Eric, here's my take:
On the formula
* Assume there are J covariates (X). Then, there should be J+1 coefficients: 1
intercept term (Beta_0) and 1 coefficient (Beta_1, ... Beta_J) for each of the covariates
= J+1 coefficients.
* [cid:image001.png at 01CAA1D1.CD653800] is the covariance matrix of the
beta's
* The covariance matrix above has dimensions J+1 by J+1 because there is an
intercept (Beta_0), and the usual coefficients (Beta_1, ... , Beta_J)
* Each entry in the matrix is a covariance of respective Beta's, i.e., [i+1,
j+1] entry in the matrix is cov(Beta_i, Beta_j). The subscript is shifted by one because
of the intercept term.
* Variance is covariance of a beta with itself, i.e., the j+1 by j+1 entry of the
matrix is the variance of Beta_j. These terms can be found along the diagonal
* SE of a beta is the square-root of the variance of a beta
If you don't have an intercept term, then the equation below should have subscripts
j,j
Some useful R functions:
* t() transposes a matrix
* %*% is matrix multiplication, e.g., A%*%B multiples A and B, assuming A and B
are matrices with correct dimensions
* solve() finds the inverse of a matrix
* diag() grabs the diagonals of a square matrix and puts it into a vector.
Best regards,
Joseph
Joseph Poj Gavinlertvatana
Doctoral student, Marketing
Harvard Business School
203 Wyss Hall, Soldiers Field, Boston, MA 02163
Ph 617.230.5907
Fx 617.496.4397
Txt/Vm 617.910.0563
Em pgavinlertvatana at
hbs.edu
From: gov2001-l-bounces at
lists.fas.harvard.edu [mailto:gov2001-l-bounces at
lists.fas.harvard.edu] On Behalf Of Lin, Eric
Sent: Saturday, January 30, 2010 4:51 PM
To: Class List for Gov 2001/E-2001
Subject: [gov2001] Formula for SE
In the following formula for SE of beta:
[cid:image002.png at 01CAA1D1.CD653800]
What does the subscripted index j+1 stand for? (this is from the gov2000 notes, lecture 8
page 69.
Thanks!
EXL