The purpose of this study would be to determine which terminology should be used in the Paparazzi interface to distinguish between links that are in a blog entry but link to other sites and reference or links to a blog entry from other blog entries. This study would also inform our development of the Paparazzi help system.
The Paparazzi interface enables users to follow blog chains in two directions. We have tried different versions of this language in every one of our prototypes. Each time, there has been confusion for our users. The model that we have is that there are different types of links associated with a particular blog entry. Some are links in from other entries. Some are links out from that blog entry to other articles or entries. Additionally, we believe that some of this confusion may come from the fact that in our user tests, written or spoken tasks are phrased using a particular set of terms and may strongly influence the user's interpretation (and amount of user misunderstanding). In some sense, this language introduces an element of artificiality to the user task as we do not know exactly how a user might describe a given concept internally. Therefore, testing language will also reveal users' mental models of blog linking to some extent.
The goal is to use terminology that makes the linking model as transparent as possible, so users can easily understand how it works. If one set of terms seems to work much better, then we will use that set in the interface. However, even if no phrase pair out performs the other, the test will not be a failure, as it will show that either both sets of terms are difficult to understand or that there is not a universal mental model to describe blog chains and the tasks of moving up and down through them.
32 participants, all with a baseline familiarity with blogs. We will then use a screening process to try to map them on a continuum of blogging expertise, which will be a factor of the participation in blogs (writing, reading, or commenting). We will then split the group of 32 into two groups along the median of this continuum.
Blocking
We will randomly assign participants to 2 different versions of the language, resulting in 2 groups of 16. However, these groups will contain an even number of people from each of the groups formed when splitting along the continuum of usage (as described above) to ensure that not all novice or expert users end up in the same condition.
Each group will get 4 different tasks, one of which has language to match the interface. Thus, groups of 16 are subdivided into 4 groups of 4, each one seeing a different combination of the tasks.
Example for one between-subjects group:
| 1 (m) | 4 | 3 | 2 |
| 2 | 1 (m) | 4 | 3 |
| 3 | 2 | 1 (m) | 4 |
| 4 | 3 | 2 | 1 (m) |
Note that the between subjects groups of 16 would not get the exact same tasks, but they would get the same "pattern" in terms of where the matching task appears.
Interface Language to Test
X response(s) to this entry (existing)
Y links in this entry (existing)
X links to this entry (proposed)
Y links from this entry (proposed)