Millisecond Forums

Inquisit, online studies, and possible fraud

https://forums.millisecond.com/Topic12587.aspx

By jbfleming - 3/7/2014

We are nearing going live with our first Inquisit web-based study. We have approached the design decisions with a mind to preventing fraud (to the extent that we can) and making it easier to detect (since it's not possible to prevent). (We are especially concerned because we are offering compensation for participation in the study.)

We do have a couple remaining questions about Inquisit's capabilities though:

1. Can we prevent someone from taking the study with a participant ID that has already taken the study? (My experience with the desktop version makes it seem like that it will let you take the same study with the same participant/group over and over again. If that's the case, can that behavior be changed or is that just how it is?)

2. And a more general question, are there best practices others who use the web platform use to avoid/detect fraudulent participants?

Our concerns with #1 are that once a participant screens is and is assigned an ID and group and are given the URL to the Inquisit-hosted study, they could use the URL more than once, or modify it to change their participant ID, or whatever. We have other safeguards in place that should help us identify that behavior and they need to provide a valid mailing address to be compensated. But it would still be better if we can avoid that situation entirely.

Thanks!
By Dave - 3/7/2014

Re. #1: No, there is no way to positively, absolutely prevent this. The reason is this: Inquisit Web experiments aren't interactive web pages (think: online survey platform) driven by an underlying central database where all responses etc. go. Inquisit Web scripts are effectively executed locally on the client machine (to ensure maximum precision and performance w/ respect to stimulus and response timing) virtually as if they were executed using a local Inquisit Lab installation -- the browser is merely the channel used to deliver the core Inquisit "engine" and the experimental materials (scripts, stimulus files) to the client computer. As such, an Inquisit script running on a client machine cannot "know" things like "How many other people have already participated in the study?", "Has this subject identifier already been used before? How often?" etc.
By jbfleming - 3/7/2014

Makes sense. And we wouldn't want the study running on the participant's computer to have any access to the study data anyway, of course. Thank you!