iScale: Studying Long-Term Experiences through Memory - Modeling Users' Experiences with Interactive Systems - page 94

In the second part, participants were interviewed about the differences between
the three graphing techniques, using a structured interview technique, called the
Repertory Grid (Fransella et al., 2003). This technique is well aligned with parallel
design, and particularly in the existence of three or more alternative artifacts, and
allows for inquiring into participants' idiosyncratic ways in which they differenti-
ate the artifacts. In this way, one can inquire into the design space from a users'
perspective (Hassenzahl and Wessler, 2000). This is useful not only for designed
artifacts, but also for methods as in the present case. Participants were given three
cards, each providing a name and a screenshot of one of the three graphing tech-
niques. Participants were first asked to identify the three techniques. Next, they were
asked to “think of a property or quality that makes two of the graphing techniques
alike and discriminates them from the third”. They were instructed to feel free to
make any combination of the three alternatives. Contrary to common practice with
the Repertory Grid Technique, we did not probe participants for the exact oppo-
site of the property they provided, but rather focused on further elaboration, when
possible. This was supported by laddering and pyramiding techniques (Reynolds
and Gutman, 1988). Laddering seeks to understand what “motivates” a given prop-
erty and thus ladders up in an assumed means-ends-chain (Gutman, 1982) towards
more abstract properties of the stimuli; in laddering we first asked the participant
whether the mentioned property is positive, and subsequently why this property is
important to him/her (e.g. “why is expressiveness important to you?”). Pyramiding,
also known as negative laddering, seeks to understand the lower level properties that
make up for a given property; in pyramiding we asked the participant to elaborate
on what makes the given technique to be characterized with the respective property
(e.g. “what makes free-hand graphing more expressive?”).

5.3.2 Analysis and Results

5.3.2.1 Understanding Free-Hand Graphing

We analysed FHG to get an idea of how people actually graph changes in product
perception and evaluation. We segmented the collected graphs in discrete units. A
unit was coded when two conditions were met: a semantic change in the partici-
pant's verbal report following a pause in graphing as observed in the video recorded
sessions (e.g., “[pause in graphing] but then I got to the point where I got new soft-
ware updates”). Pauses often suggested an initiation of a new recall. Often this was
combined with a change in the slope of the curve, but this was not always the case.
Each unit was then coded for the type of curve and the type of verbal report.
Curves were classified into four categories: a) Constant (C) signifying no change
in participant's opinion over a certain period, b) Linear (L), either Increasing or
Decreasing, c) Non-linear (NL) when there were no grounds for arguing that the
curve could be approximated by a linear one or when a single report was associated
with two discrete linear curves of different slope (see 5.4b), and d) Discontinuous
(D) when the slope was approximately parallel to the vertical axis.