iScale: Studying Long-Term Experiences through Memory - Modeling Users' Experiences with Interactive Systems - page 110

The latter, we thought, would signify the recall of experiential information (i.e., af-
fect), either through inferring this from recalled episodic information (Robinson
and Clore, 2002), or through recalling this directly from value-account Betsch et al.
(2001). These two metrics of consistency are not without limitations; future work
should expand to different facets of consistency.
Beyond the cost-effective elicitation of longitudinal data, this work provides sup-
port for the viability of survey methods that guide participants through a structured
process of data elicitation. A wealth of such procedures exists for face-to-face in-
terviewing. For instance, structured interview techniques such as triading (Fransella
et al., 2003) and laddering (Reynolds and Gutman, 1988) imprint a particular struc-
ture onto the data elicited by participants, which makes the data computational
friendly. Another example is Vermeersch' explicitation technique (see Light, 2006),
which employs a particular style of interviewing that aims at supporting the inter-
viewee to enter a state of evocation and “relive” the activity under consideration.
While such techniques, when used in face to face interviewing, are most power-
ful, they are labor-intensive and require skilled interviewers (Groves et al., 2009),
which always constrains the sample size of the study. Self-reporting approaches, on
the other hand, can have impact because one can survey large samples and, by that,
also inquire into rare experiences and atypical behaviors.
Obviously, iScale can as well produce large amount of qualitative information
that will require labor-intensive analysis given traditional qualitative data analysis
procedures like Content Analysis (Krippendorff, 2004; Hsieh and Shannon, 2005)
and Grounded Theory (Strauss and Corbin, 1998). Novel techniques from the field
of information retrieval (Landauer and Dumais, 1997; Blei et al., 2003) may prove
especially fruitful in automating or semi-automating the qualitative analysis process.
Finally, the interpersonal analysis of the graphs is definitely a subject for further
research and was addressed here only superficially.
iScale was motivated by a need for lightweight methods that provide insights
into long-term usage and related experiences. While the importance of temporality
has been repeatedly highlighted in user experience research (Forlizzi and Battar-
bee, 2004; Hassenzahl and Tractinsky, 2006), it has rarely been systematically
addressed. In two recent studies (Karapanos et al., 2008a, 2009c) we provided some
first evidence that not only our perceptions, but also the relative weight of different
product qualities change over time. So far, both academia and industry have largely
focused on initial use. This has strong implications for the quality of interactive
products. For instance, Den Ouden et al. (2006) found that an alarmingly increasing
number of returned products, in 2002 covering 48% of all returned products, are
technically fully functional (i.e. according to specifications), but they are returned
on the basis of failing to satisfy users' true needs (28%), or purely on users' remorse
(20%). These failures were not so much related to problems rooted in early inter-
actions - problems that can often be overcome through learning -, but to those that
persist over time, pointing at failures to truly incorporate the product into daily life.
We hope that iScale provides a first step towards retrospective elicitation methods