Re: odds ratio

Posted by Bruce Weaver on
URL: http://spssx-discussion.165.s1.nabble.com/odds-ratio-tp5724474p5724505.html

If you want CROSSTABS to give the same OR as LOGISTIC REGRESSION, use a temporary RECODE, as in the example shown here:

http://spssx-discussion.1045642.n5.nabble.com/compute-logit-and-compare-regression-coef-linear-and-logistic-tp5724493p5724496.html

PRogman wrote

When computing odds ratio (OR) using then Risk option in Crosstabs SPSS always sorts the variable values in ascending order.  If your variables are coded 0|1 you get the OR of the 0 value, it is the same OR as it is symmetric:
OR defined as = ad/bc

         1      |    0
      -----------------
  1 |  11 (a) |  10 (b)
  0 |  01 (c) |  00 (d)

and SPSS sorts this to

       0  |   1
      ---------
  0 |  00 |  01
  1 |  10 |  11

and still calculates OR as the first table...
Use binary linlog to get the results you want including confidence intervals, especially if you are coding variables 0|1.
(or use 0=yes and 1=no ?????)

I wish there was an option in Crosstabs (like in LogReg) to specify the reference category.

/PR