RE: Statistical Analysis of an Organizational Survey?

["Nimon H (Harry) at MSXSSC" <HN202460@MSXSSC.SHELL.COM>]

		Maybe it is just me, but I have a problem with any survey
selection range greater than three ('achieved standards,' 'exceeded
standards,' and 'did not achieve standards').  The greater the selection
range, the more ambiguous the results (IMHO).  Specifically, when the
respondent is faced with a selection of five items such as stated in this
article (very ineffective, ineffective, marginal, effective, very
effective), where does one draw the line between any of the above.  When
responding to a survey, the respondent has a set of standards in mind that
often bears little resemblance to the goal of the survey question, no matter
how well constructed.  Thus the need quantifiers within the survey.  To my
mind, attempting to focus results to even a single decimal place when faced
with such ambiguity is similar to trying to use lasers to increase the
accuracy of determining football first downs (American football that is).
The reality is that it depends upon where the referee places the ball after
unjumbling the mass of bodies laying on it.  One is either at, above, or
below the line.  It doesn't matter if 99.99999% of the ball is behind the
marker if it is actually even just touching it...it is a first down.  It
either is on the line, short of the line, or past the line.  If short of the
line, the standards have not been met and the next down is played.  If on or
past the line, the standard has been met or exceeded and the downs reset.

		Harry I. Nimon,
		Email: Hnimon@shellus.com <mailto:Hnimon@shellus.com> 
======================================================================