*
* Example 9.1 from page 291
*
open data crime1.raw
data(format=free,org=columns) 1 2725 narr86 nfarr86 nparr86 pcnv avgsen tottime ptime86 qemp86 inc86 $
  durat black hispan born60 pcnvsq pt86sq inc86sq
*
* Column (1)
*
linreg narr86
# pcnv avgsen tottime ptime86 qemp86 inc86 black hispan constant
*
* Column (2)
*
linreg narr86
# pcnv pcnvsq avgsen tottime ptime86 pt86sq qemp86 inc86 inc86sq black hispan constant
*
* Joint test of the "squares"
*
exclude
# pcnvsq pt86sq inc86sq
*
* These turning points are slightly different from those in the text because
* they're computed with full precision coefficients, not rounded.
*
disp "Maximum Point for PCNV Effect" -%beta(1)/(2*%beta(2))
disp "Maximum Point for PTime Effect" -%beta(5)/(2*%beta(6))
disp "Maximum Point for Income Effect" -%beta(8)/(2*%beta(9))