[Sigia-l] processing freelisting data

[Eric Scheid]

On 17/3/03 12:53 PM, "Christina Wodtke" <cwodtke at eleganthack.com> wrote:
> so is anyone going to say anything on this list, or is it going to degrade
> into a semantics game? Is this a list for pure theory, or are there a few
> working professionals around?

I'm working :-)

I've recently been trying out the free-listing technique that Rashmi wrote
about on B&A [1]. It's quick and simple to get data (other than a problem of
convincing the client to give the go ahead for user research).

Nonetheless I've done some informal/guerilla testing which despite not being
entirely valid due to poor audience sampling has provided some experience in
the task. A couple of tips: introduce the concept of free-listing with a
simple example, such as names of fruits or vegetables, demonstrating a
complete lack of careful thinking, categorisation, or ordering into some
kind of sequence of priority/importance/popularity. Keep it light and fast
and casual. Warn them that if they stop to think that you will take that as
the end of their list, but don't be too hard on them ... when they start to
hesitate read back the list so far, that can prompt more results.

The tricky bit comes in crunching the data. Doing it all by hand sounds
terribly tedious, so I've plonked the lot into a spreadsheet [2]. Working
out the calculations is now almost tedious. I'm thinking there must be a
better way, some software tools must already exist for doing this or
similar. Anyone know of some?

Anyways, once I've got the spreadsheet to produce the charts and graphs, the
next bit I'm scratching my head on is how to read the resulting matrix. It
helps to have it sorted by the occurrence counts. Like many of our research
tools I don't think it can give any definitive answers, only serve to guide
us in our decisions.

e.

[1] yikes, what a long URL:
<http://www.boxesandarrows.com/archives/beyond_cardsorting_freelisting_metho
ds_to_explore_user_categorizations.php>

[2] available on private request, limited to maximum of 9 respondents and
maximum of 30 terms. I'm thinking of re-doing it in a Filemaker Pro
database.