I appreciate how challenging giving a lecture can be without a visual audience, but you nailed this! It’s obvious that you know the model’s requirements / applications, and you explained the a fairly complicated model in such an easy to understand way. Very impressive lecture!
Thank you for the kind words - glad it was useful! There is a lot more to the DiD model and literature, of course, but hopefully this is a good introduction to the concept.
Thankyou so much for your explanation from the data it was so clear, I thought DiD method was complicated but you just simplified it easily. I really appreciate this video and the lecturer.
Thank you Mike, I have learnt a lot from this lecture. I thought DiD was complicated and you have not just simplified it for me alone but also gave me confidence for graduate school thesis.
Can I just say, in the past 4 years I managed to finish my bachelors and masters in finance and economics all thanks to you, I feel like u deserve some % of my salary. XD!
It is really impressive how you are able to explain those topics in a straight forward und understandable manner. Looking at your videos and the effort you put in makes me wonder how its justified that my professor is getting such a high pay for simply reading down some slides. Thanks a lot for your very appreciated help!
Thanks a lot for your lovely explanation. My questions 1- in real life experiment (quasi) comparison treatment group to control group is hardly feasible and in most situations is unethically leave second group without treatment, so can we make this quasi-experiment to compare between 2 treatment groups, meaning treatment 1 group versus treatment 2 group (comparison group) ? Taking in our consideration that pre-treatments level in both groups comparable (no difference in the starting point) as possible as other factors except in treatment allocation to eliminate bias selection. Question 2- Evaluating pre- and post-treatment in each group threatened the external validity as we did not compare it to control, meaning we can not figure out the effect of treatment A or treatment B, noting that the internal validity is not violated as we can compare the differences between both groups. Question 3- Regarding the statistics, if the normality is violated in some variables and the average can not be evaluated , using the median is it fine to figure out the significant differences or not
Thank you very much for explaining the DID. I kindly request you to explain how to conduct a placebo experiment to support the parallel trend assumption of DID. Thank you very much.
Thanks so much. You really take the time to explain it in detail. I wonder how to conduct this test on longitudinal data though. Do you have a video about that? And I wonder how to use the didregress or xtdidregress function in Stata and what the difference is between the two.
It’s very impressive to listen to your lectures! Would you please share more information about DDD(triple difference model)? Especially how shall we do the robustness test
Thanks! My understanding is that if there are more than two time points, then, DID has to be done separately for each pair of pre-test and a subsequent post-test, right?
Dear Mike thank you for posting such a useful and beatifully explained video. one question, why did you put all data into Stata? was it an alternative way of estimating a treatment effect?
Thank you for the extremely easy to understand video.. i have a data in which the control and treatment are sequencial.. ie half hour for control, next half hour for treatment, next half hour for control and so on until there are 4 control and 4 treatment groups.. also the participants were not the same. Do you think the D in D can be applied to it? Your guidance will be much appreciated.
Hello Sir, thank you very much for such a great lecture. what if to consider two or more T periods before the treatment. let's assume that they are t0 , t1, t2 etc from left to right. In this case, which t-value should be included in the formulat? the last one or t0? I hope I could explain my question. thank you in advance
I appreciate how challenging giving a lecture can be without a visual audience, but you nailed this! It’s obvious that you know the model’s requirements / applications, and you explained the a fairly complicated model in such an easy to understand way. Very impressive lecture!
Thank you for the kind words - glad it was useful! There is a lot more to the DiD model and literature, of course, but hopefully this is a good introduction to the concept.
Thank you, professor. Your video is very straightforward and clear. Keep positing more lectures of DiD!
Thankyou so much for your explanation from the data it was so clear, I thought DiD method was complicated but you just simplified it easily. I really appreciate this video and the lecturer.
This is perhaps the best lecture on DiD. Thank you so much.
👆 This comment nailed it!
Thank you Mike, I have learnt a lot from this lecture. I thought DiD was complicated and you have not just simplified it for me alone but also gave me confidence for graduate school thesis.
