Surely there is no problem, except that the participants are confusing
positive and negative values on a rating scale with things being positive
or negative in an evaluative sense corresponding to "good" versus "bad".
That is not in any way intrinsic to the rep grid method, and there is no
need whatever for them to get involved with positive and negative numbers
at all.
A very nice feature of the RepGrid2 package for eliciting and analysing rep
grid information is that as soon as a participant comes up with a construct
it is represented by a line, at each end of the line are labels with the
names of two of the elements involved in elicitation of that construct -
e.g., by the triadic method. This line is the basis of a visual analogue
scale on which each element is then rated, but the participant does not get
involved with any numbers to give their ratings. To the left of the
rating-line is a list of all the other elements. The participant simply
drags each element, one by one, to its appropriate position on the analogue
rating-line. An additional valuable feature of this computer package is
that, as the rating process proceeds, any of the element labels - including
the ones initially places at the extreme positions, can be dragged along
to new positions, to allow for the fact that it may become clear, as
rating proceeds, that, e.g., Fred is actually a bit more like Mary in this
construct than Sue is.
Of course, the program does represent the ratings by numbers on scales that
either run from 0 to a positive number or from a negative number to a
positive number, but the participant does not have to attach "good" versus
"bad" meanings to the direction of scoring. When they come to look at their
grids, it can be explained that the direction of scoring is as arbitrary as
having longitude increase, rather than decrease, as we travel westwards
around the Earth.
Not everybody can have the luxury of using RepGrid2 - e.g., they may not
have a Macintosh computer - but it would be perfectly possible to do the
same kind of rating procedure using visual analogue scales bases on lines
drawn on big pieces of paper - perhaps graph paper - and using element
labels cut from Post-It notes, so that each can be stuck onto the
rating-line in its provisonal position, yet be easy to shift if second
thoughts arise.
In RepGrid2, by the way, the default option is to have the scales quantized
into 9 levels, but I see no reason to produce tied ratings artificially by
forcing the ratings onto such a coarsely quantized scale, so I usually set
it to more than 20 levels, which acts virtually the same as a continuous
analogue scale.
==========================================================================
Tony Downing, M.A., Ph.D.
Lecturer, Dept. of Psychology,
University of Newcastle upon Tyne, NE1 7RU,
UK.
email: A.C.Downing@Newcastle.ac.uk
Phone +44 (0)191 222 6184, Mobile: +44 (0)468427481
Fax: +44 (0)191 222 5622
==========================================================================
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%