Conditional logistic regression with odds ratio

[#1]

Is it possible to run conditional logistic regression analysis with either 1-1 match or 1-M match in JMP or JMP clinical, with possibility to calculate odds ratio?

I think in SAS it is easy to program, but how can we achieve it in JMP?

Thanks!

[#2]

Hi @PL ,

What numeric data/variable already in the clinical data set are you using for the regressor? Also, what would be the categorical factor you would be using for stratification and for the response?

This will help us guide you where to go within JMP Clinical (either use a built in JMP Clinical report or to use the JMP portion of JMP Clinical).

Best,

[#3]

Hi, I attach a link here from UCLA for running conditional logit model (in SAS). The data format is very familiar with our data.

https://stats.idre.ucla.edu/sas/faq/how-do-i-do-a-conditional-logit-model-analysis-in-sas-9-1/

Could I replicate this analysis in JMP or JMP clinical?

We are assessing the possibility of purchasing JMP (or JMP clinical). So I am doing this to see if we can run the same analyses for one of our studies using JMP.

Thanks!

[#4]

Hi @PL ,

Thanks for showing an example of your data. JMP Clinical assumes a different data structure and model with demography data separated from adverse events, laboratory data, con meds, etc. Much like the CDISC data structure for SDTM or ADaM. If your data is not structured in a similar way (or put into a similar structure), then JMP Clinical would not be of any use. Also, it does not have any reports where conditional logistic regression would be applied.

As for JMP or JMP Pro, it is possible using a Choice Model.  My colleague @jiancao assisted me with this answer (he really is the one who answered it for me and I am passing it along).  Attached are two example with the results saved as JMP Table Scripts in the tables section of the JMP Table. One is 1-1 matching and the other is 1-M matching. Jian said the trick is to uncheck the default Firth Bias-adjusted Estimates option in JMP in order to match SAS output. SAS reports AIC, Wald tests while JMP shows AICc  and LR tests, but that the key outputs, such as -2LL and coeff. estimates and std. errors, do match.

Hope that helps.