Re: [Gnu-arch-users] OT: trained dependency

[Stephen J. Turnbull]

>>>>> "Miles" == Miles Bader <address@hidden> writes:

    Miles> I've seen similar numbers bandied about on slashdot (by the

Well, then they must be false!

    Miles> libertarian presidential candidate, I think!), and in that
    Miles> case, they seemed to be based on U.S. census data.

Could be: http://www.ipums.umn.edu/usa/peducation/lita.html says they
asked the simple question "how many people aged >=20 cannot read and
write?"  I don't have my account # and pwd on this box so I can't look
up the numbers at the moment.

Aside: If you have any interest in social statistics, I recommend
taking a look at that page.  It's short, it's relevant to this thread,
and it shows how people who really do know what they're talking about
talk.  It's nothing that requires a PhD to figure out, but it can help
in separating the charlatans from the pros.

    Miles> I don't know offhand how the census was done in the 1800s,
    Miles> but I wouldn't be surprised if it were biased in favor of
    Miles> population centers and the literate population

Methodologically, no, there was no such bias (remember, the primary
purpose of the census is Congressional districting---Congress critters
tend to get rather hot under the collar if their neighborhoods are
being undercounted), and the enumerators were specifically told to
count literacy.  In practice, it's kind of hard to enumerate
homesteaders with no postal address, social security number, or tax
records.  However, the numbers are small, and won't affect aggregate
literacy rate by much.  It's not like the problem where blacks and
Hispanics have historically been undercounted by 5-10% in _some_ inner
city neighborhoods, making racial gerrymandering much more effective.

Much more worrying is that census data is all basically self-reported
(the enumerators don't make any effort to verify anything but head
counts).  This would presumably cause an upward bias, though how great
I can't guess.

-- 
Institute of Policy and Planning Sciences     http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.