[Sigia-l] card sorting: dealing with multiple placements

[Eric Scheid]

On 2/6/03 9:28 AM, "Boniface Lau" <boniface_lau at compuserve.com> wrote:

>> At the beginning of a card sort there are *no* "certain locations".
> 
> I am not sure what you meant by "no certain locations".

predetermined, known before hand...

If I gave a stack of items to a user and asked them to sort them into
groups, I cannot be certain there would be certain specific categories.
Users are like that sometimes.

>> By what you've written so far I'm assuming the latter. I'd agree
>> with the former, but not necessarily the latter. Grouping, without
>> internal sequencing, can be an end in itself,
> 
> It is not about the order(sequence) of items within a group.
> It is about the order(sequence) of groups.

As you've well explained in your immediate recent message. Which when I read
it I thought "Well, that was out of left field." It's an interesting point,
worth considering...

I'd dispute your contention that order (sequence) must be observed (obeyed).
Users do indeed need to be mindful of just where their little stacks of
cards are physically on the table so as to add more cards to those stacks
(never do a card sort in the dark ;-), but they don't particularly care or
require that that arrangement of stacks observe some form of sequence.

How would you expect users to sequence the following groups:

    cutlery
    crockery
    power tools
    books
    toys
    musical instruments

Actually, expecting or looking for a sequence is itself misleading, as the
card stacks are usually arranged in a two dimensional space, not one
dimensional sequence. Further, that two dimensional space isn't always used
to form a matrix, but instead some stacks are near to others and others are
far outfield. Users often use the physical space to remind them of the
affinity of groups to each other. In the above collection of groups cutlery
and crockery would be close to each other, as would books, toys, and musical
instruments, and power tools would be "over there".

I've seen this affinity arrangement time and again for different information
types: the range of products from a hair salon wholesaler, genres of film/tv
scripts, format classifications of film/tv productions, organisation
classifications for a consultancy providing services to non-profits, market
classifications/segmentations for a company that organises sports related
social events (dinner with football greats) ...

These affinity arrangements are most often virtually random, usually because
they don't spend too much time rearranging the stacks and thus the first or
most obvious stacks are closer to the centre of the table and the rest just
expands from there, sometimes by close affinity, sometimes simply by virtue
of available space.

Occasionally, I do see some sequencing applied, but when I do it's more
often a post-hoc "tidy up". Some of these sequences are of an ordinal form,
but sometimes they are instead a story-telling sequence. My curiosity piqued
by the latter I did some further testing, and found different users formed
different story-telling sequences. Further, when I then asked them to
arrange the groups onto a two dimensional space and asked them to add in a
collection of sample elements they abandoned their story sequence and used a
multi-relational affinity arrangement again.

That's my experience though. Perhaps others have seen users develop a strict
sequencing of groups on which to rely on for the mechanics of locating their
groups on the tableaux.

So, what's your experience?

e.