[Bioclusters] Bioinformatics (potential) needs assessment form?

Catherine Letondal bioclusters@bioinformatics.org
Wed, 18 Aug 2004 18:21:42 +0200

Dan Bolser wrote:
> Hi, 
> Has anyone ever done a survey of bioinformatics needs for an
> experimentalist (molecular biology) community?
> I need to survey the (potential) bioinformatics needs of a group of
> experimentalists with a wide range of bioinformatics knowledge (from 'some
> in-depth specialist' through 'general' to 'none').
> Ideally I would like to capture both what is currently being used, and
> what would be really useful if it were being used. 
> The problem is in the last case people may not know what could potentially
> be helpful. i.e. they need training and/or the form needs to be
> educational (gulp).
> I would be very grateful if someone could point me at a stock form, and I
> would be more than willing to help try to create such a stock form from
> the condensed knowledge of this community.
> Thanks very much for any assistance,
> Dan.


I'm not sure I understand what "stock form" means :-) but since we actually run a lot of 
users studies (survey, interviews, workshops, ...) in our lab, I can tell a little 
bit about it.

We have in fact made a survey about the use of informatics in our institute (600 answers), 
but to be honest, it is a lot of work and it is the worst way of getting information 
from users I have ever experienced.
The problem is that "bioinformatics needs", even for a restricted domain of biology, are
(at least in my experience), a quite rich issue. Every single biologist works differently
and has different needs. In fact, the risk, by doing a survey, is that the information
that will be gathered by this mean is something that you already know, not more.

So, how to get more useful information with less work?

Interviews are a great way of getting data. An useful idea for efficient interviews
is to ask the user about an actual situation (critical incident technique for instance). 
Avoid generalities. Although apparently 
too specific, it's a good way to really understand what people need since they sometimes 
don't know how to describe it. And you can generalize after. Running 20 interviews
provides a rich set of data, but running only 3 to 5 is very useful as well. I also
do interviews with biologists I have known for a long time: during the interview, we
always focus on a specific topic/situation.

Another wonderful communication tool is participatory design: you can do workshops
where users brainstorm on some topic, or workshop to prototype some parts of the design.
It's an invaluable tool to understand priorities of users regarding their
bioinfomatics needs (something that you completely miss in a survey). 
(see http://www-ihm.lri.fr/~mackay/pdffiles/MasterClass2000.pdf for a tutorial)

At first sight, this approach seems to focus on design more than one requirements
specification, but it is in fact also a tool for it, because it lets users express
their needs informally.

If you really want to do a survey, it's easier to design your questionnaire
after a few interviews, and after testing the questionnaire with 3-5 users. 

Some links:

- A. Dalke presentation at BOSC (about usability) also
tells about participatory methods (http://www.dalkescientific.com/bosc2002/usability/)

- Papers and people from this workshop in Edinburgh could be of interest
(R. Stevens from myGrid was there):
Requirements Capture for Collaboration in eScience
(I can send you our position paper describing user studies for a Web server
in bioinformatics).

- A very good paper illustrating user studies with molecular biologists:
O'Day, V., A. Adler et al. (2001): When worlds collide: Molecular biology as 
interdisciplinary collaboration. In Proceedings of ECSCW'01. pp. 419-418.

Hope this helps...

Catherine Letondal