I am kinda confused on the discussion regarding what the denominator should be when calculating the mean hat value. On the practice problem sheet, there is a row on the bottom of the data which gives averages, and the result for the average of the hat values uses the total number of observations as the denominator. However, from reading the posts above, the consensus seems to be to divide the total by 1.

Bubba Gump, you list your answer for part A as being k=1 with no mention of the mean hat value. Is all this discussion in regards to a previous version of HW #17 which had a different question for part A?

a) I have k = 1.

c) h1 = .3231
...
h5 = .9692

I'm guessing its okay to compare answers?

Part A) I don't think that's right. The textbook says "k is the number of regressors in the model, excluding the constant." How did you get k = 2?

Part C) I think that's right. This is a simple regression, so I used that formula, too.

For Part C should we use the formula from page 245 of the text?

hi=(1/n) + (Xi - Avg(X))^2 / sum(Xj-Avg(X))^2

Also, for part A should the number of regressors we use in the average hat-value calculation be 2?

I am struggling with Part C.   NEAS or anyone else, can you please post an example of how to calculate the individual hat values from a small sample of points like we are asked to do in Part C of the HW? 

[NEAS: See the posting on hat values practice problem. The final exam has problems like this, so review the practice problems.]

 

A.     What is the mean hat value for these five points?

B.     What is the range of hat values for these five points?

C.    What are the hat values of the five points?