Path: utzoo!dptcdc!jarvis.csri.toronto.edu!mailrus!shadooby!accuvax.nwu.edu!tank!ncar!umigw!umbio!medsoft!f6.n135.z1.fidonet.org!Ed.Maurer
From: Ed.Maurer@f6.n135.z1.fidonet.org (Ed Maurer)
Newsgroups: comp.sys.ibm.pc
Subject: Need PC Skill Tests
Message-ID: <273.244AB19F@medsoft.uucp>
Date: 17 Apr 89 17:33:00 GMT
Sender: ufgate@medsoft.uucp (newsout1.18)
Organization: FidoNet node 1:135/6
Lines: 46


 > I am looking for written tests and answer sheets that can be used
 > to test a person's technical skills in any of the following
 > software environments:  Apple Macintosh, Microsoft Windows or
 > Presentation Manager, PC-DOS, OS/2, SunWindows, X Windows, UNIX,
 > DECwindows, or VMS.  I would also like to test the person's
 > knowledge of C, C++, and 80x86, 680x0, DEC, or IBM 370 assembly
 > languages.  Finally, I would like to test the person's knowledge
 > of compiler writing theory and practice.

Having been in charge of Corporate support for Fortune 500 (and I don't 
know of many that use tests - IBM reviewed and threw them out 10 years 
ago), I can assure you there ain't no such thing worth the paper its 
printed on. Over the years, I have had a few sales pitches for same tossed 
my way, and I always ask the producers of such junk if the illustrious Dr. 
Einstein or, more recently, Dr. Land or Stephen Jobs, or even my best staff 
would 'pass'- for which they have no answer. For more on the general 
subject, read "Essays on the Mis- Measure of Man" by Stephen Jay Gould.

I do recall a major study commissioned, I believe, by the GSA, comparing 
technical test scores with peer-management evaluations. The CE (correlation 
co-efficient) was in the vicinity of .35 (.50 being pure chance), but, 
since it was merely tax-payer money, Uncle bought it anyway.

If we in the computer science community are light-years away from true AI, 
why would you think that the Testing/Psychological community (in whose 
ranks I also reside) is any closer with a pencil and paper? Why not just 
use a lie detector? Perhaps a Ouija board?

To a large extent, technical-programming skills, like most 'human' work is 
nebulous if not <shudder> 'artistic' in nature, and any test scores mean 
less than Binet numbers in the long run. Because of real philosophical as 
well as rational problems I have with testing, I will not give you sources, 
but If you're insistant on mediocrity, I do know (having walked out on a 
job offer upon being told of the requirement) that the Miami Herald (aka 
Hearst aka "the Inquisition") uses a battery of written tests.

C:\Ed /\\aurer  > run techtest.exe
                > bad command or filename


--  
Ed Maurer - via FidoNet node 1:135/3
Medical Software Exchange BBS (305) 325-8709
UUCP: ...uunet!gould!umbio!medsoft!6!Ed.Maurer
ARPA: Ed.Maurer@f6.n135.z1.fidonet.org