Check linear relationship of logit vs predictors in JASP
One of the assumptions of logistic regression is
"continuous independent variables, or predictors, have a linear relationship with the log-odds of the predicted probabilities for the outcome" - -https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8710907/
Is there a way to check this in JASP?
The way I'd like to is by plotting the empirical log-odds vs each independent variable. But getting the empirical log-odds as a new column in my data is proving tricky. Here's an example of what I'm trying to do in JASP: https://alexschell.github.io/emplogit.html.
Side question: why isn't this assumption mentioned in Statistical Analysis in JASP: A Guide for Students by Mark A Goss-Sampson?
Comments
Hi,
Unfortunately, the JASP does not contain assumption checks yet. I was planning to expand the module after the current release, so this is definitely something we will be looking into.
Cheers,
Frantisek
Thanks, Frantisek. Any suggestions for manually adding an empirical log-odds column to the data? I found it to be quite tricky.
I'm sorry but I don't have any suggestions in this regard.
Hi František,
Can I ask a bit more about what you're planning in this regard? The box-Tidwell is the first thing that comes to mind, what else were you thinking? I was recently discussing Hosmer-Lemeshow and its cousins with Johnny here: Improvement of assumption checks section in ANOVA procedure · Issue #490 · jasp-stats/jasp-issues (github.com) NB, I didn't start the thread and the title is a bit misleading
For what it's worth, I think the discussion contains too much in one place!
Best,
Tarandeep