First Lessons from Our Study: Optimising Messaging for Returns to School

Despite two years of recovery since Covid-19 lockdowns, school systems worldwide still experience lower post-pandemic attendance levels than pre-Covid. Marginalised learners have been disproportionately affected by the pandemic. Several strategies are being used to accelerate returns to schooling. Messaging is one strategy that is often regarded as likely to be highly cost-effective. 

Our Hub-led research in Ghana tests and compares different messaging models to support caregivers in their efforts to encourage their children’s attendance at school. Read on to see what we are learning about messaging interventions, their cost-effectiveness, and how they could be contextualised, as well as what we know about messaging to increase school participation.

In this blog, we explore the following:

• What we know about the mechanisms of messaging
• What we are learning from our study (and important questions to ask along the way for anyone wanting to design a similar intervention)

And we ask the following of you, our reader:

• We invite you to connect with us to collaboratively improve our list of lessons

What we know about the mechanisms of messaging

How can the use of messaging support school attendance?

Text messaging is a simple, low-cost medium that can be used to share information that supports caregivers with their children’s school attendance. In the context of low- and middle-income countries (LMICs), text messaging:

• is typically much cheaper than methods requiring data bundles
• can reach a significant number of people — mobile phone adoption rates are increasing year on year, and non-smartphone rates are higher than those for smartphones.

Below, we focus on four mechanisms we identified through a review of the relevant literature and our preliminary research and the barriers to attendance they attempt to address.

Mechanism 1: Changing or increasing the information available to caregivers in their decision-making

The World Bank’s comparative cost-effectiveness analysis of education interventions lists sharing information on the benefits, costs, and quality of education as the only ‘great buy’. This means it is the only reviewed intervention across 150 studies in 46 countries supported by a strong evidence base that is likely to be highly cost-effective (either because of its significant benefits or because of its low costs). 

Studies like Nguyen (2008) or Jensen (2010) show that when presented with new information, households can process it and change their decisions regarding their investment (in time, money, etc.) in schooling, provided there are no other major constraints that would limit it (e.g., extreme poverty). Studies in Chile, the Dominican Republic, Madagascar, Pakistan, and Peru, for instance, have shown that providing information about the benefits of education to caregivers has increased attendance and learning at a low cost.

Beyond communicating the benefits of education, information can also be shared with the intent of targeting and shifting social and cultural norms. For instance, an intervention could specifically target​ caregivers’ beliefs about gender norms (e.g., where a caregiver might see investing in a boy’s education as more important than in a girl’s). 

Finally, information sharing can target low caregiver investment in education resulting from the gap between caregivers’ aspirations and expectations for their child’s education (e.g., a scenario where a caregiver would think: “I want my child to complete secondary education, but I expect them to only be able to complete primary”).

Mechanism 2: Supporting caregivers to address context-specific barriers 

Locally relevant information that focuses on context-specific barriers faced by caregivers also has the potential to decrease dropout rates. This could address gaps in caregivers’ knowledge about local schooling (e.g., not knowing the school calendar) or assist caregivers in overcoming obstacles to their children’s school attendance (e.g., how to deal with minor children’s illnesses). As Aurino et al. (2022) have shown, in some instances, marginalised caregivers might lack resources and human capital (e.g., exposure to formal education) to resolve some of the barriers to attendance. In this case, information provision is not so much about decision-making as it is about caregiver self-efficacy.

Mechanism 3: Increasing caregiver engagement with their child’s education

Caregiver engagement has a significant and lasting impact on student attendance, behaviour, and achievement (Desforges and Abouchaar, 2003; Goodall and Vorhaus, 2011). In particular, caregiver self-efficacy — their sense of their own ability to engage with and support their child’s learning — is fundamental.

Messaging can be used with the objective of increasing caregiver engagement, which in turn could increase attendance. For instance, messages can encourage caregivers to participate in a set of activities related to their child’s education, including visiting the school or attending school events or discussing their child’s current and future education or career plans with them.

