DESIGNS THAT FIT - Human-Computer Interaction and Management Information Systems: Foundations - page 201

Figure 10.2 is of course overly simplistic. In order to place fit within an HCI design paradigm,
it would need to be evaluated alongside other general design principles. Here are some concerns
that future research will need to address (in no particular order):

1. Fit comes at a cost: a cost to the user and a cost to the developer. The latter is straight-

forward—fit demands more design and computing resources. The cost to the user is subtler.
Consider the guideline recommended to IBM on adapting presentation according to previ-
ous navigations. The loss of consistency may reduce perceived control and cause a sense of
discomfort (Te'eni and Feldman, 2001). Such trade-offs, as articulated convincingly in the
case against consistency (Grudin, 1989), underscore the need to study fit in the context of
other design principles.

2. The interaction between types of fit is unknown. Consider, for example, the relationship

between informational fit and tool fit. Borrowing again from models of human-human
communication, one could postulate that in normal conditions informational fit would
be more important to the user, unless tool misfit crossed some threshold. If research
could substantiate such claims, designers would be better informed in designing adap-
tive systems.

3. Fit implicitly assumes task allocation between computer and human (see discussion

above). Past practices of ensuring effective task allocation are hardly considered today
in HCI methodologies. Although fit brings this issue to the forefront, it offers little in
understanding how to best allocate tasks. Research is needed to show how user charac-
teristics should be taken into account when allocating tasks, at the outset of the interac-
tion but perhaps also dynamically as the dialog evolves.

4. Computers are increasingly used to support communication and collaboration. Although

the discussion of fit is limited to the individual, the social and organizational contexts
should be considered in determining fit because the context may conflict or override the
individual aspects discussed above. For example, linking colleagues through continuous
video connections fits the cognitive need for realizing awareness and sensing the physi-
cal context that is lacking in virtual collaborations. However, in a field study in which
continuous video connections were established in public spaces, too, some workers felt
it conflicted with social needs such as privacy (Jancke et al., 2001). The challenge, there-
fore, is to learn how to study fit within a broader social and organizational context.

In conclusion, this commentary reviewed some isolated conceptualizations of fit and attempted
to bring them together into one framework shown in Figure 10.2. In doing so it examined how
current attempts can be complemented and their value to theory and practice strengthened.
Hopefully, it will serve as a catalyst for future research on fit in HCI.

ACKNOWLEDGMENTS

I'm grateful to Noam Tractinsky, Jonathan Grudin, and Dennis Galletta for their insightful remarks.

NOTES

1. To avoid confusion, I do not use “cognitive fit,” which is Davern's original label.
2. The four types are somewhat similar to Norman's classification of representations noted above.