Donations that Regress against the Mean

futurelab default header

In his agony economist column for FT, Tim Harford last week tackles a question about how to lure donation money from contributors.

The conundrum faced by the questioner was whether to suggest predefined contribution amounts and risk losing larger donations or not to prompt any amount at all and risk losing micro-contributions from a large majority.

In response, Harford references a study that prompted would-be contributors with ‘typical donation’ amounts and found that the higher the amount suggested the lesser the impact it had on the money donated (after a certain amount.) He concludes by offering 3 pieces of advice: make realistic suggestions, offer a spread of suggestions and try out a few combinations to see which works best.

I also wondered what other kinds of strategies can one adopt in such a situation – especially considering that one isn’t limited to only a couple of alternatives while seeking donations online.

One option is to show the last 5 contributions as suggestions – there’ll be the width there to suit most donors, and suggest what’s acceptable and useful while still nudging those capable of giving more. An extreme case of this strategy would be to throw up only the last contribution as a suggestion but that may lead things to getting stuck in the the low donation rut for a long time – or result in ‘donor’s block’ when faced with a high donation, until someone breaks the impasse. On the other hand, an experiment with this strategy might yield a goldmine of information.

Another strategy that I would like to explore is to inform the potential donor of the average donation until then. Considering that it’s an average, it’s unlikely to be a daunting amount. Plus, in my opinion, it will try and maximise the number of people who pay more than the average – as each person who can pay more than the average will do so (for the very reason that 80% of the people think they are above average.)

The pitfalls of the above strategy are that someone who can pay much more than the average will probably pay up only a little more than the average running at that time. Two, if the number of donors is small or if the collection is run for a limited time, this strategy may risk not optimising the collection. Plus, it tends to collect more towards the end of its run more than in the beginning.

However, in case where the collection drive is running for a long time and there are a lot of potential donors I think it would work splendidly.

I’d like to think to think of this as a ‘regresssion against the mean’ strategy. Now all that remains is for someone to check to see if it actually works as advertised 🙂

[Original pic by Globetoppers]

Original Post: