Eye Tracking for Usability
A Primer for User Experience Professionals
Agenda
Introduction
Eye Trackers
When is Eye Tracking Appropriate?
Using Metrics
War Stories
Q & A
Introduction
Hi!
Dan Berlin, MBA, MSHFID
2 years researching neuromarketing methods- eye tracking, biometrics, neurology, emotions
Eye Tracking Vendors
Two main players: &
Tobii▪ Based in Sweden, offers the same equipment for
scientific research, & has assistive technology products
SMI▪ Based in Germany, offers high-end & integrated
equipment for scientific research
Eye Trackers
1750 T60/120 T60 XL X60/120 Tobii Glasses
• 1750 & X60: old technology is old• T60 & XL: depends on your needs• Glasses: brand new• depends on IR markers• only 30 Hz• small DVR
• You probably don’t need 120 Hz• Tobii studio and Axure wireframes do not play nicely together
Eye Trackers
• RED• 60/120 Hz (also have 250 Hz model)• Use a screen up to 300”
• iView X HED• Up to 200 Hz• Uses a notebook or subcomputer• No IR markers
RED iView X HED
• Germaphobes: gotta clean that hat
• Software advantage: moving AOIs & better statistical analysis
When is Eye Tracking Appropriate?
The age old question… In usability studies
NOT during think-aloud▪ It is natural for a participant to look at the moderator▪ And they will look at parts of the screen that they are talking about
Does retrospective think-aloud alleviate this?▪ It asks participants to remember what they were unconsciously
thinking▪ More likely: primacy and recency effects (Michael Summers, TrueAction)
Allocate a few tasks to eye tracking where the user does not think-aloud
Avoid bias: make up a story for the calibration
When is Eye Tracking Appropriate?
In benchmark studies Comparing user behavior to different design or interaction
concepts No think-aloud, just explore the site▪ Can be done with a static composition or a wireframe▪ Compare, compare, compare – there are no benchmarks
Use metrics to determine if participants are looking at areas of interest▪ Not all AOIs are equal – some should be more important
to the business Static pages: 10 to 20 second exposure▪ Otherwise: the big red blob – they look everywhere
Eye Tracking Output
The typical outputs from eye tracking: Fixations & duration Time to 1st fixation Gaze plots & heat maps Areas of interest (AOIs) Pupil dilation
But first, some information from a study published the Journal of Advertising Research
Attentive vs. Emotive Advertising
Measured fixations per second (FPS) for two types of television ads Attentive = top-down processing Emotive = bottom-up processing
They found that the different types of ads were just as effective at different levels of attention (FPS) Attentive = high # of fixations Emotive = low # of fixations
So, we must consider what we are testing when examining the data Interactive Web sites necessarily require high attention
Heath, Nairn, & Bottomley. (2009) “How Effective is Creativity? Emotive Content in TV Advertising Does not Increase Attention.” Journal of Advertising Research. 49(4). 450-463.
vs.
Eye Tracking Metrics
Fixations vs. duration Basically, they are the same▪ Both measure levels of active attention and cognition
We will never know if an increased duration indicates confusion or interest
Fixations per second is the traditional measure of active attention
Gaze plots and heat maps Eye candy and not much else – but clients love them Bolster your eye candy with data!
Pupil dilation Impossible to measure accurately – don’t use it
Ok, so how should I use eye tracking metrics?
Use areas of interest to compare metrics How many fixations are in (un)important AOIs?▪ Will determine if an important AOI needs more emphasis
How do fixations in similar AOIs compare between different design treatments?▪ Will determine which design better achieves business
goals
How long does it take participants to get to a particular AOI? (time to 1st fixation)▪ You only have a few seconds to impress a user – are they
looking at that which you want them to?
Participant Recruitment
Make sure you ask about eye ailments Retina & cornea damage, eye cancer & tumors,
macular degeneration, cataracts, conjunctivitis, and nystagmus
Not necessarily problematic: amblyopia, glaucoma, and strabismus
If possible, you want to use the data from everyone you bring in Add questions to your screener to ensure you
can eye track your participants
War Stories
The most frustrating thing about eye tracking: losing the signal You’ll get a good calibration, the data starts off nicely,
then the user changes position and <poof>, there goes the eye tracking
What do you do?
“Since you are watching where I’m looking, I was trying not to look in some areas.”
Sometimes, the eye tracking data will inform you which participant should be shown the door
Creep Map
• 1 minute exposure
• These are the only hotspots on the entire page
• When asked why this design comp was given a low rating, the response: “because she’s fat”
• Oy vey
Q & A
What are your eye tracking experiences?
What have you found to be useful or annoying?
What other metrics do you use?
Thanks for having me!