Least Squares Dummy Variables estimators
ฝัง
- เผยแพร่เมื่อ 7 ต.ค. 2024
- This video explains the motivation and mathematics of Least Squares Dummy Variables estimators, and indicates their equivalence with Fixed Effects estimators.
Check out oxbridge-tutor.... for course materials, and information regarding updates on each of the courses. Check out ben-lambert.co... for course materials, and information regarding updates on each of the courses. Quite excitingly (for me at least), I am about to publish a whole series of new videos on Bayesian statistics on youtube. See here for information: ben-lambert.co... Accompanying this series, there will be a book: www.amazon.co....
![CAMPปลิ้น | EP.82[2/2] เมาท์มอยกับ 3 สาวเพื่อนซี้ ที่พร้อมขยี้พ่อหมีได้ทุกเมื่อ](http://i.ytimg.com/vi/Wwn7WnylNtQ/mqdefault.jpg)
You are the best. Econometrics seems to be one of the most obfuscated subjects I've ever studied, but (as your videos prove), it's really not that bad. Don't really get why other attempts at explanation make it so hard, or why its so rare to find good explanations. Regardless, thank you very much for your series!
Hi Kian,
Thanks for your message and kind words.
Yes you are correct; I normally include an intercept by default so that's the reasoning behind my statement.
Best,
Ben
Thanks a lot Ben. All your lectures are helping me out in my research! One que. Is this applicable even in the case of dynamic panel estimation?
So the conditions for consistency of LSDV dummy estimators differ from those of FE estimators? Because for them, homoskedasticity and no SC weren't needed.
Hi Ben
Please could you clarify something. At 2:00 you said "... technically the number of cities minus one because we don't need to include dummy variables for each of the cities".
It is my understanding that this is only required if we have an intercept in the model, since that would cause the model matrix to be column rank deficient. Would it be correct to say that you need to either:
(a) add an intercept term
or
(b) include all three levels of the indicator variable?
Loving the videos, they are all fantastic and very much appreciated by academics of many disciplines.
Good question. In the usual case we always include an intercept. I am confident the intercept is in this sense not related to our issue. We get a singular matrix if we include n dummies because the model simply cannot be estimated without a baseline. With this regard the initial intercept is irrelevant because it just adjusts our starting value if all x are zero.
Well, thank you for video. I would like to ask the missing part from my understanding. As we take all time invariant features in apha i and i symbolizes the different cities, ist that also the same thing with dummy variable estimator? In dummy variable estimator each city has their own value, yes. But in the first equation, there are also each city ( as i) for each time invariant feature. What is the difference?
Thanks
Great vid, just one thing that was a little unclear to me.
What excactly is μ in this method? Saw an example where it was used the city-specific alpha as wheight for the dummies. Is this the same?
Thank you. The video is easy to understand.
Hello Professor. Sorry for disturbing you. I am undergraduate student from Kazakhstan and I am working on my thesis. I want to estimate how family structure, number of siblings, health, teaching practices and many other variables affect the educational outcomes of school students. Each year in our country we have national tests and I used them as a measurement of educational outcome in my OLS regression, and also I used the 'receiving the scholarship' as a categorical dependent variable. I surveyed the 5 cohorts of student graduates, also I have information about the regions where they studies. So I decided to use Least Squares Dummy Variables in my Logit and OLS regressions? Can I do so? or it is possible only for panel data? My data is cross sectional since people from each cohort wrote national test only once, however I have variable 'year when they wrote the test' and 'regions'. Can I do xi: logit scholarship encourageteachers excelhealthdummy goodhealthdummy fairhealthdummy nsiblings punishteachers highclassdummy middleclassdummy singlemomdummy singlefatherdummy blendedfamdummy female i.testyear i.REGIONS and xi: regress testscore encourageteachers female city excelhealthdummy goodhealthdummy fairhealthdummy nsiblings punishteachers highclassdummy middleclassdummy singlemomdummy singlefatherdummy blendedfamdummy femalteachencourage singlemomencourageteach excellenthealthencourageteach i.testyear i.REGIONS commands in STATA and if I can how I should write the equations of these Least Square Dummy Variables regressions?
I’m a bit confused because I think the estimate for the dummy variables obtained from the dv method is for (alpha_i - alpha_{base level}). Is this correct?
Thank you Sir for your useful videos. I'd want to ask if d2 and d3 are still correlated to unemployment and criminality. I'm asking this question because the resort to LSDV is precisely motivated by the possible correlation between α (alpha) and the predictors (unemployment and criminality), am I right? Thank you in advance Sir.
That is not a problem for the estimates. They will only be inconsistent if correlated with the not specified error
Hello Ben :)
I do not get this part. I mean, what kind of variable value are you assigning to these dummy variables? What do they really meassure?
To clarify: You are not saying what those values assigned to the dummy variables mean, its just either 1 or 2, without any meaning to it.
Never mind, i got it now! Thanks for the uploads :)
What if the p-value result of LSDV and fixed effect estimation are different? Which one should I choose?
it is the same estimation in essence.
I always upvote Mewtwo
for a moment I thought I was the only one who heard it