where’s the data?

There’s a peculiar correction in today’s IHT:

An article Sept. 28 about a decision by the US Postal Service to issue
stamps featuring people still alive described incorrectly the measure of
popularity of a 1993 stamp picturing Elvis Presley and a 1995 stamp
picturing Marilyn Monroe. The Elvis stamp is considered the most popular
ever because people saved 124 million of the stamps. About 44 million of the
Marilyn Monroe stamps were saved. It is not the case that those totals
represent the number of stamps sold. (The Postal Service rates the
popularity of stamps according to how many are saved, not sold. A spokesman
said there are no reliable figures on actual sales before 2008.)

So … how does the postal service know how many stamps are saved? I mean,
the simplest way to calculate this would be to track the number sold and
subtract from that the number used (with some arbitrary cutoff date for use
based on historic patterns which say that 95% are usually used within time
period [x]). This is somewhat difficult as there’s no obvious way to tell
how many are used (counting at intake is theoretically possible, but
infeasible), and it’s made utterly impossible by the revelation that the
postal service has no clue how many were sold.

My best guess is that it’s an estimate based on the difference between the
aggregate total of postage sold in the year of release and the aggregate
average, minus the difference between the aggregate number of letters
transported in the year of release and the average number of letters
transported. But if i’m right, the number is a meaningless psuedo-statistic;
there are too many stamps issued in a year, and too much year-on-year
variation, for the number to be pinned on a particular stamp.
October 04, 2011 at 10:31AM