Update on our data tables study

Recently we conducted an online study about data tables. Many of you kindly participated in that study and so we'd like to take this opportunity to let you know a bit more about it.

First off, we were absolutely delighted with the number of participants, which was greater than we could have hoped but just the right amount to support the sort of analysis we needed. In total, we had 281 people start the study. Some of these dropped out before the end, or didn't answer any of the questions, leaving us 244 responses that we could use.

The study asked participants to answer six questions using a fictional table of data about screws. What we didn't say was that the formatting of the table changed half way through the study.

The change in formatting was the addition (or removal) of zebra striping. Zebra striping is application of shading to alternate rows, to help with readability. So, half the questions were asked when the table had zebra striping, and the other half were asked when the table didn't have zebra striping. The two versions of the table are illustrated below.

So that we could minimise any 'learning' affect, we was randomly determined which version of the table participants would see first. That is, some participants started with a plain table and moved to a striped table while the remaining participants started with a striped table and moved to a plain table.

The aim of the study was to assess whether zebra striping does in fact make reading easier. The two measures we used were accuracy (i.e. whether you got the answer right) and speed (i.e. how quickly you answered the question). We now know whether zebra striping has an affect on these two measures.

However because we've submitted a paper to a few conferences, based on the research results, we can't release most of them here just yet. In the meantime, here's a little interesting factoid. Participants who had the plain table first had a marked preference for zebra striping, whereas participants who had the zebra striping table first were less fussed either way.