StanfordHCI / bang

💥 Helping people meet for the first time, more than once 💥
MIT License
2 stars 1 forks source link

randomize masked and unmasked rounds #365

Closed phoebexy closed 5 years ago

phoebexy commented 5 years ago

Currently:

Experiment rounds 1 and 2 on the template determine when masked rounds are delivered as tasks are created for each round, mapped onto the round number. However, we want masked rounds to be delivered randomly per batch.

Potential Solution:

Create types of tasks instead of task per round. For example, when creating a task, the user has the option of tagging that task to be delivered for "masked" or "unmasked" rounds. Then, randomize the order that masked and unmasked rounds are delivered and choose the task used based on the round type.

deliveryweb commented 5 years ago

Снимок экрана от 2019-07-17 14-06-11 what about this? And I don't understand what is 'masked' and 'unmasked' rounds. They all are masked. Experiment round 1 teams === Experiment round 2 teams. That's all.

phoebexy commented 5 years ago

@deliveryweb Sorry, poor definition on my part:

By masked rounds, I mean experiment rounds where team members are working with the same team, but don't know. Right now, we specify which rounds are going to be the same team. However, this should be randomized so on some runs we have rounds 1 and 3, and on others we have rounds 2 and 4 be the same teams.

Does that clear it up?

markwhiting commented 5 years ago

@xehu I think you had proposed the switch to a fixed experimental schedule but not sure. If so, is that something you need every time, or sometimes?

In other words, should be have some kind of control for this on the batch level?

deliveryweb commented 5 years ago

I think it's done. Check it on dev pls.

xehu commented 5 years ago

@markwhiting

Just to follow up on this- Phoebe’s solution works for my purposes. My reason for fixing rounds was a temporary solution because I have different tasks that I need to deliver for experimental versus non-experimental rounds. (I.e. I wanted people in the experimental condition to debate case A, but people who are not in the experimental condition to debate case B.) At the time that I suggested the temporary fix, it wasn’t possible to predict when the experimental round would occur, so I wanted to fix the rounds in order to make a “quick and dirty” hack. What Phoebe suggested is a better long term solution.