0

I have a binary variable that I'm investigating in SPSS, inclination_to_dance. I have another linear variable as well, no_of_beers.

In SPSS, via the Analyze menu, I'm able to run a binary logistic regression against these two that gives me some kind of output that tells me whether the impact of drinking on dancing is significant, as well as its coefficient.

What I really want to know, though, is what the effect of drinking one additional beer will have on the probability that one will dance. Is there a way to investigate that probability in SPSS?

fox
  • 123
  • 3
  • 2
    Of course, there is a way. It seems you're new to logistic regression. In the table of coefficients, pay attention to **exp(B)**. You can open the table and right-click on the field, to ask "What's that?". To learn quickly the basics of logistic regression, go to Case Studies section in Help and find Logistic Regression there. Good luck! – ttnphns Apr 30 '13 at 07:00
  • ah ok, so is it correct to say that an Exp(B) of .986 associated with the `no_of_beers` variable means that each beer increases the probability of dancing by ~1%? – fox Apr 30 '13 at 07:06
  • 3
    Exp(B) of .986 says that the estimated _odds_ `Prob(dancing)/Prob(nodancing)` increases by .986 times (i.e. slightly decreases) in response to taking one more beer. – ttnphns Apr 30 '13 at 07:18
  • counterintuitive, but I'll take it! – fox Apr 30 '13 at 07:30
  • 1
    Please be aware that logistic regression assumes that the dependency (of log of the odds) on the independent var is linear. This might not be your case (e.g. first two beers provoke for dancing while next beers have oppisite effect). You can test this assumption in Logistic regression, although you'll have to work. – ttnphns Apr 30 '13 at 07:31
  • @ttnphns, please make your comments into an answer, they are a good enough solution to the problem. This is IMO borderline on topic, but I voted to leave open because there is a statistical question here about how to go from the usual logit model output to predicted probabilities. – Andy W Apr 30 '13 at 11:48
  • @ttnphns is right, what you need is to understand the basics of logistic regression. It may help you to read my answers here: [interpretation of simple predictions to odds ratios in logistic regression](http://stats.stackexchange.com/questions/34636//34638#34638), & here: [difference-between-logit-and-probit-models](http://stats.stackexchange.com/questions/20523//30909#30909), which together will provide some info on what LR is & how to interpret it. – gung - Reinstate Monica Apr 30 '13 at 14:14

0 Answers0