A while ago, I wrote a post about why participants don’t turn up to experiments. In my pilot study, I’d had a lot of no-shows, and I was writing about that. I promised to write a “part 2”, where I explained what steps I’d found effective in maximising attendance, or at least minimising inconvenience for you, the experimenter. This is that post.
Here’s the background:
In my main experiment, I wanted 60 participants (all university students), to be tested individually. Each experiment would take around thirty minutes. To make it slightly more complicated, it was important that participants believed that other participants were taking part in experiments at the same time, in different rooms. This means that being more than 5 minutes late was not feasible.
Here’s what I did:
1) I scheduled in more participants than I needed
This is a fairly basic step. The likelihood that you will have a 100% turn-up rate is very low, so booking in plenty of participants is a smart move.
2) I sent reminders
I had a lot of correspondence with participants. When they first emailed, I asked them to give their availability. Then I emailed them with a time, requesting a confirmation. If they replied, I sent them details about the experiment location. If they didn’t, I emailed to ask for a reply. A week before the experiment, I sent a reminder email, stressing that they should get in touch to cancel if they couldn’t make it. Finally, the evening before the experiment, I sent a text message asking for a final confirmation.
I worried that this level of contact might be tiresome for the participants, but they seemed grateful to be reminded. Quite a few cancelled after the “one week warning” email. Some cancelled after the “one day warning” text. That is highly preferable to not showing up, and I had a system in place to help…
3) I had backups
There were more cancellations than I’d expected. Fortunately, I’d kept a list of backup participants. Those were participants who had contacted me after I’d filled all the vacancies. I’d emailed them, asking if they’d be willing to be a backup, and if so what days they were available. This was incredibly useful, and many of my backup participants ended up taking part in the experiment.
Here are the final numbers:
I was aiming for 60 participants.
I scheduled in 65, and had a list of about 15 backups.
In the end, 58 participants took part in the experiment.