Is it a coincidence that the variance of alpha is equal to the variance of predicted mean response , or is there any connection between two?
The formulae for the variance of the mean response and the individual response, given on page 25 of the Tables, can be calculated for different values of x0. Therefore confidence intervals for the expected response when x= xo and confidence intervals for the value of a response when x=xo can be found. To determine confidence intervals for alpha, the intercept term, we substitute x0 = 0 into these formulae. Assuming that we want a confidence interval for the expected response when x0=0, (rather than the value of a response when x=x0), we use the formula for the variance of the mean response with x0=0.