Chris Blattman

Should you pay the poor to answer your surveys?

A reader about to run a large survey in Africa asks:

I’m inexperienced here, but a few professors debated the pros for payments:

  • we’re taking a big chunk of these folks’ time (there is an opp. cost to participation)
  • they probably expect it (apparently it has become rather common in Tanzania)

And the cons:

  • they’ll pay attention less (and give us worse information?)
  • participating for the wrong reason
  • we’ll be re-enforcing the standard that participants will be paid

Another thought was giving sugar or a meal to participants.

Good questions. I’ve struggled with this myself. I have three thoughts.

First, it seems to me that the concerns about worse attention or information, and the impacts on participation, are testable. Can I persuade you to do the randomized survey experiment?

Second, the biggest risk to me is the externality: you paying means everybody else must start paying. Personally I think researchers ought to compensate poor subjects for two hours of their time, but none of us has the authority to make that decision for all.

Third, I do neither. No one gets paid for my surveys. But no one goes away empty-handed. We play games for money. Risk games, cognitive games, executive function tests, public goods, or impulse control exercises–depending on the project and question. They walk away with a minimum of a dollar or two, and sometimes more, and have fun at the same time. I walk away with some unique data.


7 Responses

  1. I think it is vitally important that participants are compensated but in a way that fits the research ethos, where for me as a qualitative researcher key tenets are friendship, trust, opennes, so as to create a comfortable space for learning, sharing, and reflection.

    I’ve spent the last year listening to Zambian people’s life histories, often interviewing the same person many times. Living in a low-income settlement, forever observing local conversations, it would have been quite impractical to start paying for every conversation I heard. If I tried to distinguish between certain conversations to pay for that might bias my results as I would be preselecting ‘valuable’ evidence. And I don’t think our two-room home would have coped with a likely queue of people desperate to talk to me in a context of widespread unemployment!

    My token gestures of appreciation and friendship included: always providing refreshments (if at the market: buying bread, avocado and tea for our breakfast, and/or buying their produce (e.g. second-hand clothes and dried caterpillars), or if visiting someone’s home going with ingredients for us to cook together and the expectation of there being leftovers), developing digital photos, bringing small gifts on return trips from the UK. Unfortunately it was sometimes a bit of a struggle to make my participants financially better off, as so many of them wanted to be good hosts when I visited them, buying ME drinks, snacks and lunch… (Does gambling for money have the same problems? Or can you fix it so they win? Or is that patronising? Tricky.)

    But ‘benefit’ and making research a two-way process is not just about money. It’s also about ownership, transparency and accountability. Accordingly, I’ve recently written a ten page synopsis of my research in Bemba, emailed it to a friend in Zambia which she has printed for her mother to share with other people selling dried fish at the market. I’ve also been asked to do a radio call in programme in the local language about my research.

  2. Isn’t it less an externality issue than a social norms issue? After all, shouldn’t the externality runs in both directions- isn’t a choice between imposing the externality of not getting paid on poor subjects versus imposing the externality of having to pay on other researchers? I think your issue has less to do with externalities and more to do with social norms- it’s not the norm to pay your subjects, and if other researchers found out you were paying your subjects they would probably view that negatively and ridicule you.

    Also the argument that paying people could contaminate your results is completely wrong, the opposite is actually the case, assuming you’re looking for a random sample. If people are refusing to participate because they are less altruistic or cooperative, that’s something that could bias your results. If those same people will in fact participate for the “wrong” reasons if you give them money, that’s good for your results. In the surveys I’ve done, the reality is that almost everyone participates anyway so it doesn’t really matter.

    I think the reasonable arguments for not paying people are (as Berk alludes to) that it can be coercive, and also that you would run into problems with people who weren’t part of the sample hassling your enumerators. I’ve never paid anyone for the surveys I’ve done, but only because I think it would be difficult to get someone to fund something with that in the budget. If it were up to me I’d offer subjects something like half a day’s wage for participating.

  3. I’ve always given a small token of thanks. Once, I took a small portable photo printer, and gave respondents a photo of themselves (often with their family), which seemed to them amazingly hilarious. Later, I gave respondents from the same population a bar of soap, and it seemed to be much more appreciated than the photo had been.
    I’ve had this debate a lot with researchers in my field site. In general, foreign researchers tend to argue that something should be given at the end of the survey, while local researchers think it has a negative impact on the quality of the data and creates the wrong incentives. While I typically tend to go with local knowledge and recommendations, I just can’t bring myself to ask someone for an hour of their time and give nothing in return.

  4. Depends if the survey is purely extractive or if they stand to benefit also from the information being collected. If they survey is helping shape or inform activities in their community, then I think explaining the purpose of the survey should be enough to interest people in engagement. If it isn’t, maybe you should recheck your assumptions about the activities you’re planning. If you just want to extract information from people for your own academic purposes or to inform something that will unlikely impact their lives, then you should consider compensating them for their time. Chris’ games for money idea is interesting, but I don’t think its a realistic option for many large surveys, especially those that employ lower skilled enumerators.

  5. I usually partner with a local NGO that serves the population in question. I provide them with training on conducting survey research, compensate the organization for their time and leave a donation for increased services in the area. I usually find that treating participants like human beings with something important to share and listening to them beyond the questions on the survey is more valuable than a small amount of material goods.

  6. Personally I feel that the respondents’ time should be compensated somehow. Last year in a survey in Northern Ghana we decided to give to the respondents a bar of soap. It seemed they had appreciated it.

  7. A criterion many use is a gift that is a token of appreciation, appropriate to the context at hand. The gift should not be too high in value so as to be considered “coercive.” So, someone really not wanting to participate in your study should be able to walk away from it without much hesitation.

    A warning about sugar. The packs break in the field and sugar spills out: not the most convenient gift. Some give soaps, notebooks and pens for schoolchildren, etc.

Comments are closed.

Why We Fight - Book Cover
Subscribe to Blog