Post date: Oct 01, 2012 3:29:14 PM
I've divided this exercise into a series of steps, based on Borgatti and Halgin (2011):
Select a fairly well-defined domain, such as fruit, animals, fast-food chains, brands of a product, illnesses, dog breeds, etc.
Select around 20 members of the domain. Ideally you would obtain these by freelisting, but in the interests of time you can just choose them yourself.
Create a deck of cards. On each card, write the name of the domain item, such as FOX. On the back side of the card enter a number from 1 to M (where M is the number of items in the domain) to serve as the id number for the item.
Create several decks so that you and your team members can collect data simultaneously and independently.
Ask at least 10 people to do a pilesort task.
Hand them the deck of cards and ask them to sort them into piles according to how similar the items are. They can make as many or as view piles as they like.
Tell them if they don't know an item, just put it aside and don't sort it. (Make sure you keep track of this one and don't enter it as a regular pile)
If the respondent wants to put an item into more than one pile, let them. Just take a blank card, write the item's name on it, and give it to the respondent.
If the respondent asks for help, just tell them that there is no right or wrong answer, we are just interested in their perceptions of the items and which ones are similar to each other.
When they are done sorting, do the following:
photograph the piles just as the respondent left them
Ask the respondent to name or describe each pile. For example, you can point to a pile and ask "So what kinds of journals are these in this pile?". Record these comments in your notes or via voice recording
Record the contents of each pile. You can do this simply by writing the pile number followed by the numbers of the items that went in it, like this:
Enter the data in a text file using the format of this file: holiday pilesorts.txt
Run Anthropac for Windows on the data file