4

I want to run a simple probability matching gambling experiment: Each round has the participant pick Up or Down. The computer picks Up with a probability (p=0.7). If the participant gets it correct, they win 1 credit. If they get it wrong, they lose 1 credit. This runs for 100 rounds. It needs to run online.

The problem is that I don't think Qualtrics can do this. Qualtrics can branch and do randomness, but that randomness is geared toward randomization of a survey. The only way that Qualtrics can do this is to do 2^100 branching. This is not good.

Does anyone know how to use Qualtrics to implement this type of study?

Otherwise, we will use something else with Amazon Turk. We believe that the most direct route is to create our own HTML page that implements this as a javascript. None of the online systems that we looked at seem to work well. In pacticular: http://www.tatool.ch/ is good, but requires a java server. http://www.webexp.info/ is too complicated.

Jeromy Anglim
  • 30,741
  • 11
  • 93
  • 221
chongman
  • 143
  • 3

1 Answers1

1

This would be easy to implement in Inquisit. Inquisit has a web interface and is particularly designed for branching and randomisation of trials. If the coding sounds too complex, you can always contact Inquisit and pay them to program the task. From my experience, the cost is not that great for simple scripts.

You can see an example of the Iowa Gambling Task in Inquisit here. That's a lot more complex than what you are describing, but it should give you a feel for what is possible with Inquisit.

You could set up Inquisit with Mechanical Turk as you would most external surveys. See this discussion of mechanical turk.

Community
  • 1
Jeromy Anglim
  • 30,741
  • 11
  • 93
  • 221