4
$\begingroup$

I have a question regarding quasi-complete separated data.

One example of quasi-complete separation is a dataset, where all x < 2 have y=1, all x > 2 have y=0 and some x=2 have y=0 and some x=2 have y=1.

I have a dataset with kind of the following situation: for all x < 2 y=0, for all x > 5 y=1. In the rage x=2 to x=5 y is sometimes zero and sometimes one.

Is this a case of quasi-complete separation as well?

Does is matter for logistic regression if the range of x=2 to x=5, where y varies, is small?

An example how my data look like:

enter image description here

$\endgroup$
1
1
$\begingroup$

Your textual description ("for all x < 2 y=0, for all x > 5 y=1. In the rage x=2 to x=5 y is sometimes zero and sometimes one") does not match your figure. All data displayed in the figure have x<2, and many have y=1. Nonetheless, neither your textual description nor the figure have separation or quasi separation. For both your description and the figure, the data where y=0 and where y=1 overlap. Logistic regression will have no problems with these data.

$\endgroup$

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.