Adding variables to the model one by one, or at the same time What are the benefits of adding the variables into a model one by one, as compared to adding them all at the same time?
As I see in most research, the first model that is tested is consisted of all the control variables, then in the second model the first predictor is added, then the second one, and so on, until all predictors are in the model.
Are there any benefits in using this approach, or would it be ok to have just do two models - one with just the controls, and one with controls + all other predictors. 
(p.s.) - am interested in this with regards to logistic regression
 A: This is a habit that stems from linear regression, where this sequence of models can be interpreted as ways of estimating indirect effects. Unfortunately this trick does not generalize to non-linear models, like logistic regression (e.g. Buis 2010). Still the habit persists, as you have observed. Unless you are interested in direct an indirect effects, I would just avoid all the complexities involved with comparing non-linear models and just report one model. If direct and indirect effects are of interest then you can choose between different methods, some references are given below:
M.L. Buis (2010). Direct and indirect effects in a logit model. The Stata Journal, 10(1), pp. 11-29.
http://www.stata-journal.com/article.html?article=st0182 (free)
Hicks, R. and D. Tingley (2011). Causal mediation analysis. The Stata Journal 11(4), 605-619.
http://www.stata-journal.com/article.html?article=st0243
Kohler, U., K. B. Karlson, and A. Holm (2011). Comparing coefficients of nested nonlinear probability models. The Stata Journal 11(3), 420-438. 
http://www.stata-journal.com/article.html?article=st0236 
Sinning, M., M. Hahn, and T. K. Bauer (2008). The Blinder-Oaxaca decomposition for nonlinear regression models. The Stata Journal 8(4), 480-492.
http://www.stata-journal.com/article.html?article=st0152 (free)
