs + 1 Masters in education; Curriculum development - 1 PhD in Biology + 1 Masters in English. I have run the operations for DIRECTV Lat Am - so know how to run large operations and an a board member of Mantra For Change - a non-profit that works with over 30,000 schools. 

Currently, the team is women dominated ( aside from the Founders, the entire team ( including interns) is female , different religions and geographies represented). Low income schools in general have significant number of lower caste, minorities - and this is our target market.As we deploy our solution, the plan is to hire locally from the community where the low income schools operate for academic support, account management and sales. Having representation from the community is integral to our business model and in fact serving the lower income group is the purpose of the organization

Multiple reasons for applying to Solve (in no particular order) -

a) We have tested our method with 650+ kids, 50,000 hours.  In a comparison with control group, we have seen a 27% better academic outcome of the bottom decile.  Having said that, we need guidance with rigorous methods to test at scale. 

b) Access to academic mentors - if our solution does not produce sufficient improvement in academic outcomes at scale, folks who can help us course correct

c) I have worked on social enterprises over the last 8 years. Hopefully, can contribute to the group as well as get inputs from the peer group

d) Access to funds

e) Getting selected to an MIT program would dramatically improve credibility with the schools that we want to help

f) A 9 month program where I can associate with MIT - that would be awesome!

Technology - Scital platform would have individual formative assessment but learning will be in groups and collaborative. You have data from formative assessments, from activity reports as well as chat conversations in breakout rooms. We need to use all of these for providing learning pathways as well as optmising group composition.  Adaptive learning platforms just take assessments as an input.  We need to go beyond that. Will need help in thinking through the algorithms.

Monitoring and evaluation: Consistently delivering improved learning outcomes is key - that is the mission.  Having robust monitoring and course correction in case the performance is not good enough is extremely important. We will need help in putting this together

Financial: The impact investor pool in India is tiny. Need access to a larger pool. Also, will need help in putting together the cost accounting systems in place, review financial plans periodically.

MIT Faculty:

a) Dr Grace Lin, Learning Scientist and assessment designer: Would love to get mentoring support for putting together competency framework and assessments

b) Ira Fray, Learning game designer - would love to get inputs on activity design, nudging students so that participation improves consistently

c) Alex Hargroder - would love to get his guidance on project based learning

d) Prof Peter Senge - While we are focused on improved learning outcomes, this cannot happen without institutional improvement. There is an interdependency within the school ecosystem as well as the community. Would be great to bounce ideas with Prof Senge

Solver Alumni:

a) Practical Education Network , STEM education in West Africa - would love to learn from their experience and collaborate on an on-going basis

b) 40k plus: We have a similar mission.  Associating with them would shorten our learning curve

c) Dost Education: We dont have a model for parent engagement.  Need help in putting this in place

Scital's target segment is the low income group in India.  This group does very poorly in STEM - over 60% of grade 6 kids cannot divide a 3 digit number by a 1 digit number.  Scital's primary intervention is in STEM education in this segment - basically collaborative activity based learning in STEM and English.

Technology - Scital platform would have individual formative assessment but learning will be in groups and collaborative. Formative assessment will be based on competency framework.  You have data from formative assessments, from activity (or project) reports as well as chat conversations in breakout rooms.   Adaptive learning platforms just take assessments as an input.  We intend to go beyond formative assessment and also incorporate extent of participation, the types of conversation to provide learning pathways. For instance, would the performance of a group improve if one kid is replaced, should remedials be via peer tutoring or should the teacher have a one one one. The idea is to be highly data driven and create models that will help us optimize learning outcomes.

Technology - Scital platform would have individual formative assessment but learning will be in groups and collaborative. Formative assessment will be based on competency framework.  You have data from formative assessments, from activity (or project) reports as well as chat conversations in breakout rooms.   Adaptive learning platforms just take assessments as an input.  We intend to go beyond formative assessment and also incorporate extent of participation, the types of conversation to provide learning pathways. For instance, would the performance of a group improve if one kid is replaced, should remedials be via peer tutoring or should the teacher have a one one one. The idea is to be highly data driven and create models that will help us optimize learning outcomes.