search

carpentries-incubator / logistic-regression-public-health

Logistic regression for public health
https://carpentries-incubator.github.io/logistic-regression-public-health
Other
1 stars 4 forks source link

Ethics of direction of coding in intro to binary response variable #9

Open ailithewing opened 3 years ago

ailithewing commented 3 years ago

As this is specifically for public health, could you reconsider the direction of coding for your example:

E.g. 'In this context, we would consider an individual that still smokes a “success” and an individual that no longer smokes a “failure”.'

Seems weird to consider still smoking as a 'success' in this context....

Just me?

ezherman commented 3 years ago

Hi @ailithewing, I have tried to solve this issue by including a callout box at the end of the episode, in which the SmokeNow variable is re-levelled. I hope that hereby students are reminded that "success" and "failure" are just labels in logistic regression, while learning how to change the direction of coding if they so wish.

Does seem like a good compromise? The alternative would be to re-level inside the episode, but I personally don't feel this step is essential enough to be part of the main steps (in this context, as the outcome variable does not represent an intervention that aims to encourage people to give up smoking).