Sampling Grids

Harald Seelig (
Fri, 15 Jan 99 15:58:25 +0100

Here is what I have been trying to figure out for a few weeks:
How can grids of different individuals be compared? And what about grouping and
sampling grids?
My suggestion is that there is a possibility to sample grids by
modifying the gridmethod and using appropriate statistical methods. But not
having read all of the theoretical issues that support the background ideas and
ideals behind PCP I`m not sure if modifying the gridmethod wouldn't lead away
from the initial ideas.
Because of this, the first thing that needs clarification is: Is a grid that
could be compared (1:1) to another still 'a grid ` or is it (only) a special
sort of Osgood's semantic differential?
Considering given constructs (instead elicited - in every interview), there is
the problem that - even if a subject is able to use the given dichotomy - there
is no validation that their use by one subject would be similar to that of other
subjects as well for their use in another situation (point of time): One solution would
be to find 'stable' and/or 'definite' constructs. But this, I think, would lead
away from constructivism in the end. Even if there are possiblities to find
definite (superordinate) constructs, there still would be a problem with the
comparability of the 'styles' in which elements and constructs would be combined.
So the second question (for me) is: How much of the methodological weight of the
dichotomy corollary could be kept if modifiying a grid is considered.
Would there be any support by taking the commonality corollary into account?
Would it be necessary to think of other kinds of dichotomies?
By coming this far the next questions come to mind. What would the information we
receive from (given and) sampled grids be like? What kind of interpretation
would be possible? And thinking of the data: is there dichotomous (nominal) or
rated, or ranked (ordinal) or numerical information? What kind of statistical
methods are appropriate?
I think all this is manageable. So I tried it. The results I found seem to be
okay to me, but I'm not sure if 'giving up' the individuality corollary wouldn't
disqualify the whole thing.
Hints and opinions appreciated.



Harald Seelig
Vierlinden 1
79102 Freiburg