Thanks for the video! One thing I am not completely clear about: what is the difference between homoskedasticity and zero conditional mean? Could you clear that up?
Hi guys. Hope you're doing well. I've tried my best to answer your question, but bear in mind that I am also studying this and I am by no means an expert. I think homoskedasticity refers to all random variables in a sample having the same variance, whilst heteroskedasticity is the opposite. An example of heteroskedasticity is that of the effect of years of education on wages. At higher levels of education, for example holding multiple degrees, you are able to access a wider range of jobs: the highest paid and most intellectually demanding ones, but also those which are open to people with little to no education, who can only access entry level jobs with lower wages. Zero conditional mean is a condition which is necessary for the OLS to be an unbiased estimator. The zero conditional mean is another way of saying that given the random variable, the expected value of the error term is zero. Therefore we cannot tell whether it is an over or underestimate. If there is a correlation between any of the independent variables and something unmeasured but which is contained within the error term, the zero conditional mean assumption is violated and the OLS estimator becomes biased. Let me know if this helps.
In your notation, you write E(u_i | H_i). I am assuming that for H_i is actually Hi and is just height, a single random variable. I wanted to confirm that the error term is also a single random variable and could be written as u and doesn't need an index and shouldn't be written as u_i since that's just a data point and not a random variable which has moments. The data that you have plotted are realizations or draws from the joint distribution of two random variables Earnings and Height and the error term is a single random variable u with various realizations. Can you confirm this please? If I am wrong and u_i is an index then can you please explain what the index runs over and what are these family of error random variables? Thanks
very good explanation and said very fast to not waste time
Thank you for this, it was just what I needed to find the connection between the two, also concise presentation!
u explained it better than woolridge
Thank you so much. Keep going you are doing great job!
Thanx a lot. It was really helpful to me!
great video and clear explanation. thanks so much!!
thanks for the video
Thanks for the video! One thing I am not completely clear about: what is the difference between homoskedasticity and zero conditional mean? Could you clear that up?
same question!
Hi guys. Hope you're doing well. I've tried my best to answer your question, but bear in mind that I am also studying this and I am by no means an expert.
I think homoskedasticity refers to all random variables in a sample having the same variance, whilst heteroskedasticity is the opposite. An example of heteroskedasticity is that of the effect of years of education on wages. At higher levels of education, for example holding multiple degrees, you are able to access a wider range of jobs: the highest paid and most intellectually demanding ones, but also those which are open to people with little to no education, who can only access entry level jobs with lower wages.
Zero conditional mean is a condition which is necessary for the OLS to be an unbiased estimator. The zero conditional mean is another way of saying that given the random variable, the expected value of the error term is zero. Therefore we cannot tell whether it is an over or underestimate.
If there is a correlation between any of the independent variables and something unmeasured but which is contained within the error term, the zero conditional mean assumption is violated and the OLS estimator becomes biased.
Let me know if this helps.
wow wow, chill out
hope this all makes sense - and we are DONE - my-man
Deepak Sivaraman i really hope so :D
Amazing 👏
thanks man... very helpful
Welcome
Thank you.
In your notation, you write E(u_i | H_i). I am assuming that for H_i is actually Hi and is just height, a single random variable. I wanted to confirm that the error term is also a single random variable and could be written as u and doesn't need an index and shouldn't be written as u_i since that's just a data point and not a random variable which has moments. The data that you have plotted are realizations or draws from the joint distribution of two random variables Earnings and Height and the error term is a single random variable u with various realizations. Can you confirm this please? If I am wrong and u_i is an index then can you please explain what the index runs over and what are these family of error random variables? Thanks
Anshuman Sinha I’m confused about the same thing
Im confused as well...
tysm
slow down. I did not understand nothing
Russian Power!
;)