This article first published in the Financial Times in April 2016

Not all charities are good causes. This may sound surprising, because we’re used to thinking of them all as being somehow virtuous, but they vary in their effectiveness. Smart donors know that good intentions alone aren’t enough.

Take, for example, the important goal of reducing cases of diarrhoea in Kenya — a major cause of child death. One solution is to provide chlorine at the water pump for people to add to their water when they collect it. Another is to deliver chlorine to households so people can add it there.

Both of these sound pretty sensible, but researchers at the Massachusetts Institute of Technology have found that providing chlorine at the water source prevents more than twice as many cases of diarrhoea for a given sum of money. Put another way, a lot of children will get a potentially fatal disease if donors make the wrong call about which programme to fund.

Figuring out which programmes are good and which are simply mediocre is clearly important. That relies on collecting robust evidence — applying science to the problems we’re trying to solve.

So where should well-intentioned donors turn? I refer you to Zen and The Art of Motorcycle Maintenance, which talks of how “the scientific method is [how] to make sure nature hasn’t misled you into thinking you know something you actually don’t know”. Philanthropic donors are often misled by nature and by their instincts. Sometimes they fund work which doesn’t really need funding, such as supporting organisations which have so much money in the bank that they could fund themselves for several years without any new money.

Others fund work that sounds worthy, but actually exacerbates problems: for example, an anti-drug campaign in the US that may have increased marijuana use. Sometimes, donors fund activity which achieves results, but isn’t as good as an alternative (like the Kenyan example).

To give well, donors need sound evidence — starting with evidence about the problem. For example, how many children can’t read and where are they? This is the purpose of the Annual Status of Education Report which tests the reading and arithmetic levels of children in 300,000 households across India (aser means ‘impact’ in Hindi), and hence shows where educational need is greatest.

Second, you need evidence about how best to solve that problem. The most valuable evidence here is from head-to-head comparisons of competing approaches — such as in the Kenyan water example, though sadly these are rare. Third, evidence about what the ‘intended beneficiaries’ actually want: there are legion stories of outsiders showing up to build, say, a school when what the community really wants is the bus to come more often. And fourth, evidence about other activity to solve the problem: in the UK, animals in The Donkey Sanctuary get more than three times as much funding each as people with mental health problems — it’s hard to imagine that they really have three times the need.

Smart donors therefore consciously use two strategies: use the evidence (assuming it is adequate), or if it isn’t, build it. Dubai Cares, a billion-dollar foundation set up by the ruler of Dubai Sheikh Mohammed bin Rashid al-Maktoum focused on primary education in developing countries, aims to spend 65 per cent of its annual budget on programmes based on existing evidence, and the other 35 per cent on building evidence which Dubai Cares and others will use.

A programme based on existing sound evidence might distribute free bed nets to prevent malaria in sub-Saharan Africa. Such programmes have performed well in numerous rigorous evaluations. Though even then, one has to watch out. In some places, there are reports of mosquitoes changing their feeding times to circumvent the bednets.

A quick way to find some evidence-informed programmes is through The Life You Can Save, an NGO founded by the ethicist Peter Singer (disclosure: I’m an unpaid adviser).

Building evidence includes doing basic research, such as the Human Genome Project, which was bankrolled by the Wellcome Trust in order to keep the ‘human code’ in the public domain. It also includes medical research and evaluations of social and environmental programmes: for example, the Monument Trust, one of the Sainsbury Family Charitable Trusts, is funding an experiment in Birmingham to evaluate various ways of reducing crime by ‘low harm offenders’.

All this evidence then needs to be available and findable, which is not trivial. For example, the AllTrials campaign aims to ensure that the results of all medical trials are published — not just the ones that flatter the researchers.

Finding robust evidence about a pressing problem can be invaluable to your own giving and that of others. Use the evidence if it exists, and build it if it doesn’t. That way we can avoid being misled into wasting our donations.