Displaying 1 result from an estimated 1 matches for "cd_i".
Did you mean:
c_i
2005 Sep 13
4
plot(<lm>): new behavior in R-2.2.0 alpha
...AOV"
situation, something probably pretty rare nowadays in real
practice, but common of course in teaching ANOVA.
As you may remember, in a balanced design, all observations have
the same leverages h_{ii}, and the plot R_i vs h_ii is really
not so useful. In that case, the cook distances CD_i = c * R_i ^2
and so CD_i vs i {the old "4-th plot in plot.lm"} is
graphically identical to R_i^2 vs i.
Now in that case (of identical h_ii's), I think one would really
want "R_i vs i".
Question to the interested parties:
Should there be an automatism
``when h_...