Mechanism 4: Increasing teacher competencies and shifting teachers’ attitudes

Although less talked about, messaging interventions can target teachers, addressing behavioural barriers relating to teacher competencies and attitudes (de Hoyos et al., 2021; Muralidharan and Sundararaman, 2010; Piper and Korda Poole, 2010).

Beyond likely and obvious resource-related barriers like a very low teacher-student ratio or issues like teacher absenteeism, there are competency-related barriers to teachers supporting attendance. For example, teachers might lack know-how about identifying, integrating or tracking absentee children. Additionally, teachers’ attitudes towards absentee children could also be a barrier to higher attendance levels. For instance, long-term absentee children might be discouraged from either attending or fully engaging in class because they are afraid of being disciplined by their teachers. Messaging interventions targeting these barriers would then seek to increase teacher competencies and shift their attitudes.

Where does that bring us? A high variance in efficacy

Despite some promising signals and examples in the existing literature, research results are very mixed as to whether improved attendance can be achieved — there is a high variance in the efficacy of these different interventions. Some studies have even found instances where messaging can have adverse effects on caregiver engagement with education.

While it is clear that no solution should be seen as a silver bullet, this is even more true for messaging where its low-cost appeal might unhelpfully encourage ineffective use cases. 

This variance in effectiveness is, therefore, a critical research gap.

So what makes messaging effective?

Little research has provided a detailed assessment of why some messages are effective in some contexts and with some recipients and not in others. There are significant unanswered questions regarding how different types of technology can be used most effectively, the contextual factors that influence uptake of the calls to action contained in the messages, and how these messages should be personalised to be more effective.

Our Hub-led research in Ghana tests and compares different messaging models to support caregivers in their efforts to encourage their children’s attendance at school. Through our delivery mixing research and innovation methods, EdTech Hub is hoping to learn about factors that make messaging a cost-effective and contextually appropriate way of supporting school participation. 

Below, we present the first lessons we are learning from our preliminary research. These lessons are flexible and easily integrable snapshots which could help a team when designing a messaging intervention which seeks to increase school participation. However, they do not prescribe a specific way of implementing the said intervention. 

Four lessons we are learning from our preliminary research

Effective messaging interventions that support attendance typically:

1. Understand recipients’ context and profiles to target the right behavioural barriers to participation.

The intervention should be responsive and sensitive to the context, needs, and profiles of the recipients to target the right behavioural barriers to participation. 

This means understanding the rhythms of recipients’ lives, the systems they’re in, and the factors driving or preventing participation in formal education. This includes how the messaging interacts with other external factors and interventions that affect the recipient. For instance, the use of messaging has been shown to be most effective when it fills a contextual knowledge or information gap that exists (e.g., the benefits of education are unknown or not prominent in the recipient’s mind).

Additionally, for the messaging to be effective, the recipients must have the means and motivation (pre-intervention) to act on the information received. Messaging has been shown to have negative impacts where the messages crowd out motivation or pressurise individuals. Caregivers may need a base level of human capital and resources (e.g., exposure to formal education) to respond to the messages and change their educational investments in meaningful ways; without it, messaging can induce increased stress and lower self-efficacy, cresting overall negative effects on attendance.

Ask these questions first: Who are the recipients, and what differing characteristics of the recipients might warrant different interventions (e.g., content)? What competing priorities in the systems recipients find themselves in would drive or prevent a behaviour shift? Based on our understanding of recipients’ lives and their context, which behavioural barriers should the messaging intervention address? Which aspects of the programme may cause additional stress, and how can you avoid it?

Example: Spend time understanding the recipients and the systems in which they exist. Assess whether the desired behaviour shift is realistic given the constraints and competing priorities that may have a greater impact on participation. Provide time and contextually relevant information.

2. Adapt delivery based on recipient profiles (e.g., caregivers, teachers, marginalised learners, and literacy levels, household structure) and the technologies available to them (e.g., text messaging, internet-based messaging apps).

