iScale: Studying Long-Term Experiences through Memory - Modeling Users' Experiences with Interactive Systems - page 111

as a viable, lightweight alternative to the expensive longitudinal methods. Only by
that, we can start to fully account for the notion of User Experience as a temporal
phenomenon.

5.6 Appendix - Temporal Transformation

In section 5.4.2.3 we used users' consistency of temporal information of reported ex-
periences across repeated recalls as a metric of reliability of the recall process. One
question, however, relates to whether participants' accuracy in recalling temporal
information remains constant across the full timeline, from the moment of purchase
of the product to the present time. The participant's accuracy might be affected by
the amount of contextual information surrounding the experience that is available at
the moment of recalling. Theories of recall have suggested that recent experiences
(Koriat et al., 2000), or experiences associated with important milestones (e.g. the
purchase of the product) (Barsalou, 1988) might be more easily accessible. If such
biases exist, they will affect the reliability test as differences in the consistency of
recalled information might be due to pertaining to more or less salient periods and
not due to the reconstruction process. In the presence of such biases, the tempo-
ral distance between the two coupled experience reports elicited in the two distinct
sessions should be transformed to account for the accessibility biases.

1.00 .00

R Sq Linear = 0.657
y = 1.47x + 0.27

-.50

.80

-1.00

.60

-1.50

.40

-2.00

.20

-2.50

-3.00

.00

-3.00 -2.50 -2.00 -1.50 -1.00 -.50 .00

.00 .20 .40 .60 .80 1.00

Log perceived time (x position on iScale timeline)

Perceived time (x position on iScale timeline)

Fig. 5.10 Relation between actual time (reported time for a sketched node) and position of
node along the iScale's x-axis: (a) linear relation, (b) power-law relation. Actual time (days)
is adjusted to the full time of ownership of the product for each participant.

We attempt to assess the existence of accessibility biases through examining the
way in which participants used the timescale of the tool, (i.e. iScale's x-axis). Par-
ticipants graphed linear curves through adding nodes in the graph (see figure 5.2a).
Each node can be characterized by two properties: a) the actual time (participants ex-
plicitly annotated for each node the approximate amount of days, weeks, or months
after purchase that this node represents, and b) the perceived time (the position of
the node along the x-axis of iScale).