Can I just say, in the past 4 years I managed to finish my bachelors and masters in finance and economics all thanks to you, I feel like u deserve some % of my salary. XD!
Haha - thank you Omar! Very glad to be helpful.
It is really impressive how you are able to explain those topics in a straight forward und understandable manner. Looking at your videos and the effort you put in makes me wonder how its justified that my professor is getting such a high pay for simply reading down some slides. Thanks a lot for your very appreciated help!
Great video! Very easy to understand! Please do another one on when we need to consider other control variables!
This has been so helpful!!! Thanks so much. I had struggled to understand the concept. But this lecture has made it so easy
Glad it helped!
Thank you for this, makes it easier to understand the concept.
This is a great video for a first-time exposure to D-I-D.
Thanks a lot for your lovely explanation. My questions 1- in real life experiment (quasi) comparison treatment group to control group is hardly feasible and in most situations is unethically leave second group without treatment, so can we make this quasi-experiment to compare between 2 treatment groups, meaning treatment 1 group versus treatment 2 group (comparison group) ? Taking in our consideration that pre-treatments level in both groups comparable (no difference in the starting point) as possible as other factors except in treatment allocation to eliminate bias selection. Question 2- Evaluating pre- and post-treatment in each group threatened the external validity as we did not compare it to control, meaning we can not figure out the effect of treatment A or treatment B, noting that the internal validity is not violated as we can compare the differences between both groups. Question 3- Regarding the statistics, if the normality is violated in some variables and the average can not be evaluated , using the median is it fine to figure out the significant differences or not
Man what a great teacher, Fantastic.
Your explanation is so clear! Thank you so much!
Glad it was helpful - thank you!
Thank you so much. Great Video !!!
Amazing video! Thanks for creating such valuable content
Very informative lecture. It made did very easy to understand. Thank you!
Thats a great video Mike!
Thank you very much for explaining the DID. I kindly request you to explain how to conduct a placebo experiment to support the parallel trend assumption of DID. Thank you very much.
Thank you.
THANK YOU SO MUCH FOR THIS!!
Thank you!
Well done... very helpful explanation! Thank you.
Thanks so much. You really take the time to explain it in detail. I wonder how to conduct this test on longitudinal data though. Do you have a video about that? And I wonder how to use the didregress or xtdidregress function in Stata and what the difference is between the two.
thanks
Thank you. It was very impressive lecture!
This is really nice
It’s very impressive to listen to your lectures! Would you please share more information about DDD(triple difference model)? Especially how shall we do the robustness test
Thanks! My understanding is that if there are more than two time points, then, DID has to be done separately for each pair of pre-test and a subsequent post-test, right?
very helpful! thank you
Dear Mike thank you for posting such a useful and beatifully explained video. one question, why did you put all data into Stata? was it an alternative way of estimating a treatment effect?
Thanks for sharing the information, it was very useful. I have a question, how relevant is the R2 value in this impact evaluation method?
Thank you for the extremely easy to understand video.. i have a data in which the control and treatment are sequencial.. ie half hour for control, next half hour for treatment, next half hour for control and so on until there are 4 control and 4 treatment groups.. also the participants were not the same. Do you think the D in D can be applied to it? Your guidance will be much appreciated.
Thanks! helpful
can we have more than two pre-treatment and post-treatment periods
can you attach the sample workbook you use please
Hello Sir, thank you very much for such a great lecture. what if to consider two or more T periods before the treatment. let's assume that they are t0 , t1, t2 etc from left to right. In this case, which t-value should be included in the formulat? the last one or t0? I hope I could explain my question. thank you in advance
Thanks for the video. Can you do another video using SPSS or R
Yes, I will work on an example in R soon. Thanks!
@@mikejonaseconometrics1886 Ok thanks, looking forward to it!
Shouldn't the slope change starting from the point of treatment? (13:26)
I see - this is for two time points only - fair enough
Really liked the video. Can you please share the referencees as mentioned by you?
Yes, thanks for the reminder! I've linked to a great new book, and an old article worth reading.