[newbie dumb question] Why is that the assumption of E(u)=0 essential for interences?

#1
If E(u)=0 is violated, how would it affect the inference of parameters, such as by using t-statistics?

My understanding is that if E(u)=0 is violated, then the distribution of the parameter is not normal, therefore t-statistic is invalid. But why is that the distribution of the parameter would not be normal under such violation? Thanks for helping out!
 

Dason

Ambassador to the humans
#2
Re: [newbie dumb question] Why is that the assumption of E(u)=0 essential for interen

What is 'u'? Would you mind explaining what you mean a little bit more?
 
#3
Re: [newbie dumb question] Why is that the assumption of E(u)=0 essential for interen

Oh yes, I'm sorry for the confusion.
By E(u), I mean the error term of the regression equation.