Re: Constructs & attitudes

Fri, 16 Aug 1996 14:05:28 +0200 (MESZ)

Dear Bob,

> If it
> were practical and ethical I would be fascinated to see an
> example(s) of your grids.

it's not very practical to send a grid. But note, that we had the
standard procedure (described by Kelly), except that we collected
ratings on the constructs instead of binary responses.

> Presumably the grids had the same
> constructs (somewhat akin to the questions covered in the
> questionnaire).

This would be little informative, each participant provided his/her
own constructs and elements.

> If not, the raters did a remarkable job
> absorbing 70 grids (a single grid can be enough to drive one to
> distraction).

I agree.

> Can you give me an idea of the levels of correlation which were
> obtained?

They were in the range of .15 to .30. The higher ones for dimensions
describing social behavior. This is not overwhelming, but at least for
some of the five dimensions systematic and significant.

> To my mind your results support the proposition that
> a grid can provide a representation of anothers construing, to
> the extent that 'accurate predictions' can be made based on grid
> data.

I agree, but I would more moderately speak of 'better than chance'
predictions. Anyhow, our subjects had no information about the target
except the grid, and they had to deal with more than 70 grids.

Thanks for your encouraging comments
Kind regards

Rainer Riemann Phone: <0>521 106 4529
Universitaet Bielefeld Fax: <0>521 106 5844
Fakultaet fuer Psychologie e-mail:
und Sportwissenschaft
Postfach 100131
D-33501 Bielefeld