Mary’s helpful guide to soliciting research participation on the ‘net

In my years on the ‘net, I’ve seen any number of people want to interview others or get them to take surveys for everything from a short high school or undergraduate paper through to graduate research projects and books. And they so seldom manage to meet basic ethical guidelines for making sure they aren’t wasting their participants’ time at best or endangering them at worst. Hence this article.

In addition, this article may help research participants better assess requests: are researchers telling you what you need to know? Have they considered your interests as well as their desire to Find Something Out At All Costs?

Full disclosure: I am not a research ethics expert, I am simply a researcher helping you get the basics right. Please seek expert advice if you have any doubt about the safety or integrity of your research.

Why do I need to do this stuff?

Because you’re so often asking people sensitive stuff, that’s why!

Look, I have some sympathy for the “it’s just questions about something-seemingly-small!” myself. I ask people questions about their linguistic intuitions. “Which sentence reads better to you, A or B?” There’s nothing less fun than completing a 31 page ethics application to get approval to ask people about which sentences read better.

But look, all research, at best, takes up people’s time. You owe people something for that. In addition, quite a lot of the research people are recruiting for on the ‘net wants to get into harassment of women, political affiliations, sexual experiences, why people write slash. That kind of stuff? That kind of stuff in the wrong hands loses people jobs and relationships. You owe people serious, well thought out harm mitigation for that.

So, ethical research recruitment lets people know what they’re getting into, whether it is a boring half hour sharing linguistic intuitions, or sharing potentially damaging information with a reseracher.

The bare minimum

All researchers asking for participation should share this information:

  • Who are you?
  • Who do you work for or who commissioned this work, if not yourself?
  • How can I get in contact with you, and how can I get in contact with who you are working for?
  • What is the purpose of the research?
  • What is the status of the research? Is this sheer curiosity that made you whip up a survey in five minutes, or a pilot study, or the main game?
  • What kind of effort do you want from me? (Interviews versus surveys. Five minutes versus many hours. You get the idea. Tell me upfront what my time investment is.)
  • When you’re done, where can I see the results?
  • Will the results be made public and in what form? (A peer-reviewed article? A PhD thesis? A pop science book? On your blog?)

Some of this might be the sort of thing you want to put on a webpage you can link to, so you can leave short advertisements like “Hi, I’m looking for help with X, and thought readers here might want to help because of Y, if you need to know more, please see LINK.”

You;d be amazed how many people miss the “When you’re done, where can I see the results?” step. Even if they’re asking people for 20 hours of interviews or something like that. For anything but the most trivial investment of time, letting people read your results is the minimum reward required.

Also, results being made public can often be good: the subject’s work is contributing to the sum of human knowledge! So don’t consider this necessarily a bad thing in and of itself.

Institutional research

If you are doing research at the postgraduate, postdoctoral or faculty level, research using human subjects (and other animal subjects for that matter, but you aren’t likely to be recruiting them on blogs) requires ethics approval by an institution-level ethics committee in most institutions.

So, when soliciting participants for research that has ethics approval, provide the following info:

  • All the bare minimums plus
  • A statement citing your ethics approval in whatever manner is usual. Your committee probably has boilerplate. Typically this will name the institution, give a reference number for your experiment and provide contact details for the ethics committee.
  • If your ethics committee approved a recruitment advertisement, use it! If it’s long put it at the other end of a link if that’s OK with them.
  • If your ethics approval requires that you disclose a bunch of things, also state them or place them at your info link if allowed.

If your institutional research didn’t require ethics approval (some institutions might, for example, have a blanket policy covering low-risk things like linguistic intuition questionnaires) find whatever boilerplate they let you use instead, if there is any or say something sensible along the lines of “This questionnaire comes under the XYZ University Low Risk Experimentation Policy [link].”

Basically, if you are doing research on behalf of an employer state either that you have ethics approval, or if not, why not (eg, your institution has no committee).

No committee but doing something sensitive?

If you’re doing sensitive work outside the oversight of ethics committees, here’s the start of your checklist!

