univariable logistic regression

Discuss statistics related things

by anna » Fri Oct 05, 2018 8:08 pm

I would like to perform univariable logistic regression for several variables independently of each other with a binary outcome. So far, I have only been able to do this in jamovi by analyzing each variable separately, which takes a lot of time. If I use the drag and drop menu and drop in multiple variables the resulting analysis is a multivariable regression. I would also like to not only give the OR and the CI but be able to show the n of each level of each variable (e.g. n patients with medication A, n patients with medication B). Is there a way to do this?
Posts: 2
Joined: Fri Oct 05, 2018 7:57 pm

by jonathon » Fri Oct 05, 2018 10:43 pm

hi anna,

yeah, there's not an easy way to analyse a lot of variable in the way you describe. how many variables do you typically analyse in this fashion?

i've been meaning to add support for 'duplicating' an analysis, and then you can duplicate an analysis a few times, and then just change the parts you need.


User avatar
Posts: 815
Joined: Fri Jan 27, 2017 10:04 am

by Ravi » Fri Oct 05, 2018 11:23 pm

About your second question, is it a classification table that you're looking for (in the "Prediction" menu)?
User avatar
Posts: 140
Joined: Sat Jan 28, 2017 11:18 am

by anna » Sat Oct 06, 2018 4:36 pm

Hi jonathon and ravi,

thank you for your quick reply. I usually need up to 50 variables, which makes it a lot of individual analyses and a lot of copy pasting. Also I would like to specify what I meant: e.g. I want to analyze the risk factors for a specific binomial outcome. The risk factors being smoking, non smoking, ex smoking etc. and then 49 other potential risk factors. I would like to be able to give the odds ratios and p values for each of these variables (e.g. in the case of smoking always compared to non smoking individuals and so on). I would also like to be able to give the number of smokers, non smokers, ex smokers... to be able to interpret the data quality the regression is based on. I understand the number part is quite difficult? The most important bit for me would be to have one table summarizing the odds ratios, p values etc for all potential risk factors.
Best, Anna
Posts: 2
Joined: Fri Oct 05, 2018 7:57 pm

Return to Statistics