0
$\begingroup$

I want to calculate the crude and adjusted odds ratios for exposure to occupational risk factors such as aluminum and fossil fuels in my case control study. My cases are 180 demented patients and I have 370 controls. Which type of logistic regression model should I use? When I adjust for age and education the odds ratios get bigger, isn't that wrong?

$\endgroup$
0
$\begingroup$

The type of model you should use depends on the way the dependent variable (DV) is measured. It appears that your DV is dichotomous (demented/controls) which would indicate "regular" logistic regression.

It is not necessarily wrong that the odds ratios (ORs) increase when you control for demographics. The change in the ORs when adding control variables depends on the relationships among the variables. They can go up, go down or stay almost the same.

$\endgroup$
  • $\begingroup$ Thank you! another thing is that when i enter the risk factors which are also binary, should i enter all of them simultaneously in the model or just one risk factor in a time, then when i want to adjust fo age, sex and education should i enter all the confounding factors at the same time or they should be intered seperately? $\endgroup$ – s nadery Dec 29 '18 at 14:25
  • $\begingroup$ In the model summary for logistic regression the Cox& Snel R square and also Nagelkerke R squares get zero, what does that mean? $\endgroup$ – s nadery Dec 29 '18 at 14:27
  • $\begingroup$ How to add variables depends on what you are trying to find out. R^2 of 0 is a bad sign, either you coded something wrong or the model is very poor. $\endgroup$ – Peter Flom Dec 29 '18 at 17:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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