Heteroscedasticity: dealing with the problems caused

[@larissacury7714]

Hi, very interesting! thank you!! Do you also have a video on violating normality of residuals ? (accounting big and small samples)

[@hasben0]

Ben is the man💯💯

[@1982sadaf]

Just to double check. So heteroscedasticity due to omitted variables causes Biased estimates, but heteroscedasticity not due to omitted variables only is not Best, meaning it is un-biased?

[@lastua8562]

Heteroscedasticity causes biased estimates only if additional variables are correlated with existing x1. If you feel like you have included all relevant variables, you may well assume the model is unbiased, i.e. blUE. It may not be Best, because of the heteroscedasticity of errors.

[@marcuschiu8615]

@@lastua8562 I don't think heteroscedasticity can be compared with biased/unbiased estimates. An estimator is unbiased when 𝐄[𝑒𝑟𝑟𝑜𝑟|𝑥] = 0. An estimator is dealing with a heteroscedastic population when 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒(𝑒𝑟𝑟𝑜𝑟|𝑥) = 𝑛𝑜𝑛-𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑥. I do NOT think Heteroscedasticity can cause an estimate to be Biased. Also if a model contains a regressor variable that is correlated with a missing variable, that model can still be unbiased, it's just no longer BEST, hence no longer BLUE, but it's still unbiased. An example can be shown at kzbin.info/www/bejne/e2qsYmR9oah3qMk

[@YeungyBassMasta]

@@marcuschiu8615 Marcus is right, omitted variables lead to under/overestimated coefficient estimates on the regression. This means that the coefficients are biased (upwards or downwards). Heteroskedasticity is not typically a result of OVB, it's just a result of the form of the residuals. It will lead to standard error bias however, which simply means that the estimate may not be statistically significant.

[@voiceofusmanau]

what did u say in the last part of video when u say i have used X1 when infact i should use.....X??? I did not get it. 11:30-11:35

[@lastua8562]

x1 is X. Your reference to 11.30 is not in this video.

[@Marchinoft]

Advise: if you want to explain something you are supposed to speak slowly and clearly.....not always to in this video....

[@SpartacanUsuals]

+Marco Boschi My apologies if my speaking wasn't clear here. Sometimes I have a tendency to mumble. Particularly when saying the word heteroscedasticity! Best, Ben

[@Marchinoft]

Ben Lambert Thanks.I had a very bad teacher of econometrics and I have to take an exam soon....so I am still sensitive to these teaching details, sorry. ;-P

[@vp4744]

Ben, I wanted to thank you for the videos but didn't know where to put this comment. Thought this might be a good place since the reason I watch your videos is because you don't mumble. People complaining about your mumble have not had Russians or Eastern Europeans teaching this subject matter. Those are the lucky ones. Cheers.

[@Wandapenny2022]

Get some manner. This guy is teaching millions for free. You are not entitled to criticize his teaching quality if you don't pay him.

[@orderinchaos]

You can play videos at 75% or 50% if you want, that's in the settings.