0
$\begingroup$

I have data on Liver Disease Patients, with the response variable being whether the patient has liver disease or not(1 or 0) and 7 predictor variables, such as age, gender, protein levels etc.

Now, the purpose of my study is to first conduct binary logistic regression on the entire data and check for the significant traits. Next, I need to split the entire data with respect to age(the two groups being <=40 and >40) and fit binary logistic regression on both groups. Next, I need to check if the co-efficients from the regression equations of these two groups are same or not. How do I do this? I'm allowed to use the softwares R and Minitab for this purpose, by my curriculum. I'd be glad if anyone could suggest some ideas.

$\endgroup$
1
  • $\begingroup$ It would be much more instructive for you if you tried to answer some part of your question and then edited your question telling us what you tried, how far you got and what you do not understand. Also please clarify if you are dichotomising age because you have been told to by superior beings since it is a dumb idea and we would suggest better ones if free to do so. $\endgroup$
    – mdewey
    Mar 19, 2017 at 14:24

1 Answer 1

3
$\begingroup$

Before using logistic regression make sure you intensively study the subject. You started the analysis on the wrong foot. A few things to study:

  • keeping continuous variables continuous
  • how to specify interaction terms in the model (and not split the data and fit two models)
  • how to interpret interactions
  • how to determine if $Y$ is truly binary or represents an oversimplification of liver disease severity
  • how to compute the sample size needed for reliable modeling

These are covered in my course notes and book - see http://biostat.mc.vanderbilt.edu/rms

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.