-logit- and -logistic-, what’s the difference


Both -logit- and -logistic- are used to estimate binary logistic regression models. Thier difference lies in the reports that they display — -logit- reports coefficients, while -logistic- reports odds ratios. Since the odd ratios can be computed (in terms of the coefficient, b) as e^b, the choice between the two is just a matter of preference. To illustrate, I have posted below the results of  -logit- and -logistic- using womenwk.dta, which is used in Christopher Baum’s An Introduction to Modern Econometrics Using Stata.


6 Responses

  1. Thank you! I needed a simple explanation for this question!!

  2. You could also use the “glm” command…

    glm work married, family(binomial) link(logit)

    and add the option “eform” on the end to give the odds ratios. Not adding this option will give the coefficients. This command also lets you fit different types of models (eg. poisson models with a log link)

  3. Hi your website is very informative and interesting. Just found it today and i’ve already learned a lot. By the way, the logistic command also has an option ‘coef’ to display coefficients while the logit command also has an ‘or’ option to display the odds ratios. : )

  4. Thank Mitch for your reply. I know what you mean but since I have a lot of variables for each country and several countries in my data set, I cannot do like that for that will only allow means for 1 variable to be tested at one time. In addition, if I choose the simple t test, I have to repeat 6 times for each single country since there are 6 pairs to be compared. My supervisor also told me that the most sensible way is to use double loops i.e. one for the 3 variables and one for the 4 groups. As I am new to programming in Stata, I cannot figure out how double loops work. I also try to send the question to Statalist but the question has been always bounced. Do you have any idea?

  5. Dear Mitch,

    I am having a problem with double loops and hence would be much grateful if you could give me a hand on that. The whole story is that I have a data set will all firms for a particular country during the 1980-2007 period. Let’s suppose that each firm has 5 variables – profitability, growth, size, deviations from target leverage (either above-target leverage (A) or below-target leverage (B)), and financing gaps (either financing surplus (FS) or financing deficit (FD)) (actually I have a lot more variables).

    Now I want to calculate the mean for each variable – profitability, growth, and size for each of the 4 groups – A & FS (1), A & FD (2), B & FS (3), and B & FD (4) using the summarize command in Stata. Next I want to do t tests to see whether the means for each variable are statistically different between each pair – (1) & (2), (1) & (3), (1) & (4), (2) & (3), (2) & (4), and (3) & (4). I know that I have to use double loop in this case but do not know how it works since I am rather new to programming in Stata.

    Thank you very much for your help.

    • Hi Cuong,

      If you only want to test for the means, you do not have to calculate them explicitly. Check the -ttest- command, “help ttset”. This might be what you need:

      Two-group mean-comparison test: ttest varname [if] [in] , by(groupvar) [options1]

      Also, I would suggest that you join Satalist, where you will have better chances of getting the best answers. You would be surprise how generous the people are in giving advice. See my post on Statalist, Stuck? Hello Statalist.

      Hope this helps.

      Best,
      Mitch

Leave a Reply