[Sigia-l] Card sorting with complex content

[Jennifer Healy]

>c/j wrote:

>Does anyone know of any research that specifies the
>number of individuals needed for a card sort in order
>for the findings to be somewhat "valid"? 

IBM developerWorks [http://www-106.ibm.com/developerworks/] has a useful
tutorial on card sorting. According to them, "Interviewing from four to
seven participants in an informal study can get you the data you need." 

>My sense is that I should be able to get away with a fairly low number
>within each major segment (perhaps around 3)

I've found that sample sizes for card sorting are typically similar to those
for usability testing, as this is a similar exercise. If your segments are,
indeed, drastically different from one another, three individuals per
segment should be more than sufficient.