Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: $Revision: 1.6.2.16 $; site inmet.UUCP
Path: utzoo!lsuc!watmath!clyde!burl!ulysses!gamma!epsilon!zeta!sabre!petrus!bellcore!decvax!yale!inmet!janw
From: janw@inmet.UUCP
Newsgroups: net.politics
Subject: Re: Orphaned Response
Message-ID: <7800688@inmet.UUCP>
Date: Fri, 15-Nov-85 23:18:00 EST
Article-I.D.: inmet.7800688
Posted: Fri Nov 15 23:18:00 1985
Date-Received: Tue, 26-Nov-85 21:27:18 EST
References: <797@whuxl.UUCP>
Lines: 49
Nf-ID: #R:whuxl:-79700:inmet:7800688:000:1997
Nf-From: inmet!janw    Nov 15 23:18:00 1985


[ tim sevener whuxn!orb]
>A common statistical fallacy is to cite overall averages or per capita
>figures.

To think that merely *citing* any correct datum can be a fallacy, is
itself a fallacy. A pretty obvious one.

>For example, ten people could have a thousand dollars in total,
>with one person having $991 and the rest having $1 apiece.

Right. Now try and  repeat  the  same  example,  pro  rata,  with
*calories*.   (Say,  2000 a person a day - 1000 would kill them).
One person consumes 19,820 calories a day, the rest 20  calories
a  day.   Impossible ? Well, that should tell you why per capita
figures make more sense for nutrition than income. (Under certain
conditions,  they  can  be  quite useful for income, too.  It all
depends on additional constraints).

>Thus per capita figures are very misleading and almost never used
>by social scientists doing serious comparisons of goods distributions.

But used *a lot* for comparing food consumption in different nations.
Or at different periods in one nation.

>> (4) Now, since you are interested in statistics, try  and  verify
>> the  following  theorem:  "If  the average person is hungry, then
>> *some real people* are hungry, whatever the  distribution".  See,
>> averages do tell you something. In fact, per capita  figures  are
>> universally and correctly used in this field of study.

>Your theorem is obviously correct that if people on average are hungry,
>therefore some people must be hungry.

Good. A point of agreement.

>However your next statement,
>that mere overall averages are perfectly OK is wrong.

Since I never made that statement, and find it meaningless,
I have to return it to you.

I said they are universally *used*: a matter of fact that you 
do not dispute. I also said using them is *correct*, i.e. not 
a fallacy. I can also add they are *useful*, i.e. yield
meaningful conclusions. This is demonstrated by the theorem 
above and its application to China in my original article.

		Jan Wasilewsky