Field Trial

1
DESIGN ASPECTS OFMOBILE USER INTERFACES
Pekka Parhi Department of Electrical and
Information Engineering University of
Oulu pekka.parhi_at_ee.oulu.fi Wireless Cities
2006 Oulu, Finland
2
Mobile vs. Desktop UI Design

• Designing UIs for mobile devices is quite
  different from designing for fixed terminals
• Different characteristics

3
Mobile Applications

• Context of use
• Dynamic, unpredictable
• Often crowded and noisy
• Users attention divided among several tasks
• Supported activities
• Frequent, short-duration, immediate
• Focused few specific tasks can be carried out
  very well

4
Mobile Devices

• Physical issues
• Limited screen space
• Limited input capabilities
• Simple interaction
• Awkward text entry
• Technical issues
• Processing power, memory, storage space, battery
  life
• Challenges with wireless networking
• Slow, intermittent connection

5
Presentation Focus

• One-handed interaction
• UI design to support one-handed
• use of mobile devices
• Target size study for one-handed thumb use on
  touchscreen devices

6
Why One-Handed Interaction?

• One hand occupied
• Attention divided among tasks

• Unstable environment
• Two handed use unnatural

7
Interaction on Existing Handhelds

• Smartphones

• Input Hardware Buttons
• Interaction Keypad-mapped functions
• Directional navigation

• PDAs

• Input Touch Sensitive Display Hardware
  Buttons
• Interaction Software targets for
• direct manipulation
• Directional navigation

8
Suitability for One-Handed Use?

• Smartphones

• Compact form, proper-sized keys
• Interaction limited to keypad mapped menus and
  directional navigation
• Not efficient nor user-friendly

Touchscreen UIs for one-handed interaction?

• PDAs

• Touchscreens offer greater flexibility for UI
  design
• UIs traditionally designed for pen-based,
  two-handed interaction

9
Related Work

• Interfaces for One-Handed Use on PDAs
• AppLens LaunchTile
• Karlson 2005
• Command-Based Gestures
• Direct Manipulation Gestures
• MessagEase Nesbat 2003
• Scalable Keypad for Text Entry
• Thumb-Based
• Hardware

10
Related Work

• Thumb-Based Hardware (cont.)
• Microsofts Ultra-Mobile PC (formerly Project
  Origami)

11
Related Work

• Thumb-Based Hardware (cont.)
• Touch key phone
• (NTT DoCoMo Mitsubishi)

12
Direct Thumb Interaction

• Limited Screen Space
• UI targets should be as small as possible without
  degrading performance and user satisfaction
• No previous target size studies for one-handed
  use on small touchscreen devices
• Studies exist for desktop-sized displays and
• pen-based interaction on mobile devices

• Optimal Size for Targets???

13
Target Size Study

• Objective

• To develop guidelines for targets that
• maximize performance and user preference
• during one-handed thumb use on
• small touchscreen devices

14
Two-Phase Study (1/2)

• Phase 1 Discrete Targets
• Single-target selection tasks
• Similar to clicking a button or
• selecting a menu option

• Phase 2 Serial Targets
• Multi-target selection tasks
• Similar to text entry

15
Two-Phase Study (2/2)

• Participants
• 20 (17 Male, 3 Female)
• 18 regular cell phone users
• 6 regular PDA users
• All right-handed
• HP iPAQ Pocket PC used
• Tasks performed standing
• Total time 40-45 minutes
• Including instruction, both phases and
  questionnaires

16
Phase 1 Discrete Targets (1/2)

• 5 target sizes (3.8, 5.8, 7.7, 9.6, 11.5 mm)
• 9 locations (screen divided into a 3x3 grid)
• Each target size tested 5 times per location

17
Phase 1 Discrete Targets (2/2)

• (1) tap green button, (2) tap actual target x
• North ltgt South movement
• Measures
• speed, accuracy, hits distribution, user
  preferences

• Constant distance between green button and target
  x
• X surrounded by distractors
• Lift-off selection strategy
• Auditory and visual feedback

18
Phase 2 Serial Targets

• 5 target sizes (5.8 13.4 mm), 4 locations
• (1) Tap green, (2) Enter 4-digit code, (3) Tap
  END
• Measures
• speed, accuracy, user preferences

19
Results Discrete Targets (1/3)

• Speed
• Differences between all sizes
• were statistically significant
• Error Rate
• No difference between 9.6
• and 11.5 mm target sizes
• Significant differences
• between other sizes
• Location had no effect on either speed or accuracy

20
Results Discrete Targets (2/3)

• Hits Distribution
• Hit area increased with target size
• Users trade off speed for tap accuracy
• Right-leaning trend for targets on the right side

(white boxes buttons dark gray boxes area
enclosing 95 of hits per location gray dots
successful hits black dots erroneous hits)
21
Results Discrete Targets (3/3)

• User Preferences
• Center region was the easiest
• Objects on the left side and bottom right corner
  were the hardest

Mean comfort rating for each region (1-7 7
most comfortable)
Mean of the smallest comfortable target size in
the region
22
Results Serial Targets (1/2)

• Speed
• Differences between all sizes
• were statistically significant
• Error Rate
• 5.8 mm differed significantly
• from target sizes 9.6 mm
• No difference between
• other target sizes
• Location had no effect on either speed or accuracy

23
Results Serial Targets (2/2)

• User Preferences
• NE region was the most
• comfortable
• Minor differences

24
Discussion

• Speed continued to improve significantly with
  even the largest targets in both phases
• No difference in error rates with target sizes
• 9.6 mm (discrete) and 7.7 mm (serial)
• Error rates were much higher in serial tasks than
  in discrete tasks (9.6 mm target 5.0 vs. 2.8)
• Limitations of the study
• One posture used for performing tasks (standing)
• One touchscreen-equipped mobile device (PDA)

25
Conclusion

• Target size recommendations for one-handed use of
  touchscreen-based handhelds
• 9.2 mm for single-target pointing tasks
• 9.6 mm for multi-target pointing tasks
• Recommendations based on error rates data along
  with user preferences
• As well as hits distribution data for
  single-target tasks

26
References

• Parhi P, Karlson A, Bederson B (2006)
• Target Size Study for One-Handed Thumb Use
• on Small Touchscreen Devices.
• Proc. MobileHCI 06, Espoo, Finland, to appear.
• The presented study was done in UMD during Fall
  2005.

27

• Thank You!
• Questions?