Don't take David's remarks too seriously.  Either he was writing with
tongue firmly embedded in cheek, or he was being more than usually
curmudgeonly.  See my comments on his comments, below:

On Wed, 25 Jun 2003, Dianne Worth wrote:

> I found the answer to my own question: One would subtract/add 4 to
> the -3/+3 scaling, NOT based on whether the statement was positively
> or negatively worded, but because the Likert-type scale had (for
> example) Extremely Unlikely as -3 and Extremely Likely as +3 for one
> statement; but as +3 and -3 in another part of the questionnaire.
> This converts them so they are all headed in the same direction.

Rich Ulrich has already commented on THIS point.

> As far as your comment about the holy water (I like that!), does the
> following clarify? "..assume that, on the basis of a great deal of
> prior experience in encoding, the terms 'extremely,' 'quite,' and
> 'slightly as linguistic quantifiers have been associated with more
> or less equal degrees of intensity.." (Osgood, et. al. 1957, p. 29)

   < snip >

> --- David Heiser <[EMAIL PROTECTED]> had written:

> >   Wow!!
> >   I didn't know you could take nominal data,

Semantic differential data are ordinarily AT LEAST ordinal, and are
often not far from interval-scale;  at least, not far enough for the
departure to be noticeable in the general noise.

> > sprinkle a little holy water (from your supervising professor's
> > closet) and lo and behold it becomes ratio data that can be analyzed
> > by regression analysis.

Regression analysis pretty well wants interval-scale data, at least
approximately;  it does NOT require ratio-scale data.

  < snip, the rest >

 -----------------------------------------------------------------------
 Donald F. Burrill                                         [EMAIL PROTECTED]
 56 Sebbins Pond Drive, Bedford, NH 03110                 (603) 626-0816

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to