quietly {
  /*
     The Covratio flags observations that have a major impact on
       the generalized variance of the regression coefficients.

     Values < 1 is increasing the estimated generalized variance and 
       this is generally unfavorable.  Values greater than 1 typically 
       decrease the variance.
  */
  regress  y  x
  predict covr1, covratio

  regress y1  x
  predict covr2, covratio

  regress y2  x
  predict covr3, covratio

  local n=10
  local j=1
  local covrp=1+((3*(`j'+1))/`n')
  local covrm=(1-`covrp')*-1

  #delimit ;
  graph twoway (scatter covr1 obs),              
    ylabel(0(.5)2, format(%3.1f))                               
    xlabel(1(1)10)                               
    title("Original Data")                       
    ytitle("") xtitle("")                        
    yline(`covrm' 1 `covrp', lp(dash) lw(thick)) 
    name(covr1, replace) nodraw
  ;
  #delimit cr

  #delimit ;
  graph twoway (scatter covr2 obs),              
    ylabel(0(.5)2, format(%3.1f))
    xlabel(1(1)10)
    title("Scenario #1")                         
    ytitle("") xtitle("")                        
    yline(`covrm' 1 `covrp', lp(dash) lw(thick)) 
    name(covr2, replace) nodraw
  ;
  #delimit cr

  #delimit ;
  graph twoway (scatter covr1 obs),              
    ylabel(0(.5)2, format(%3.1f))
    xlabel(1(1)10)                               
    title("Scenario #2")                         
    ytitle("") xtitle("")                        
    yline(`covrm' 1 `covrp', lp(dash) lw(thick)) 
    name(covr3, replace) nodraw
  ;
  #delimit cr

  #delimit ;
  graph combine covr1 covr2 covr3,
    title("COVRATIO Index Plot")
    name(covratio, replace)
    nocopies
  ;
  #delimit cr

  drop covr1-covr3
}