Personal Attribute Judgments - Modeling Users' Experiences with Interactive Systems - page 36

Ta b l e 2 . 4 Number (and percentage) of attributes being adequately modeled by the resulting
views of the averaging analysis and the two proposed procedures.

Analysis Procedure No of views No of attributes (%)

Averaging analysis 1 65 (43 %)

Procedure proposed in this chapter 5 62 (41 %)

Procedure proposed in chapter 3 2 85 (56 %)

Ta b l e 2 . 5 Number of attributes that are adequately modeled by the resulting one or two views
for each participant.

Subj. Description view a view b remain

1 User 1 6 - 1

2 User 2 10 - 1

3 User 3 - - 13

4 User 4 6 4 1

5 User 5 - - 10

6 User 6 6 - 2

7 User 7 9 - 3

8 User 8 6 - 4

9 User 9 10 - -

10 User 10 9 - 1

11 User 11 6 - 3

12 Concept Developer 1 11 - 1

13 Concept Developer 2 6 - 4

14 Concept Developer 3 6 - 1

15 Documentation Expert 7 - 1

16 Interface Designer - - 2

17 Market Expert 8 - 1

18 Project Manager 7 - -

19 Software Expert 1 6 - 1

20 Software Expert 2 5 - 2

21 Usability Expert 1 - - 9

22 Usability Expert 2 6 4 -

Third, the technique did not explicitly optimize the goodness of fit of the diverse
models for the analyzed attributes. As a result, while the technique succeeded in
identifying the differences between users and designers in terms of overall dissimi-
larity, it might fail in accounting for more attributes than the traditional “averaging”
analysis. In the next chapter, a new procedure is described that explicitly aims at in-
creasing the number of attributes that are adequately modeled based on two criteria
introduced below. In this procedure, two goodness-of-fit criteria are defined for as-
sessing whether or not an attribute is adequately predicted by a given model: a) the
amount of variance R 2 in the attribute ratings accounted for by a given model, i.e.