*
* Example from page 168, section 6.1
* Regression through the origin
*
cal 1971
open data table6-1.prn
data(format=prn,org=columns) 1971:1 1980:1
*
* Run the regression through the origin, and with an intercept. Compute a 95%
* confidence interval for the beta coefficient each way. Note that, given the
* order of the regressors in the 2nd regression, the coefficient is %BETA(2) and
* the std error is %STDERRS(2).
*
linreg y
# x
*
* Compute the scale for a 95% confidence interval for the t distributions. These
* have different degrees of freedom for the two regressions. You can look these
* up in a table, or use the %INVTTEST function. This takes two arguments: the
* first is the desired tail probability (thus 1-.95=.05 for 95% confidence bands),
* and the second is the degrees of freedom, which is just %ndf for the last
* regression run.
*
compute t95 = %invttest(.05,%ndf)
disp "95% confidence interval" %beta(1)-t95*%stderrs(1) %beta(1)+t95*%stderrs(1)
linreg y
# constant x
compute t95 = %invttest(.05,%ndf)
disp "95% confidence interval" %beta(2)-t95*%stderrs(2) %beta(2)+t95*%stderrs(2)