Supplementary Exercise 1.144 of IPS7e
-------------------------------------
(continuation of Exercises 1.42 and 1.72)

Percents of residents aged 65 and over in the 50 US states (in 2000). 


(a) The extreme outliers at each end of the distribution would appear as 
points off the roughly straight line formed by the other points. The
positions of the outlying points relative to the line depend on whether
one uses a normal probability plot or a normal quantile plot. In a
normal probability plot, the lower extreme value is above the line
whereas the upper extreme value is below the line. In a quantile plot,
the positions relative to the line are reversed (because the axes are
switched).


(b) Minitab commands:

 PPlot 'over65';
  Normal;
  Symbol;
  FitD;
  Grid 2;
  Grid 1;
  MGrid 1.

Comments:
---------
As expected from the discussion above, the plots shows the central
values of the distribution to be reasonably well within the indicated
bounds around the straight line, whereas the lower point is very much
off the line (and the bounds) and the upper point is off the line but
not quite off the bound. This matches our previous discussion which
indicated that the lower outlier (Alaska) was by far the strongest. The
A-D normality test gives weak evidence (P=0.044) against the normal
distribution. Without the lower outlier, the data appears to conform 
reasonably well to a normal distribution; try to redo the probability
plot without the value for Alaska!