  • All the bare minimums plus
  • Are respondents going to be anonymised in your personal/researcher copy of the data? Are you stripping any associated names, IP addresses, email addresses and similar? If not, what are you keeping and why?
  • How are you storing the researcher copy of the data?
  • Who has access to the researcher copy of the data? (Yourself? Your boss? All of your boss’s present and future employees? The Internet?)
  • When do you plan to delete the researcher copy of the data, if ever?
  • Are respondents going to be anonymised in the published results? If not, what identifying information will you publish and why?
  • Can a respondent withdraw their participation and be deleted from your data or transcripts? How do they do it? How long do they have to do so?

There are all kinds of other factors that ethics committees would get you to look at, basically, what capacity for harm does your research have? How are you mitigating that harm? What risk to your participants is left?

Risks include: physical health risks; mental health risks (more common with online data gathering, eg, triggering questions); exposing people to relationship disruption or breakdown, or abuse (by, eg, asking them to discuss infidelity); exposing people to criminal prosecution (eg by asking them to discuss illegal drug use); exposing people to civil liability (eg by getting them to discuss breach of contract), exposing them to job loss; denying them the best treatment or resources (by, eg, giving preferential treatment to patients or students or employees who agree to take part in the research, thus harming others); and coercing participation in general. And there’s one question that frankly stands out to me as a member of the apparently rare species Lady on the ‘Net, which is “are you studying an over-studied population and if so, what benefit does this extra research have for them, as opposed to for you?”

One of the most obvious mitigation strategies is anonymity of your subjects in reports, and eventual data destruction of any private identifying data. But as you can see from the examples related to coerced participation, it isn’t the only strategy you might need. List your possible harms, list your mitigations, let the potential subjects decide if the research is worth it to them.


I wrote a similar post focussed on software development a few years back, in that case mainly focussed on “prove to your subjects that their participation is not a waste of their time.”

Categories: education, ethics & philosophy

Tags: , ,

5 replies

  1. Great article. I have seen some less than wonderful studies/surveys promoted and quit many surveys half way through due to some of the issues you mention above.

  2. Excellent!
    A less important point but one that’s often annoyed me: if you only want respondents from the US, do us the courtesy of saying so up front rather than have a question on page 35 asking us to give our zip code or select from a drop-down list of US states.

  3. Great post! I would also add that for some projects it’s worth considering a minimum age restriction – there may be more ethical problems with under -18s.

  4. Kirstente: definitely, and even legal concerns.
    More broadly, in academic research there are populations called vulnerable populations. For my ethics applications they include under-18s, prisoners, patients, particularly one’s own patients if you do clinical work, pregnant people and indigenous people.
    Some of these classifications are in themselves concerning: to give a linguistic example indigenous and other minority/oppressed language communities often actively seek the assistance of linguists in documenting their language and reducing them to “subjects” at all isn’t ideal, their status is actually that of co-investigators, or in some cases bosses (as in, they commissioned the research). But ethics committees seldom analyse it that way.
    Another better known case is that the efficacy and safety of various drugs in pregnancy is seldom assessed partly because the barriers to getting approval to recruit pregnant participants in clinical trials are enormous. And there are definitely benefits as well as potential harms to treatments being found to be quite safe or safer-than-the-alternatives for pregnant people and, if they desire, their fetus.
    However, at least knowing which populations are considered vulnerable and when and why helps with ethical design.
    Miss Prism: given that “wasting your subjects’ time” is itself an ethical concern, that one is definitely important. Any disqualification questions/conditions should be first up and also stated in the information page. (Occasionally one’s experimental design requires that people don’t know they’ve been disqualified, or you can only do so after looking at all their answers, in which case, actually you don’t usually tell them at all. I can see arguments each way on that, you don’t want them to feel obliged to come back and do your survey “right” next time. This can be a case where paying people a small amount for their time is reasonable, to offset the risk that you won’t use their data.)
    Relatedly, if you do want international respondents, allow for, eg, monetary amounts to be stated in various currencies. It’s not only time-consuming for me to translate, say, my income into USD, it’s also probably misleading, since I am not in the US job market and shouldn’t be analysed as part of it. Likewise, allow both metric and imperial amounts to be input if needed, and provide prompts in both metric and imperial, if it’s necessary to give or receive measurements. (Obviously without a full translation of your survey into many languages, you will already be limited to speakers/readers/writers of your own language.)

  5. I’m glad to know several new and important questions I would not have thought to ask a researcher about the anonymising of their database, who would have access to it and for how long, and where I would be able to see their results for myself. This is hugely useful stuff.

%d bloggers like this: