Dear Bezon,

It is unfortunate that up until now I have not heard anyone doing marginal effects on SPSS successfully. I would recommend using Stata instead and which is more presentable, less crowded view and easier to report marginal effects too. Although Stata is mostly command based, there are versions with GUI based views. 

Once you import your data on stata, just write "logit [dep.var] [indep. vars]" and enter. After that, run "mfx" and there you've got the marginal effects.

Hi Bezon.  I assume you mean a binary logistic regression model, and that you are fitting it with the LOGISTIC REGRESSION procedure in SPSS.  If you fit your model via NOMREG instead, you can use the TEST sub-command to generate fitted values at any desired combination of explanatory variables, and you can also compute differences between two fitted values, etc.  The output for each TEST sub-command will include a SE, 95% CI, and p-value, IIRC.  (I don't have SPSS on this machine, so can't provide a worked example right now.)

Alternatively, you could use the !MLEcomp macro I wrote.  It is described in the article linked below, and the syntax files and demos can be found at the second link.

HTH.

http://www.ncbi.nlm.nih.gov/pubmed/22610390

https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/olscomp_mlecomp

Dear Bezon,

It is unfortunate that up until now I have not heard anyone doing marginal effects on SPSS successfully. I would recommend using Stata instead and which is more presentable, less crowded view and easier to report marginal effects too. Although Stata is mostly command based, there are versions with GUI based views. 

Once you import your data on stata, just write "logit [dep.var] [indep. vars]" and enter. After that, run "mfx" and there you've got the marginal effects.

I agree with Muthoka - it is very common (even in quality papers) to see incorrect interpretation of individual predictor influence. Very often exp(regression coefficient) is believed to provide elasticity - but for example with Poisson models (used with count data) this is not correct. "Statistical and Econometric Methods for Transportation Data Analysis, Second Edition" (Washington et al., 2010) is one exception, where the authors describe in detail how it should be done (and they also explain the difference between elasticity and marginal effect, which is often overlooked, or even interchanged).

Can you kindly tell me that how can we calculate elasticity of Dependant variable in a binary logistic model by using SPSS ?

I will have an idea soon. Still learning the stuff myself.