I’ve used and loved Donna Spencer’s card sort spreadsheets. If you’re here, I’m guessing you have too.
Recently my team ran a follow-up card sort to dig deeper in an area of a taxonomy where we didn’t yet have much clarity. We got a little more, but the bottom line was that the participants had widely varied, often inconsistent, mental models of the subject. As a result, we weren’t able to do much standardization of their categories.
I found myself wishing I had a tool to generate a co-occurrence matrix. So, I built one. OK, actually first I hacked one. I hard-wired some formulas together for that particular sort to generate a matrix that showed how often any two cards were put into the same group. When we looked at the result, the team felt that it helped them identify the cards that were the best representatives of the most useful groups.
But a hack just won’t do. So I built a real co-occurrence matrix that works with any of Donna Spencer’s card sort analysis spreadsheets. That took me a bit off my blog posting cadence, but I think it’s worth it.
Here are a few of Donna’s reactions:
I think it’s a good way of exploring results. I think it is limited in that people may put two cards together in a group that only are together because they are held together with other cards in a group. So it could be a bit misleading to assume that a participant meant that the two cards belong together.
This is great – still takes effort to interpret, which is a good thing, but really useful.
I hope you like it too.