Every two months, 800 women gather in a church courtyard in the village of Tritriva, Madagascar, to receive cash from the Malagasy government. Mothers of six- to 10-year-olds get the payment only if their children have regularly attended school. For those with children under five, it’s unconditional – but they are given information about family health and nutrition.

With more than half of Madagascar’s children chronically malnourished, it is vital these women take note. But the problem is not just financial. Breaking long-term habits, such as spending the bulk of your income on rice, is extremely difficult – especially, according to recent research, for those living in extreme poverty.

This is where nudge theory comes in. It is about using insights from behavioural science to identify reasons why people make bad choices, such as smoking or failing to pay taxes on time, and then testing small changes in the way choices are presented to “nudge” them into making better decisions. In the school example above, it was about optimising how these families spent the money they received from the government.

A team of researchers from the World Bank and ideas42, a behavioural insights consultancy, is about to launch a series of experiments on the mothers of Tritriva. One will involve them selecting stickers with pictures of nutritious foods; in another, some of the women will be appointed as community leaders and trained in parenting and children’s health. The team will measure the success of these experiments to find out the best way to persuade Madagascan women to buy healthier foods.

It may seem insignificant – a distraction to the serious structural issues global development practitioners face, such as corruption and poverty. But Varun Gauri, head of a new global nudge unit run by the World Bank, the Global Insights Initiative (Gini), defends nudge theory. “Why not make all programmes as effective as possible, even if it doesn’t turn a very poor country into a Scandinavian country overnight?” he says.

Many of the problems governments and NGOs in developing countries are trying to fix, he says, are at least partly behavioural. Another example comes from South Africa, where more than 5 million people have HIV. Teenage girls are three times more likely to be infected than boys their age. A team from ideas42 found that many girls were choosing to have sex with older men, believing them to be safer sexual partners, when they were in fact more likely to be HIV positive. The team then found that the girls were better able to retain knowledge about which men are more likely to have HIV if they played a simple computer game about HIV-risk, than if they were simply given information to read.

Sometimes it’s about getting the incentives right. Humans tend to think in the short-term, and many people in developing countries will not buy deworming pills even when they are very cheap. A study in Udaipur, India, found that by offering people free lentils they would turn up en masse to get their children immunised. The nudges also have to be exactly right. For example, in Costa Rica, telling people that they use more water than their neighbours is likely to encourage them to use less of it; but telling them they use more water than others in their city has very little impact.

Gini was launched in October 2015 under the premise that “development policy is due for a redesign based on a more realistic understanding of how people think and behave”. Saugato Datta, managing director at Ideas42, believes the standard approach to development policy has relied too heavily on the principles of economics – the price of goods and the level of subsidies, for example. “A lot of policymakers trained as economists,” he says. “They have not really tried to look at some of the other stuff.”

He also points to the “implicit paternalism in development policy” – too many programmes are decided from on high by people who think, “This is good stuff we are giving people, so they have got to use it”.

Toilets, for example. Nowhere is open defecation more prominent than in India, where more than 600 million people have no access to a toilet. But even where proper sanitation has been installed, “people tend not to want to use latrines,” says Gauri. It is a difficult habit to break, exacerbated by issues such as India’s caste system.

The team at Gini will be helping the Indian government work out how to meet its target to eliminate open defecation by 2019, measuring the success of certain interventions, such as asking people to commit to using toilets, and “no toilet, no bride” – an initiative that asks parents to ensure their daughters are not married to men whose villages have no toilet.

No longer is development policy just about taxes, subsidies and laws, says Gauri. That the World Bank has set up its own nudge unit really signifies to NGOs, developing countries and their governments that the time to start using a behavioural approach has arrived, he says.

Nudge theory does have its critics. One of the fiercest charges Gauri faces is that of manipulation – that behavioural scientists are taking advantage of people’s weaknesses to make them do things that are not in their best interests.

It is a charge Rachel Glennerster, executive director of J-PAL, a global health charity, feels should be taken seriously. Speaking at a behavioural insights conference hosted by the UK behavioural insights team in London in September 2015, she called for humility on the part of behavioural scientists. Glennerster urged experts not to assume they know best and discount local knowledge – citing how expert advice given to farmers in Kenya on fertiliser maximised yields but did not maximise profits for them.

The other major criticism of nudge theory is that nudges might have unintended consequences. However, nudge supporters point out that every time a government or NGO rolls out a new law, policy or practice, these are also likely to have unintended consequences. “The idea it can be fully unpaternalistic is a bit of a pipe dream,” says Datta, of Ideas42. At least with nudge theory, which is often based on randomised controlled trials, governments and NGOs can gauge the impact of new projects before they’re scaled up.