# PS3.4 for Example 3.8 on conf. intervals 
open data3-1
ols 1 0 2
# generate predicted price 
genr phat = price - $uhat 
# retrieve the number of obs. 
genr n = $nobs 
# retrieve degrees of freedom 
genr df = $df 
# retrieve sigma squared 
genr sgmasq=$ess/df 
# get mean of sqft 
genr xbar=mean(sqft) 
# get s.d. of sqft 
genr sdx = sd(sqft) 
# reset sample range 
smpl 1 1 
# calculate sxx 
genr sxx = 13*sdx*sdx 
# other variables for confidence interval calculation 
genr x0 = 2000
genr temp1=((x0-xbar)^2)/sxx 
genr temp2=(temp1+(1/n)) 
# calculate using equations 3.28 and 3.29 
genr sysq1=sgmasq*temp2 
genr sysq2=sgmasq*(1+temp2) 
# take square root for standard errors 
genr sy1=sqrt(sysq1) 
genr sy2=sqrt(sysq2) 
# predict mean y for x0 
genr ymean0=52.351+(0.13875*x0) 
# compute bounds for confidence interval using equation 3.28 
genr ymean1=ymean0-(2.179*sy1)
genr ymean2=ymean0+(2.179*sy1)
# compute bounds for confidence interval using equation 3.29 
genr y1=ymean0-(2.179*sy2)
genr y2=ymean0+(2.179*sy2)
# compute large sample confidence interval, that is, plus/minus 2 sigma 
genr sgmahat = sqrt(sgmasq)
genr y3 = ymean0 - (2*sgmahat)
genr y4 = ymean0 + (2*sgmahat)
print -o n df sgmasq xbar sdx sxx sysq1 sysq2 sy1 sy2 ymean0 ymean1 \
ymean2 y1 y2 y3 y4