If you must use logit regression, dichotomize your dependent variable!

This doesn't sound like you need logit. If the DV is a scale variable, and the predictors are categorical, you can analyze the data with an ANOVA-like technique, or linear regression on factors. 

As Andres said above, if you MUST, for some reason, use logit, you need to dichotomize the outcome variable, perhaps using a mean- or median-split. This, however, throws away a tremendous amount of variability, and therefore information, so it should generally be avoided. 

@Paul Yarnold. Even if I incur in the risk of sounding naive, for what does CTA stand? I cannot find the full term anywhere. Thanks.

Ah, I see. Thanks.