Marcello Pagano <[email protected]> wrote,
> Bill Gould in 1999 answered the question of how to store pairwise
> correlations of a varlist in a matrix--basically calculate the
> covariance matrix and then convert it into a correlation matrix. The
> problem with that answer is that it does not use all the data--basically
> the difference between -corr-and -pwcorr-. Does anyone have a better
> answer, one that uses all the data, like -pwcorr- does?
One solution would be the following Mata function,
==============================================================================
real matrix pwcorr(string scalar vars, |string scalar touse)
{
real scalar n
real matrix X
real matrix P
string matrix V
V = tokens(vars)
n = cols(V)
P = J(n, n, 1)
for (i=1; i<=n; i++) {
for (j=1; j<i; j++) {
if (i!=j) {
st_view(X, ., (V[i], V[j]), touse)
P[i,j] = P[j,i] = corr(variance(X,1))[2,1]
}
}
}
return(P)
}
==============================================================================
The frunction takes two arguments. The first is a list of variables (not a
varlist, i.e., "m1 m2 m3", not "m*"). The second is argument is optional.
It is the name of a single variable that has zero and nonzero values.
Observations with nonzero values will be used.
Here's the function in action:
. sysuse auto, clear
(1978 Automobile Data)
. replace mpg = . in 1/20
(20 real changes made, 20 to missing)
. replace weight = . in 21/30
(10 real changes made, 10 to missing)
. mata: pwcorr("mpg weight displ")
[symmetric]
1 2 3
+----------------------------------------------+
1 | 1 |
2 | -.7304624386 1 |
3 | -.7407515152 .8711929724 1 |
+----------------------------------------------+
If you want to save the matrix in Mata matrix, code
: whatever = pwcorr("mpg weight displ")
in Mata.
If you want to save the matrix in a Stata matrix, code
. mata: st_matrix("whatever", pwcorr("mpg weight displ"))
-- Bill
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/