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

[Todd R.Warfel]

On Sunday, June 1, 2003, at 07:28 PM, Boniface Lau wrote:

> But for participants to successfully group things, they have to
> observe the order among groups.

I've done a few hundred card sorts on content inventories (which is 
what I'm referring to when speaking of card sorts in IA) and have never 
had an issue with participants observing an "order" among groups. The 
participants observe piles, but not an order in the sense of a 
traditional order (e.g. sequence).

> For illustration, let's say there is a stack of cards...
>
> We know that "21" is already on the table. And it is ordered as the
> second of the three groups. So, we need to place "28" onto the second
> group. That is observing order. Thus, we maintain the three groups in
> the following order: (12), (21, 28), (33).
>
> Thus, successful grouping requires that order be observed...

The problem with this illustration is that you're assuming the 
participants will place the content cards in a sequential order, or 
some order within the group. This is not necessary for a successful 
content inventory (card) sort. Perhaps this is where you've 
misunderstood the process of how a typical card sort on content 
inventories is done. As mentioned above, typically, participants lump 
the content cards into piles that make a group, not a sequential order 
as you're sighting above (first card, second card, third card). (if 
you're using something like EZSort, then you're correct, they would 
typically create a structure, or have to create an order. This is not a 
shortcoming of the model, but rather a shortcoming of the application, 
as the application won't allow for piles. However, this hi-fi model is 
less reflective of their mental model then the typical piles (groups) 
created using the low-fi paper technique. This is where low-fi out 
performs - it is typically closer to their mental model. Additionally 
the low-fi method allows for groups of users, whereas EZSort only 
allows for one at a time. To find out more, read 
http://messagefirst.com/downloads/ModelingOrganization.pdf)

Perhaps if you could answer the following questions, this would give me 
a better understanding of your frame of reference.
Have you done content inventory card sorts in the past, or are you 
referring to a different type of card sort?
Were these sorts done using an application or machine, as you've 
referenced before, or were they done using a paper-based model? Did you 
notice any differences?

In the paper-based model, participants should not be instructed or 
encouraged to group them in a fashion as you've described. If they do 
this on their own, then they should be reminded that it's not necessary 
to create a "site map" or structure, but rather to put the content into 
"like groups" that they feel make sense. If they respond that creating 
an order within the groups helps them think clearer, then the response 
should be that this is fine, but that they're not required to do this 
during the exercise.

> No. Just that I am aware of the fact that successful grouping requires
> order to be observed.

In my experience of content inventory (card) sorts, I've never found 
that (sequential) groupings, as you've described above, was required or 
necessary to obtain successful results. I'm basing this on:
* several hundred sorts
* small to enterprise products
* small businesses to fortune 500s
* consumer, educational, commercial, financial, medical products
* intranets
* web sites to applications

I'm not saying it never happens, as I'm sure it does. What I am saying 
is that it's clearly not a requirement.

Cheers!

Todd R. Warfel

_//message first [method second]
-=========================-
User Experience Architect
message first
[P] (607) 339-9640
[E] twarfel at messagefirst.com
[w] http://messagefirst.com
-=========================-

In theory, theory and practice are the same,
but in practice, they're not.