[ taking it out of the HTML script ]
On 8 Sep 2001 22:14:58 -0700, [EMAIL PROTECTED] (Jay Warner) wrote:
[ ... snip, details of excellently devising a pragmatic measurement
on the job]
========= start
"Years later I discovered what I thought was a Likert Scale, so I
called what I had done a Likert scale. Otto Kral, whom some older ASQ
and ASA people might remember from 3M, called it 'making a ruler.'
Which it is. It is the initial stages of attaching numbers to
non-numerical items.
"Now, this scale did not apply to attitudes. It used decimals (i.e.,
it was continuous). It had limits of 0 and 6, so if I wanted to do an
analysis, I might have had to do a transformation on the data prior to
assuming Normality.
"Are you folks, my statistical elders, trying to tell me that I didn't
use a Likert scale? Perhaps I should call it a Kral scale? "
======== end excerpt
I should say that this is a bit weird: that you seem to be
offended if you don't have a 'Likert Scale' - as if that
was some peak achievement. Well, you did not have
a Likert Scale; but you also did not have an IQ scale,
nor a normal z-scale, nor a Rasch scale -- and those
are each approximately as meaningful.
Right, you don't have a Likert scale because it is not
made up of the sum of similar items. You don't have
a Likert item, or even a Likert-type item, because it is
not one item in a set of parallel items which are designed
to be summed. And you don't have headings/anchors
that sound 'Likert-like' (or you haven't named them).
You had, perhaps, a graded 9-point score, with 3 or
5 descriptive anchor points, and you kept physical
reference samples on hand. You say that
you eventually demonstrated a reliability
of no-more-than-0.5 point difference (95% of the time?)
with your co-worker, even when challenged with the
sneakiest samples your co-workers could invent.
Sounds awfully good to me. Call it a ruler? Okay.
(No 'Likert item'-by-my-standard was ever that
reproducible between two raters.) Call it what you will.
I'd say, some people might be willing to call it Likert,
but for those people, there is probably (I think) NO
criterion in mind beyond having an ordered points.
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
