The assumption is the following. $$E[Y_i]=\beta _0 +\beta _{1}X_i, Var[Y_i]=\sigma ^2, Cov[Y_i,Y_j]=0, \forall i\ne j $$

Where $\hat \beta_0$ and $\hat \beta_1 $ are the least squares estimators.

I want to prove that $E[\hat \beta_1]=\beta_1$ and I am looking at my professor's work,

enter image description here

I get the first two rows, but I do not understand how he got the third row.

understand that

$$E[\Sigma (x_i-\bar{x})y_i]=\Sigma \left( E[(x_i-\bar{x})y_i] \right)$$

but I don't know how he got rid of the Expected value part.

May I get some help?

  • $\begingroup$ Because $x$ is non-random and $y$ is random( the denominator in $\hat\beta_1$ is just a constant). $\endgroup$ – StubbornAtom Feb 3 '19 at 6:26
  • $\begingroup$ Can you elaborate on that, please? How can you tell that x is non-random and y is? $\endgroup$ – hyg17 Feb 3 '19 at 6:46
  • $\begingroup$ That part must have been mentioned right at the start, because otherwise how can you proceed? $\endgroup$ – StubbornAtom Feb 3 '19 at 6:47

$X$ is a variable, but not a random variable. To be more rigorous, I would prefer conditioning on $X$, i.e, \begin{align} \mathbb{E}[\hat{\beta}_1|X] &=\mathbb{E}\left(\frac{\sum(X_i - \bar{X})Y_i}{\sum (X_i - \bar{X})^2} |X=\mathrm{x} \right)\\ &= \left(\frac{\sum(X_i - \bar{X})(\beta_0 + \beta_1 X_i)}{\sum (X_i - \bar{X})^2} |X=\mathrm{x} \right)\\ &=\beta_1 \end{align}

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.