Re: Best Methodology for Search Usability Testing?

Hi Doug,

I did quite a bit of testing of search some years ago.
It's best if the person truly needs the information, so try to build a
situation where they need specific answers. This may mean that they do several
searches as they refine the answers they get -- and that's good.

You didn't mention the type of search. For example, if it's related to commerce
/ shopping, this is a great time of year. You could ask the participant to list
2-5 people they need to buy gifts for and the kinds of things they might like.
Then ask them to use the interface to find actual objects they would like to
buy. This leaves the categories / words pretty natural to each participant. You
can add an incentive in the sense that if they find something they truly want,
you will buy it for them. If you have a prototype, you might want to limit your
recruiting to people who only want to buy certain things, like shirts or books
-- that you have in your prototype.

If it's something else, like a book topic -- then you can build your questions
around certain answers. Like, locate the most recent virus, or discover the
string to use to accomplish X. 

If you look through Jakob Nielsen's site, there may still be a reference to a
study we ran on search. At that time, we were looking at how people looked up
technical information in Sun's AnswerBook.

Meghan
--- [email protected] wrote:

> Hi,
> 
> I'm interested in conducting a usability test for a new search strategy.
> We're trying to incorporate some of the newer type-ahead and
> pre-catergorization technologies you see in google toolbar or ask.com.  I
> have a talk-aloud and screen record method that I've used for most usability
> test, but I'm not sure this is the best way for testing something like
> search.
> 
> Nielsen maintains that you must observe "how people search" to get accurate
> findings, but I'm worried about creating an artificial, experimenter-biased
> situation.
> 
> It's hard to say "Search for anything" or "Do what you would normally do"
> because the user isn't in a normal situation if in a test environment
> 
> Any thoughts?
> 
> Doug
> _______________________________________________
> This is the BayCHI Discussions mailing list, [email protected]
> To change your subscription options, or to unsubscribe, visit
> http://baychi.org/mailman/listinfo/discussions
> 



      ____________________________________________________________________________________
Get easy, one-click access to your favorites. 
Make Yahoo! your homepage.
http://www.yahoo.com/r/hs 
_______________________________________________
This is the BayCHI Discussions mailing list, [email protected]
To change your subscription options, or to unsubscribe, visit http://baychi.org/mailman/listinfo/discussions