2

I'm using SPSS for binary logistic regression. The result showed that one of the variables has a large beta coefficient and standard error.

Here are some results:

enter image description here

Can some explain to me what is the reason for that?

additional info: the variable that have large beta coefficient and standard error are actually part of the categorical of age (i.e. age were being divided into 4 categories (e.g. 20-25, 26-30 etc). So in this regression binary is being used for 3 of the categories and 1 become the reference.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Palmer
  • 61
  • 5
  • 2
    Are all your variables standardized? Otherwise we can't really say anything about the scale of the coefficient estimates without further details. – Firebug Nov 25 '16 at 10:47
  • 1
    You should post (much) more details, but also search this site for Hauck Donner. – kjetil b halvorsen Nov 25 '16 at 11:10
  • 2
    Or if you cannot understand the answers for Hauck Donner as @kjetilbhalvorsen suggested try searching for separation – mdewey Nov 25 '16 at 12:01
  • additional info: the variable that have large beta coefficient and standard error are actually part of the categorical of age (i.e. age were being divided into 4 categories (e.g. 20-25, 26-30 etc). So in this regression binary is being used for 3 of the categories and 1 become the reference. – Palmer Nov 25 '16 at 16:05
  • 1
    Have you followed up on either my suggestion or that of @kjetilbhalvorsen – mdewey Nov 25 '16 at 16:33
  • 1
    So show us a table of the response variable versus the age categories? – kjetil b halvorsen Nov 25 '16 at 16:36
  • Additional info should be added as an edit to the post, please! And, you should not divide age into categories keep it as continuous and maybe represent it with a spline so the effect do not need be linear. – kjetil b halvorsen Sep 07 '17 at 09:30
  • Possible duplicate of [Enormous coefficients in logistic regression - what does it mean and what to do?](https://stats.stackexchange.com/questions/48739/enormous-coefficients-in-logistic-regression-what-does-it-mean-and-what-to-do) – kjetil b halvorsen May 16 '19 at 09:09
  • I'm voting to close this question as off-topic because the OP has been asked for details and has not supplied them. – Peter Flom May 16 '19 at 12:03

0 Answers0