I am working on a data set where disease related information exists for 600 patients while demographic data is available for 300. I want to run multivariable logistic regression to see associated factors for disease variables and adjust it for demographic variables. How to do it with missing data?
Suri Shivali Of course, the easiest solution would be "listwise deletion", which would not create any kind of bias if data is missing at random. This would also mean that you would lose quite a lot of information.
Check out the available literature on "data imputation". This is the process of replacing missing values. Of course, this is based on different kinds of algorithms. In your case it sounds that you simply do not have the demographic data and this might be hard or impossible to impute.
Please go through the article below:
https://www.researchgate.net/post/How_can_I_fill_the_missing_climatological_data
You can impute 15% missing data at the most.
Hi Suri Shivali ,
follow the suggestion of prof. Horst Treiblmaier . An example is attached.
Anyway, I agree that " In your case it sounds that you simply do not have the demographic data and this might be hard or impossible to impute".
Is there a chance for you to analyze the data with Bayesian statistics?
- Badges
- Science topic