How did you come up with h5= 0.6?
I computed using the formula in page 245:
hi = 1/n + [(Xi-Xbar)^2/ summation from j=1 to n of(Xj - Xbar)^2]
shouldn't the answer for h5 = 1.2? However, if h5 = 1.2, I can't compute for sqrt(1-h5) since this will be a square root of a negative number. Please help.

for H5 this is my solution:
1/5 + (4/4) = 1.2 ???

I am still confused with the explanation, I have been reading the text so many times and it was stated there that "Suppose, however, that we refit the model deleting the i-th observation, obtaining an estimate of SE(-i) of sigma-E that is based on the REMAINING n-1 observations." My interpretation for this is that we will NOT include (5,0) in computing the regression, and thus the standard error of residuals is based on the first four observations only, that is, the standard error is 0.866. But based on this thread, it seems to be otherwise. Kindly enlighten me on this.

I can get SE(-5)=.866, but in order to get the E5* to match I believe I need to use h5=.6, which I'm unable to do. In the formula for Ei*, the only term that uses the (-i) model fit is the SE(-i) term, is this correct? For h5, I get (Xi-Xbar)^2 is 9, and the sum of the squares as 15. 1/n = .2, so h5 = .8. If we are supposed to use the (-i) model for hi as well, then I'm getting the the square as 12.25, and the sum of squares as either 5 (not including the 12.25) or 17.25, which neither give me h5=.6

Please help

To scomurphy,
I am afraid the calculation of h(i) is based on the value of X, not the value of Y.