Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# Re: st: Question: Ordered logit- suspicious odds ratio

 From brendan.halpin@ul.ie (Brendan Halpin) To statalist@hsphsun2.harvard.edu Subject Re: st: Question: Ordered logit- suspicious odds ratio Date Mon, 16 Apr 2012 14:25:42 +0100

```On Mon, Apr 16 2012, Stefanie Kneer wrote:

> and in particular whether there exists a social norm effect. P11101 is
> thus my dependent variable and as it is discrete I was planning to run
> an ordered logit but I am encountering difficulties when trying to
> calculate the odds ratio. As can be seen the social norm effect, and
> thus whether you feel more comfortable when other people around you
> are not working, too is massively big (27). That looks kind of
> suspicious to me.  Could you give me an advice on why this is the case
> and what I should do about it?

> p11101       Coef.   Std. Err.      z     P>z     [95% Conf.      Interval]
>
> regionunempl  -2.432835   .2575799    -9.44     0.000    -2.937682  -1.927988
> unempl_empl   -1.553812   .0582462   -26.68     0.000    -1.667972  -1.439651
> interaction    3.328277   .5334905     6.24     0.000     2.282655  4.3739

I don't think the big OR is an issue. Think of it as the effect of
regionunempl being -2.432835 for those not unemployed and -2.432835 +
3.328277 = 0.895442 for those who are (a lot then depends on the range
of regionunempl).

You can't really interpret interactions independently of their
lower-order elements.

A second remark would be that with so many values in your dependent
variable, linear regression would likely work very well. At the very
least you should use it as a sanity check on your ologit results.

Regards,

Brendan

--
Brendan Halpin,   Department of Sociology,   University of Limerick,   Ireland
Tel: w +353-61-213147  f +353-61-202569  h +353-61-338562;  Room F1-009 x 3147