Issue with native PCIbex randomization

[KateK]

Hi Jeremy,

We used *randomize(startsWith(“experiment”))* for a study and ended up with a bizarrely distributed sample.

Out of 25 participants distributed across four conditions, for example, we ended up with 10 participants in one condition, and only 1 participant in another condition, which seriously skewed our data.

Could you help me understand why/how this happened? Is this a bug, or an issue with our build? You can see the code here: https://farm.pcibex.net/r/xgObnF/

Thanks,
Kate

[Jeremy]

Hi Kate,

Did you run this experiment like this? There’s an incompatibility with PennController 2.1, you should upload version 2.0 to your project’s Modules folder

Regarding group distribution, note that by default, if multiple participants open your experiment at the same time, they will all be assigned the same group, resulting in imbalance. You can control group assignment by passing it as withsquare in the URL of your experiment

Jeremy

[Author]

Posts