Page 1 of 1
simple correlation coeficient
Posted: Thu Dec 13, 2012 4:43 pm
by msrahman
Is it possible to calculate simple correlation coefficient( not autocorreltioan) and its SIGNIFICNCE LEVEL between two series in RATS? Thanks.
sincerely,
msrahman.
Re: simple correlation coeficient
Posted: Sat Dec 15, 2012 10:32 am
by TomDoan
msrahman wrote:Is it possible to calculate simple correlation coefficient( not autocorreltioan) and its SIGNIFICNCE LEVEL between two series in RATS? Thanks.
sincerely,
msrahman.
you can compute it in many ways. For instance:
cmom(corr)
# x y
disp %cmom(1,2)
As to the significance, what assumptions do you want to make? The standard asymptotics are that sqrt(%nobs)*rho~N(0,1) under the null of no correlation. You could also look at the F-statistic from a LINREG to Y on CONSTANT and X (or X on CONSTANT and Y) which will give a similar, but not identical value.