 |
 |
Hi All! I am a baby in statistics and would please want some support in this my dillema. I know is may sound so simplistic to many of the elite who are on this forum. I have conducted a CRT involving over 50 clusters randomised into intervention and control. Within the cohort, I randomly selected 2800 sample from the original 8000 and over population within ONLY the intervention group for a certain marker which could be A or B. 10% of the 2800 tested positive for the marker (either A or B) and over 65% of these are positive for marker B. In my analysis, i am seeking to find what factors are associated with testing positive for the marker B. I was thinking of using random effects logistic regression models but not sure whether this is appropriate because of the small outcome size and the problems with logistic regression random effects models. Would I be justified in using LR-RE models and what is the justification?. Are the some particular problems i might encounter especially when i would want to even split the data and check association with wealth quintiles? Any support and suggestions about potential problems will be most welcome. Thanks. You may leave the list at any time by sending the command SIGNOFF allstat to [log in to unmask], leaving the subject line blank.