Presentation
(based on Shneiderman chapter 11)
[
lecture notes | CSC 397 | Pete Sanderson | Computer Science | SMSU ]
Table of Contents
Balancing Function and Fashion
Use of Color
Display Layout and Design
Error Messages
Anthropomorphism
Resources
Chapter 11 of Designing the User Interface Third Edition, by Ben Shneiderman.
Balancing Function and Fashion
This is Shneiderman subtitle for the chapter
If one is sacrificed to the other, most users will reduce or discontinue use of the system.
User perception of the system is important
Presentation styles are developed with the balance in mind
Several dimensions of presentation style are discussed
- Use of color
- Display layout and design
- Error message design
- Nonanthropomorphic design
Return to the Top
Use of Color
Color serves many purposes:
- Draw user attention (e.g. to warning)
- Distinguish between items
- Emphasize item groupings
- Evoke emotion in user
- Make the display more attractive (relative to audience)
Here are a few color guidelines from Shneiderman (p. 398-402)
Design in black and white. Layout, not color, should be first consideration. Use white space or lines to group and separate groups. Consider highlighting text in bold or italics.
Use color to differentiate items in dense layout. Or to allow greater density. Balance with desire to design in monochrome.
Color scheme should support task. If task OA have associated colors, corresponding interface OA should have same colors.
Be aware of traditional/cultural meanings associated with colors. Balance with coloring to support task.
Use dynamic color changes to support dynamic status changes. Depends on task, but will draw user attention to the status change.
Color scheme should be applied consistently. Otherwise, information may be misinterpreted (e.g. nonrelated items encoded in same color may be mistakenly grouped by user).
Limit number of colors to four in a display and seven in an application.
Consider user eyesight deficiencies. Consider colorblind users, and avoid color pairings that are incompatible in anyone's vision (too close or too far apart on color spectrum -- red and blue are far apart; it is difficult to focus on both at once).
Allow user control over coloring. Provide alternative schemes with easy selection mechanism.
The textbook web page for Chapter 11 provides many links to color usage guidelines.
Return to the Top
Display Layout and Design
Complex but important endeavor, since this is a key factor for UI and system success
There are benefits to performance and error rates:
- Both can be improved dramatically through effective display design.
- One form fillin system from Bell telephone saved 3.3 seconds per form by switching from narrative to structured display. Total savings over system life were nearly 80 person-years!
- Density seems to affect performance rates more than error rates.
Effective display design requires combination of:
- Task knowledge
- Graphic design,
- Principles of perception,
- and more
Most important overall guideline: design primarily for task, secondarily for real estate.
- Provide all necessary data in correct sequence
- Organize related items into groups
- Consistent and orderly formatting (such as placement and justification)
Style should be: (in alphabetical order)
- appropriate
- cohesive
- comprehensive
- consistent
- distinct
- elegant
Different displays for different types of user, even with same system, may be most effective. (e.g. expert users deal with information-dense displays much better than novices)
Some common guidelines for display design:
- Consistency is essential for multi-display systems: format, terminology, location
- Context is important for multi-display sequences, so user has idea of how far along he/she is
- Organize related items into visually separate groups
- Place items and groups on display in simple visual manner (columns, justification)
- Information density should be appropriate to task and user
- Data needed for user to make decision should be visible on display
- Many others too numerous to list (e.g., Shneiderman p 386).
It is possible to define and measure layout metrics.
Tullis study develops 4 task-independent metrics for character displays:
- overall density,
- local density (around each character,),
- grouping (e.g. size of character groups),
- layout complexity.
Example of a task-dependent metric: layout appropriateness
- How well users can follow linear sequence through UI to task completion.
- Lots of jumping around while specifying input, wastes time.
- This includes distances traversed by mouse.
Return to the Top
Error Messages
Frank and Ernest express the popular perception of error message design.
A new era in error message design? Haiku error messages
Reduce need for error messages through good design
- Input by selection rather than fill-in.
- Usability testing to discover and address frequent errors
Consider novice users, when designing error messages
- Anxiety about system
- Not confident of skills
- Not familiar with system
Poorly designed error messages lead to:
- Negative feelings about system (one negative experience can offset a lifetime of positive)
- Reduced use of system
- Reduced productivity
Software development process can address messages:
- Message guidelines
- Message QA
- Messages included in design
- Usability testing
Shneiderman guidelines for error message design:
- Provide specific information about what went wrong.
- Indicate how to correct the problem, if possible.
(Why not have system correct it automatically?)
- Use a positive tone, avoiding negative or violent terms.
- Avoid anthropomorphism (see below)
- Support user feelings of control over system.
- Provide error information with means for user to get more details.
- Use "user-centered phrasing".
- Allow user to control audible error signals.
Return to the Top
Anthropomorphism
Designing computer responses to appear as if they were coming from a human.
Recommended only for design of children's game and educational software
Reasons for non-anthropomorphic UI:
- Anthro deceives and confuses users: expectations are high if machine leaves impression that it has human-like intellectual capabilities.
- Clarify differences between humans and computers. We treat them differently.
- Reinforces perception that designer is responsible for errors, not computer.
- Anthro can make some users anxious about using computers.
- Anthro can externalize success: reinforces feeling that success is result of machine instead of user mastery.
- Easier to make user feel in control.
Many studies conducted.
- Compare first-person ("I"), second person ("you") and neutral (just the facts...) approach
- The "I" approach generally determined to be worse than the others.
Example: Shneiderman-directed study
- CAI for jr-high students
- Some got "I" interaction, others got "you", others got neutral.
- Each described using computers as "hard" or "easy" both before and after lesson.
- 5 (of 36) changed from "easy" to "hard". None were using the "you" interface.
- 2 (of 36) changed from "hard" to "easy". Both were using the "you" interface.
Example:
- Arithmetric drill software for 3rd graders (24 in all)
- neutral feedback (# right and # wrong) vs. judgemental positive feedback ("excellent").
- Latter did not improve performance or satisfaction
The word "you" can be used at least when user first begins session (especially for guidance). May change to impersonal thereafter.
Return to the Top
[
lecture notes | CSC 397 | Pete Sanderson | Computer Science | SMSU ]
Last reviewed: 23 November 1998
Peter Sanderson ( pete@csc.smsu.edu )