User Eye Motion Handheld

User Eye Motion with a Handheld Personal Digital Assistant: A Pilot Study: 

User Eye Motion with a Handheld Personal Digital Assistant: A Pilot Study Robert Krull, Binod Sundararajan, Michael Sharp, and Liza Potts October 1, 2004 PCS Rensselaer Polytechnic Institute

Project Goals: 

Project Goals Demonstrate issues in: Screen layout of PDAs Collecting eye motion data for handheld displays Layout issues: Information density Color Icons and text Workspace and framework areas Eye motion issues: Similarity to desktop machines Feasibility for handheld displays

Background: Display of Information: 

Background: Display of Information Desktop displays (800 x 600 pixels) are noted for being crowded PDA displays are smaller (240 x 320 for the larger ones) and the pressure to crowd them is even greater Scrolling or splitting PDA information across several pages produces reading delays. Research indicates that PDA screens may not be overwhelmingly bad compared to phone-sized screens

Background: Social Implications: 

Background: Social Implications Elementary and high school education Games Applications Medicine Hospitals ePocrates Diagnostics and patient management Personal Beaming Time management

Research and Screen Size: 

Research and Screen Size Palmtops Ability to make smaller sizes Classroom use

Research and Screen Size: 

Research and Screen Size Ka-Ping Yee, 2003 Movable keyhole Larger virtual space 3-D calendar

Research and Screen Size: 

Research and Screen Size Testing out expandable Palms

Form and Content in Displays: 

Form and Content in Displays Normal displays contain both form and content. (We concentrated on form in this study.) Tullis metrics on form: overall density, number of groups, and size of groups, etc. Hornof's eye motion studies controlled for 'out of bounds' content. Krull and Rubens' eye motion study found user problems with high information densities, lots of color and semantic complexity.

Tullis Metrics: 

Tullis Metrics

Tullis Metrics: 

Tullis Metrics Overall Density: 65% Local Density: 60% # of Groups: 7, 4 Size of Groups: # of Items: Layout Complexity: 3

Tullis Metrics: 

Tullis Metrics Overall Density: 45% Local Density: 50% # of Groups: 7, 3 Size of Groups: smaller # of Items: Layout Complexity: 1

Eye Motion and Displays: 

Eye Motion and Displays Voluntary attention: reading, searching Involuntary attention: high contrast, movement, color, particularly in the periphery Faster than people can report verbally Several strategies and processes at once Where should people look?

Measuring Eye Motion: 1: 

Measuring Eye Motion: 1 Limits of resolution: ~ 2 degrees of visual angle; 0.4' is 15% of the PDA height and 19% of its width. 2 degrees Eye ~ 1 foot .4' Dimensions are not to scale.

Measuring Eye Motion: 2: 

Measuring Eye Motion: 2 The horror, the horror. Equipment has improved, but can still be balky. Small displays push the boundaries of eye tracking resolution. Tracks can be difficult to interpret.

Measuring Eye Motion: 3: 

Measuring Eye Motion: 3

Measuring Eye Motion: 4: 

Measuring Eye Motion: 4 A head-mounted ISCAN Baseball Cap Mounted Eye and Scene Imaging Assembly collected a continuous stream of data based on reflections of an infrared light from the cornea of the test subject. The equipment allowed test subjects to move their heads without disrupting data collection. The data stream was saved for later extraction of eye fixation points and statistical analysis. The recordings of horizontal and vertical eye movements were obtained by detecting the reflection from the left cornea of test subjects. Once the users’ point-of-regard was calibrated, we began data recording. Test subjects pressed the Page Down key or the space bar on a computer keyboard to be shown each task screen and a keystroke logger recorded the time.

Measuring Eye Motion: 5: 

Measuring Eye Motion: 5

Test Situation: 

Test Situation PDA screen shots were shown on a desktop monitor. Images were adjusted for viewing distance. 26 images were organized into 8 groups of related content, but groups varied in layout. 2 versions of the 26-image sequence Instructions and images were shown simultaneously in a PowerPoint presentation.

Welcome to the PDA Eye Motion Study: 

Welcome to the PDA Eye Motion Study Thank you for participating in our study. You will search for information in a series of PDA (personal digital assistant) screens. Instructions for each search will appear on the left and the PDA screen will appear on the right. When you have completed each search, press the spacebar on the keyboard to go on to the next search. When you are ready to begin, please press the spacebar.

Slide20: 

Under which category can you find cameos? Press the spacebar to go on.

Slide21: 

What program lets you see moving video? Press the spacebar to go on.

Slide22: 

Which program controls the wireless card? Press the spacebar to go on.

Slide23: 

Is flow control set to software or hardware? Press the spacebar to go on.

Slide24: 

How is the transmission rate set? Press the spacebar to go on.

Test Subjects: 

Test Subjects Graduate and undergraduate students Considerable computer experience 5 subjects for one sequence and 4 for the other (n = 9) 18,000 – 22,000 data points for each subject 700 to 900 data points per screen

Hypotheses: 

Hypotheses H1: Information densities  + response time (Test subjects should spend more time and more eye motion.) H2: Color  + response time H3: Chaotic arrangements  + difficulty locating text-entry fields

Navigation: 

Navigation How do people navigate PDAs? What does navigation look like? 2 degree approximation to some types of reading The eye does not 'mirror' the conscious think aloud Some 'straying' and anticipatory eye motions Eye-tracking

Findings: 1: 

Findings: 1 Users were surprisingly quick: ~ 10 seconds per search task Faster than the Krull and Rubens CRT subjects Considerable practice effect: faster with practice (r = -.63; p andlt; .01)

Slide29: 

Sample Eye Tracks: S4, Seq. 1, Screen 15, color

Sample Eye Tracks: S2, Seq. 2, Screen 15, b/w: 

Sample Eye Tracks: S2, Seq. 2, Screen 15, b/w

Sample Eye Tracks: S6, Seq. 1, Screen 24, b/w: 

Sample Eye Tracks: S6, Seq. 1, Screen 24, b/w

Sample Eye Tracks: S2, Seq. 2, Screen 24, b/w: 

Sample Eye Tracks: S2, Seq. 2, Screen 24, b/w

Findings: 2: 

Findings: 2 Density metrics: adaptation of Tullis metrics showed generally high densities (over 60% filled) Because of insufficient variation in densities, testing for the effect of density is not meaningful. Color screens were not consistently slower.

Findings: 3: 

Findings: 3 Complex layouts may produce a larger amount of scanning and more retrogressive scanning than do simpler layouts. But users seem to be able to scan extensively and yet quickly. Retrieval times and eye movement are likely to be affected by several different factors (task difficulty, screen design, and screen content). By adding eye tracking to other performance measures, we may gain a clearer understand of user behavior.

Conclusion: 

Conclusion Our eventual goal will be… Closer to in vivo use Continue with content Begin looking at performing tasks Eye-tracking (A first for eye tacking?): Spelling into thin air (Another first for eye tracking?): Spell checking

Future Research: 

Future Research Color Information Density Grouping and Layout Methods Usability Testing

Questions?: 

Questions?