Displaying 1 result from an estimated 1 matches for "hat_i".
Did you mean:
yhat_i
2003 Mar 30
1
simple test of lme, questions on DF corrections
...e can then analytically calculate the
maximum-likelihood results for various quantities, getting the
standard-looking results that the ML estimate of the within-group
random error is
sigma^2 = sum(e_ij^2)/N Eq. g.1
and the ML estimate of the between-group random error is:
sigma_b^2 = sum(b.hat_i^2)/N_groups Eq. g.2
(where b.hat_i are the fitted estimates of b_i, and e_ij are the
residuals after fitting).
Note that these formulas are slightly different than the standard
unbiased least-squares results, which divide by the number of
degrees of freedom corrected for the number of fit p...