Personal Attribute Judgments - Modeling Users' Experiences with Interactive Systems - page 28

John sends 3 documents
for print

He collects his documents
and returns to his office

He places his finger on the touch sensor,
and all his documents are getting printed

Fig. 2.4 An example storyboard describing one of the concepts used in the study.

layers of interaction with the printer: the option of user identification at the printer
(Direct print - Scroll list - Pin code), different user identification mechanisms (Scroll
List - TouchToPrint - Badge), and the option of document selection at the printer
(TouchToPrint - Direct print - Touch & Select). The order in which the triads were
presented was counterbalanced between participants.
For every triad, participants were asked to “think of a property or quality that
makes two of the products alike and discriminates them from the third” . Designers
were instructed to think of ways that would be meaningful to users. Users were
defined as employees at a university department. This process was repeated until a
maximum of six attributes were elicited for each triad.
After attribute elicitation, participants were asked to remove duplicate attributes
and rank the remaining attributes according to their importance. Finally, participants
were asked to rate all products on their personal attributes, as well as on preference
and dissimilarity . In contrast to the traditional Repertory Grid approach, we em-
ployed paired comparisons instead of semantic differentials, as this was a priori
expected to deliver more stable results (de Ridder, 1996). While in the semantic dif-
ferential technique only one product is being rated and thus being compared to an
implicit reference point, in paired comparison two products are being compared on
a specific attribute. Out of the six products one can form up to n(n-1)/2 = 15 pairs.
To reduce the number of pairs we employed a balanced incomplete design (Sandt
Van de, 1970; Furlan and Corradetti, 2006) with 9 total pairs and every of the six
products participating in 3 pairs. According to this design, the total number of pairs
can be calculated from equation 2.1:

n 2

=

TotalNumberO f Pairs (2.1)

where n is the number products and λ the number of pairs in which we want each

product to participate in.