NY Times Story on Pilot Population Decline

OTOH The number of
discrete values in the set can easily skew the median.

Yes, and so which one is most appropriate would depend on whether a
=value= is likely to change much, or whether the =number= of values is
the important consideration. In the case of income distribution, the
former is probably more important. In the case of school test grades
from year to year, the latter might have more significance.

Jose
--
Get high on gasoline: fly an airplane.
for Email, make the obvious change in the address.

On Sat, 28 Apr 2007 13:29:13 GMT, Jose
wrote:

OTOH The number of
discrete values in the set can easily skew the median.

Yes, and so which one is most appropriate would depend on whether a
=value= is likely to change much, or whether the =number= of values is

The problem with income is we don't know if they include every income
that is different than the others or if they put them into brackets.
"I think" it's the latter, but have not been able to find any
information on how they come up with those figures. Most of those
numbers don't have a lot of meaning if we don't know how they are
derived. I remember reading that some incomes are exempt, or not
included.

As the old saying goes, "Statistics don't lie, but statisticians do."

the important consideration. In the case of income distribution, the
former is probably more important. In the case of school test grades
from year to year, the latter might have more significance.

Jose

The problem with income is we don't know if they include every income
that is different than the others or if they put them into brackets.

That only matters when figuring out the mode. For the median, if the
brackets are properly weighted it will work out just fine.

Jose
--
Get high on gasoline: fly an airplane.
for Email, make the obvious change in the address.