- Data: The results of studies in which people were asked to make the same sort of decision many times (say, at least 30 times), and in each case the decision has to be somewhat arbitrary because the subject isn't being rewarded for anything, nor is there a right answer. For example, subjects might be asked to press one of two buttons, or to push a joystick in one direction or another. Ideally, the dataset would have not just each subject's choice on each trial, but also other contextual information such as reaction time, cursor movements if the subject is using a computer mouse, etc.
- Context: I'm a statistician and psychologist interested in predictive models of decision-making, and I'd like to try my hand at predicting decisions, within subjects, in a task where the decision to be made is essentially arbitrary. I hope that by so doing I can learn about the tricker factors involved in more meaningful decisions.
- Region: Any.
- License: Preferably free-as-in-freedom.
- Format: Preferably free-as-in-freedom.
- Authority: Any.
- Requirements: No other requirements.
- Non-answers: There have been many studies in which subjects are asked to generate random sequences, as by pressing one of two buttons with the goal of making the choice of button as random as possible. I'm interested in cases where subjects aren't told their choices should be random. The fewer instructions and cues on how subjects should choose, the better.
-
You might poke around sites like mturk.com if you want to create a study of your own, or see if anyone has published an mturk random task dataset.– user3856Sep 23, 2018 at 17:30
-
@BarryCarter Yeah, I'll probably be doing my own study (which will probably be on Mechanical Turk) whether or not I can find previously collected data.– KodiologistSep 23, 2018 at 17:43
Add a comment
|
1 Answer
I'm now conducting such a study, which I call Donkey. The data is freely available under the ODbL and the DbCL.