# aaargh!

anima@devi.demon.co.uk
Sun, 14 Jan 1996 18:42:55 +0000

Firstly,

On rereading my last posting to our mailing list, I see that I stated that
the expression for % Matching Scores is

>Matching scores are given by
>MS% = 100 + ((-100 . Sigma d) / (e . r))

No, it's not: I'm so sorry! The correct version is:

MS% = 100 + ((-100 . Sigma d) / (e . (r-1)))

Secondly,

Chris Evans' response to F.Reid Creech's original enquiry states inter al.,
of my mailing, that:

>I don't agree with him (viz. me, Devi) about the
>choice of matching score over correlation here, surely the ordinal
>relationship between ratings of elements is the crucial thing here,
>not the end of the pole toward which the ratings are clumped.
>However, I think his argument applies perfectly to explain why matching
>scores or some other distance measure should be used when considering
>relationships between ELEMENTS rather than CONSTRUCTS.

Well, he's probably right, and when I get my head round the issue (for I am
an intuitive rather than mathematical-derivation-led user of statistical
expressions!) I expect I'll be fully convinced.

I see what he's getting at, but can't for the time being get away from the
thought that, since the purpose of a grid is to _create_ construct
descriptions by eliciting them as "that which expresses meaning of an
element by giving a position to the element on several constructs
simultaneously" (for constructs don't exist as numeric ratings "in one's
head", as it were), the relative location of elements towards one end of a
construct pole rather than another matters even when one is considering the
relationship between _constructs_.

Er. Dunno, really. Are there any clear thinkers out there who could help
resolve this one?

Kindest regards,

Devi Jankowicz

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%