oTree Forum >

Prolific: Allocate participants to different session configs

#1 by gmdcastillo

Hi all,

I'm planning a session on Prolific and I'd like feedback from those who have experience running interactive experiments there.

I have several session configs, each using different apps. When subjects join from Prolific I'd like to allocate them to one of them. So for example, say I have config1 with app1 and app2, and config2 with app3 and app4, I'd like the first X subjects to do config1 and the next Y subjects to do config2. (In my case, I actually have more than 10 configs...)

I thought about three solutions:

1: I create a single study on Prolific for X subjects and direct subjects to a room. In the room I create a session with config1. Once I have my X subjects, I change the session in the room to config2, and increase the number of subjects on Prolific to X+Y. The Y subjects are directed to config2.

Pros: it's easy and it ensures than subjects cannot participate to both config1 and config2 since I am using a room.
- if one of the X participants returns the study while Y is running, Prolific will try to send me a new participant who will end up in config2 and not in config1.
- it requires me to manually change the session that runs in the room mid-study.

2: I create a single study on Prolific that directs people to my oTree room. I create one session for each config. In the room I run a single app that redirects people to one my sessions.

Pros: same as above + I don't have to fiddle with what is running in the room mid-study.
Cons: same as above - I don't have to fiddle with what is running in the room mid-study.

3: I generate two sessions, one with config1 and one with config2. I create two studies on Prolific: study1 points to the first session and study2 to the second.

Pros: if someone drops out in config1 and Prolific tries to send me someone new, they'll end up in the correct study.
Cons: if I do not run the two studies at the same time, participants could do both study1 and study2. I could exclude subjects who participated to study1 from study2, but that means I need to wait for study1 to finish before doing study2, which goes against my requirement of running everything at the same time.

(I cannot use treatments like we usually do because, while the apps in the different session configs are similar, they have different numbers of rounds and different timings.)

What do you think?


#2 by clintmckenna

Its not the most elegant solution, but could you make a "treatment room" where everyone is sent to from Prolific, then they are randomized to one of your config treatments. Then based on this, they are each split up and redirected to one of your 10+ configs (each of which have a separate room set up)?

#3 by qxiao (edited )

I am unsure how good this solution would be, but I would probably have only one config and manipulate which apps participants go through with a condition variable and upcoming_apps[]. To ensure that at least X participants finish your config1 before participants are assigned to config2, I would include a session-level variable as a counter and add 1 whenever a participant completes the study. Once this variable reaches your desirable N, participants are given the condition that would direct them to the other config. You won't get the exact number of participants since there will be participants with config1 who are still in the process of completing the study, though.

Write a reply

Set forum username