On Tue, 31 Jan 2012, clarodina(a)lycos.com wrote:
ols d1 const d2 and the std error of d2 is 0.351919
d1
1 -0.9371
2 -0.40054
3 1.963222
4 -0.3328
5 -0.82194
6 1.085598
7 -0.9985
8 0.763004
9 -0.94785
10 0.047204
d2
1 -0.26785
2 -0.67142
3 -1.07196
4 0.891262
5 0.558466
6 -0.26347
7 0.822125
8 -0.17638
9 0.586626
10 -0.3612
Use s = sqrt[ ∑ ((e_i)^2/(n-2)) / (∑ (x_i - xbar)^2) ] for
manually calculating the std error and the value is 0.88583
No it's not!
ols d1 0 d2
d2dev = d2 - mean(d2)
sx = sqrt(sum($uhat^2)/($T-$ncoeff) / sum(d2dev^2))
Allin Cottrell