This means:

• Understanding the diversity of recipients (e.g., low vs high levels of literacy, language spoken, language read, special needs, etc.).
• Taking into consideration literacy levels, including thinking through how a person who cannot read texts gets the messages read to them. Drafting content in the most-read language (as opposed to the most-spoken).
• Drafting content that is easy to understand and tone-aware, including considering drafting messages and revising the content in collaboration with a group of recipients.
• Understanding the impact on the reach of different technologies (SMS, WhatsApp) and media (text, audio).

Ask these questions first: Who are the recipients? What technology is available to them? What is their possible interaction with the messaging intervention? Will the messages reach the right people in a single-phone household?

Example: Develop content in the most-read language. Consider the channel that’s most accessible to our audience. For instance, WhatsApp is cheaper for the implementer but requires the internet, while SMS is more suited to remote areas.

3. Consider different behavioural principles to increase the nudging potential: The intervention should use tested and proven behavioural insight techniques, which can potentially make the call to action more effective. 

This means integrating principles of behaviour design, where possible. With the right content and timing, the potential to shift behaviour is higher, as well as the possibility of making the changes stick. 

For instance:

• Messenger: ​​People respond best when the message is from a credible, knowledgeable source or someone they trust.

Ask these questions first: Who is your message from? Who would caregivers and teachers respond best to? Have you considered the cultural approach to authority, safety, trust, and power and clearly signed or branded your message?

• Attractive: We are more likely to do something if we find it attractive and rewarding. What we find rewarding differs from person to person — it depends on our unique blend of internal and external incentives. Evidence suggests that nudging in education is most effective when interventions are aligned with participants’ beliefs and behaviours (Damgaard & Nielsen, 2018), drawing on culturally relevant examples (Madaio et al., 2019).

Ask these questions first: What are the incentives in the system you’re looking at? Are they internal or external?

• Social: People look to the actions and behaviours of others to determine what they should do, especially when they feel uncertain. 

Ask these questions first: How might you surface the social norms that could shift behaviours of the people you’re trying to reach?

• Timing: How a new behaviour is timed and how it interacts with what already exists in a typical day / week / month etc., is key. Giving participants prior notice of a planned messaging intervention has been shown to be more effective. Avoid busy periods (e.g., harvest season) and consider what macro-context may cause additional stress on caregivers.

Ask these questions first: How can you contextualise and personalise the time for sending messages? What existing behaviour can you hang on to (e.g. getting to work in the morning)? What existing event can you align with?

• Easy: We are more responsive when things are made easy for us. Our message should be simple, clear, and concise. 

Ask this question first: What is the one call to action?

• Regular: The frequency and duration of the intervention should aim to balance fatigue and provide enough content for recipients to take action.

Ask these questions first: What rhythm and frequency will you use to set up the habit?

Example: Make the storyline of the messages simple and straightforward enough to be delivered in a few words. Include an ‘action point’ that is easy to understand. Consider timing reminders about attendance at weekends for parents and on weekdays for teachers. Be intentional about how many messages are sent in a week, and think about how many messages participants can receive before fatigue sets in. 

4. Iterate to learn and allow for gradual learning about what works. 

Within our work, this means building the optimal intervention iteratively (e.g., using experimentation) and integrating intentional learning moments with the team.

Ask these questions first: How are you going to learn what works? How can you act on evidence as we implement and avoid limiting ourselves to the ‘one-stop’ learning moment at the beginning and/or at the end? What is a good learning rhythm for the implementing team? 

Example: Set a learning rhythm with the implementing team, and allow pivoting when relevant. Test the messages with a sample of recipients before rollout. Constantly review messages and check to see if participants accept them.

Help us to improve our list of lessons

EdTech Hub wants to keep learning and refining our lessons list for this study. Our goal is to develop further insights into cost-effectiveness so that the evidence produced can subsequently support other organisations in implementing this type of messaging intervention more effectively. Our preliminary results will be available this summer.

So, let’s work together. We invite you to connect with us to collaboratively improve our list of lessons. If your work has produced insights into the effectiveness of messaging interventions that seek to improve attendance, reach out! Our study will run until the end of June 2023, at which point we will seek to organise discussions around our findings.