Empirical Study of Prototype Attributes: Fidelity and
Medium
A 2 x 2 empirical study of early stage prototype attributes'
effects upon user testing results, using the following working
definitions of "Fidelity" and "Medium" attributes:
Fidelity - the amount of similarity between a prototype
and the final product, including characteristics such as methods of
interaction, visual appearance, level of detail, content, etc.
High-fidelity prototypes are more similar to the final product than
low-fidelity prototypes.
Medium - the type of platform on which the prototype is
presented (i.e., on a computer screen using mouse and keyboard
methods of interaction, on a piece of paper using Wizard-of-Oz-ed
methods of interaction).
Motivation & Methodology:
- DENIM
is a prototyping tool designed for web designers that creates
prototypes combining the prototype attributes of low-fidelity and
computer medium, breaking the more traditional coupling of
low-fidelity with paper medium and high-fidelity with computer
medium.
- This study sought to investigate the separate factors
of fidelity and medium in terms of their effects upon user testing
results.
- We ran a user testing experiment in which participants carried
out several tasks on two different websites under two different
fidelity conditions (low-fidelity and high-fidelity) and in two
different media (paper and computer).
Procedure:
- Early-stage online banking websites were created, purposefully
incorporating usability problems in order to elicit a large number
of comments from users.
- Each participant tested the two websites in the same fidelity
but different media.
- Participants performed five typical online banking tasks:
setting up an account, e-mailing themselves a statement, setting
up automatic bill payments, finding the value of foreign currency,
and moving money between accounts.
| |
Low-fidelity Prototype |
High-fidelity Prototype |
|
Website A |
|
 |
|
Website B |
|
|
Data:
- We recorded participants thinking aloud and prompted them to
make additional comments at the end of each task.
- Behavioral data (e.g., confusion or frustration) was noted and
followed up with users at the end of the task, as were unusual or
incorrect paths.
- Comments were extracted from the behavioral data.
- Similar comments were then categorized into issues.
- Issues were categorized into Nielsen's heuristics by web design professionals.
Analysis:
- We then tested the independent variables of fidelity
and medium in a factorial design, repeated-measures experiment.
The 2-by-2 design counterbalanced low- and high-fidelity, computer
and paper media.
Findings:
- Low- and high-fidelity prototypes are equally good at
uncovering usability issues.
- Usability testing results were also found to be independent of
medium, despite differences in interaction style.
The Take-Home Message
Designers should choose
whichever medium and level of fidelity suit their practical needs
and design goals, as discussed in this paper:
Publication
Walker, M., Takayama, L., and Landay, J. (2002) High-fidelity or low-fidelity, paper or computer medium? Proceedings of the Human Factors and Ergonomics Society 46th Annual Meeting (in press).
PDF (465 K) | HTML
Human Factors and Ergonomics Society Presentation
PPT (2874K)
|
