1
$\begingroup$

I am conducting analysis to assess agreement between self-report and lab data on adherence to a certain drug intervention. I know that medication adherence in the population of interest can be influenced by variables such as age, sex, socioeconomic status, etc..

I need to conduct logistic regression to determine to what extent these variables explain the level of agreement between self-report data and lab data. What type of regression should I use to achieve this?

$\endgroup$
5
  • 2
    $\begingroup$ Maybe I'm missing something, but I don't understand the question. Can you give an example of what you call "types" of logistic regression? $\endgroup$ Commented Feb 3, 2019 at 0:09
  • 1
    $\begingroup$ I was only aware of one type of logistic regression myself $\endgroup$
    – Ingolifs
    Commented Feb 3, 2019 at 2:32
  • 1
    $\begingroup$ I think the OP means binary, ordinal or multinomial. $\endgroup$
    – Peter Flom
    Commented Feb 3, 2019 at 13:07
  • $\begingroup$ It is not clear what the dependent variable is and what kind of variable it is (binary, nominal, ordinal)? $\endgroup$ Commented Feb 3, 2019 at 13:19
  • $\begingroup$ Could you please clarify what "level of agreement" is and how you measure it? $\endgroup$
    – whuber
    Commented Feb 3, 2019 at 14:27

1 Answer 1

1
$\begingroup$

Assuming that by "type of logistic regression" you mean binary, ordinal or multinomial, it depends on the nature of your dependent variable.

If agreement for each person is a dichotomy - agree vs. not - then you want binary logistic. If agreement is ordinal - e.g. agree completely, agree somewhat, did not agree at all (or it could be a variation of this) - then ordinal logistic is a good start.

If agreement is a count (days per week) then you don't want logistic at all, but some sort of count regression. Poisson or negative binomial would be starting places.

And if agreement is measured some other way, then please clarify by editing your question.

$\endgroup$

Your Answer

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

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