Difference in Difference Regression in Stata using "diff" command by Samsun 

Thank you so much for watching this video. We are glad that you found it helpful. Please like this video. Share with your friends. Subscribe to our channel! Please stay tuned!

Thank you so much. Good question, I can not figure out how we can use the fixed effect DID model using the diff command yet. I have another video on the 'reg" command: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-qu1uPVxC5_o.html. using reg command you can easily run a fixed effect DID model. For example, if you want to run state fixed effect, then use i.state after reg (depvar/s) (covariate/s). I will make a video very soon on the comparison between diff and reg, why which one is more preferable than another. Please stay tuned.

@@Learn-Coding-with-Us Thanks for your reply! I have been trying to specify fixed effect diff-in-diff model for my PhD thesis. It would also be helpful if you explain how to use time effects and individual/cluster fixed effects by showing both the regression specification and STATA code. Waiting to see soon.

@@mdhasibreza5161 I will upload two videos on the comparison and fixed-effect model by this week. Could you please subscribe to our channel? It will be helpful for you to receive a notification once the videos will be uploaded.

Hi, please find the video on fixed-effect diff-in-diff using reg command: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-N_7nnhqpPbU.html

Hi Maykon, thanks for watching my video and comment on it. To my knowledge, the only difference is xtreg is used for panel data and you can use diff command for repeated cross-sectional data. This pdf might be helpful for panel data and xtreg command for difference-in - differences models. people.ucsc.edu/~aspearot/Econ113F14/Lecture%2010%20F14%20A.pdf

@@Learn-Coding-with-Us Thank you! I have another question: the dummy variable identifying individual i was ommitted of my model by collinearity with interaction dummy. How to proceed? What kind of problem the omission of the variable can cause?

@@maykondouglas2294 Hi, Can you clarify what kind of dummy variable it is? What is the other variable you are using for the interaction? I have an answer for you.

@@Learn-Coding-with-Us Yes. I'm using three dummies: tempo (0 for pre-intervention and 1 for post-intervention), bndes (0 for no affected by the intervention and 1 for affected) and dummydiff (tempo*bndes). Dou you believe bndes was built incorrectly? If you prefer, we can discuss these points by email

@@maykondouglas2294 , you are totally fine to discuss here. Between tempo and bndes, tempo is your time dummy variable, right? and bndes is your treatment dummy variable, right? Time variable means you have some time period. Within this time period, there is an intervention happens at some point. Then you should code 0 is equal to the pre-intervention time period and 1 is the post-intervention time period. And treatment variable means some people are being treated (=1), and some not(=0). I am suspecting that you may code both tempo and bndes from the same variable. It leaves you collinearlity problem. But, you have to have two different variables. From these two variables, you have to code one for the time dummy variable and another for the treatment dummy variable. Are you using diff or reg or xtreg command? By the way, could you please subscribe to our channel?

Hi, please use the example dataset from the STATA. You can also use your own data. I can not make my data publicly available. I just showed how I have used my data to run diff-in-diff regression to make you guys confident. I am so sorry about that.