If you want immediate practical ideas, my short answer is to follow Givewell recommendations.
If you want to know how I personally think about choosing charities, read on.
My starting position is issues-agnostic. I don’t have any prior commitment to a cause related to my family background or life trajectory. I tend to be suspicious of causes for which I have personal experience which may distort my assessment of the global significance of a cause (but I have nevertheless funded some such causes).
The scale of the modern world and all its problems can seem disempowering compared to the valley world of our ancestors, where saving 10 lives in your valley saved 10% of the “world’s” population. There are two ways of responding to this: you can limit your perceptual horizon to some local valley - medicine, education, the UK, etc (this is what most people do); or you can search globally for the most effective interventions across all horizons. The latter strategy is difficult, but it seems unavoidable if you want to do the greatest possible good.
In principle, I want to choose causes with high expected value, where “value” is beneficence (“quantity of good done”). I might want to diversify across several causes, if I’m risk averse in choosing my portfolio of donations; whether I should be risk averse is not obvious (and discussed further below).
Although this principle is helpful in theory, the quantification of beneficence is difficult in practice. How do you compare relatively certain interventions (say delivery of proven vaccinations) with risky research (say developing a new vaccine)? Can beneficence in different domains - medical, educational, environmental, etc – be usefully compared at all?
Furthermore where you are trying to build civil institutions or improve a culture, it may be difficult to measure any direct effects. Expected value may then be the wrong category or level of abstraction.
Because beneficence is difficult to quantify, I spend little time on explicit numerical estimates. Maximising expected value is my goal, but it’s not my operating framework. I prefer to think in terms of heuristics, which increase my chance of getting a reasonable answer but with no pretence of optimisation. (This is analogous to
my approach to investment. In principle, I want to rank discounted cashflow valuations of every potential investment at risk-adjusted discount rates. In practice, I don’t use discounted cashflow valuations; my decisions are based largely on heuristics.)
Unpopular causes UK hospitals and animal welfare and children's charities are all worthwhile, but there are always plenty of people who will fund these. Your money can have a bigger impact if applied to unpopular causes. You want not "worthy" causes, but "icky" causes - those which most philanthropists will not touch. In other words, the non-substitutability of your money for other philanthropic or government funding is important.
Avoid crowding out Focus on people or situations which are not likely to be helped by for-profit enterprise, government, local philanthropy and community, big foundations (Gates etc), and other small donors who got there before you. Very often one or more of these is often better placed to help than you are.
Positive-sum interventions A troubling feature of many interventions - for example in education and poverty reduction - is that they may have zero-sum properties. Better educational or economic opportunities for some lucky people may entrench and increase the disadvantages of their neighbours. Other interventions - for example most medical and nutritional initiatives - may have higher expected benifence because they do not have strong zero-sum properties.
Leverage This can be financial leverage – giving to a cause may attract further funds from other philanthropists or government. Or it can be intellectual leverage - research which doesn’t do much direct good may change the political climate in a way which makes others (government, or private enterprise) do the good.
Diminishing marginal utility & “room for more funding” The first dollar to each cause probably does more good than the millionth dollar. Often an intervention seems very cost-effective, but money is not the binding constraint, in other words the cause does not have room for more funding.
One possible example of the "room for more funding" problem is restorative surgery programmes in the third world (cleft palates, etc). "$200 to save someone's sight" seems a very cost-effective intervention. But these programs are often not short of money: the binding constraint is the availability of volunteer surgeons, which money cannot relax (except possibly in the very long term). Another possible example: “Microfinance
,might do better on its own terms if less money went into it.”
Diversify, at least a little Some commentators suggest that one should identify a single charity with the highest expected beneficence and allocate all one’s donations to that. The argument is that risk aversion and diversification are inappropriate, because failure of your single best charity is merely disappointing, rather than financially threatening. The world already has a diversified portfolio of charities.
I disagree with this, because it ignores ambiguity. It is often difficult to measure how much good you’re doing, or even whether you’re doing any good at all. Ambiguity justifies some diversification (subject to each of your choices having room for more funding).
Another (unflattering) rationale for diversification is that it represents a compromise between the egoist and altruist in me. The altruist is risk-neutral: he’s not distressed by choosing the charity with highest expected beneficence but very high risk, and then seeing it fail. The egoist is concerned with doing at least some good in his own lifetime, so he diversifies.
Another use for diversification is that it facilitates knowledge generation and makes salient comparatives more available.
SOME BAD HEURISTICS
The following are factors which may suggest themselves, but which I think are poor criteria.
Administration costs Except in the most egregious cases, the proportion of income which a charity spends on staff salaries or fundraising is not a useful criterion in assessing expected beneficence. (Would you judge the entertainment value of a movie by the salary bill for the actors?)
Sustainabilty? The ideal intervention would promote self-sustaining change, so that the need for charitable support disappears. Rather than distributing food, teach the people to fish, etc. This is a desirable goal, but it’s very hard to achieve – so hard that it is probably not a reasonable requirement.
Feeling good By feeling good I mean doing things which give you a warm glow. One actuary advised me – apparently sincerely – that I should give mainly to children’s charities, because nothing would make me feel better than helping children. I think this is a very poor criterion. High expected beneficence may often involve trying to help people I don't particularly like, in ways I am not sure will do any good. I should expect to feel uncertain and anxious about it.
Appearing good By appearing good I mean being involved with popular charities supported by politically well-connected people which get you invited to gala dinners and celebrity events and Buckingham Palace. I have no interest in any of that.
Being good By being good I mean a moral or religious motivation to be good, as distinct from doing good. I don’t really have this. I'm more concerned with outcomes than motivations: consequentialist, not deontologist. (This doesn't explain where the concern for outcomes comes from. Perhaps to seek to do good is already to believe in God, whether one wishes to believe or not.)
Because of the high degree of ambiguity, I diversify at least a little (although much less than I do in investment). I diversify by cause, by continent, by size of organisation and by the extent of my own involvement in the charity's affairs. I contribute to some established organisations with proven effectiveness (taken largely from Givewell’s top recommendations), and one higher-risk start-up with a small high-calibre staff and preposterous global ambitions. (No I won't tell you what it is. My funding is anonymous even to most of the staff. I may tell the story one day if it works!)