I am working on two dependent outcomes, the first dependent outcome is 2 categories(yes/no).
The second dependent outcome is 4 categories. I want to perform statistical analysis to see the relation between each outcome and a number of independent variables( continuous, categorical, binomial) in the following scenarios:
1- the first outcome (2 categories) with each independent variable.
2-the first outcome with all the significant variables from step 1.
3-the second outcome (4 categories) with each independent variable.
4- the second outcome with all the significant variables in step 3.
Which is the suitable statistical test for each scenario?
For the 2 outcome DV you want logistic regression(binary)
For the 4 outcome DV you want multinomial logistic regression.
It matters a lot whether you want to predict or explain something in what you are doing. The methodology is quite different. Here are some possible approaches:
https://www.google.com/search?client=firefox-b-1-d&q=Difference+between+statistical+explanatory+a+predictive+models
https://www.google.com/search?client=firefox-b-1-d&q=mutinomial+logistic+regression+adaptive+lasso+r
Best wishes, David Booth
I agree with David Eugene Booth above, you can do that easily in R using StatsNotebook for R, it's a free download at the moment.
Thank you David Eugene for your reply. I'm doing the test to find if there is a relation between some patient's characterstics (IVs) and the treatment response (DV). I've read that I should use Chisquare test if I have 1 categorical DV and 1 categorical IV. So, I'd like to know in which scenarios I should use Chisquare or Logistic Regression teste.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis