Eye tracking study recruitment — managing participants with vision irregularities

Considerations for test participants with vision irregularities and how it can impact your results.

Part 2 in our series of expert eye tracking advice.
This article comes from the years of experience Dr Tim Holmes has conducting a wide range of eye tracking research. Dr Holmes has a PhD in Visual Neuroscience and Attention from Royal Holloway, University of London, and is a valued creator of neuroscience programs for academic and commercial clients.

There's no point eye tracking if participants can't see properly – factor corrective lenses into your study design.

Did you ever calibrate a participant and start a recording only to have them say "oh, hang on a minute, I need my reading glasses for this"? Every study I've ever done with a decent sample size has thrown up something new or unexpected in the vision disorder category, let's begin with glasses.

First of all, a quick word on the types of corrective lenses out there. Single prescription lenses contain one type of lens and typically correct for myopia/nearsightedness (minus numbers on the prescription) or hyperopia/farsightedness (positive numbers on the prescription). Additionally, some might include corrections for astigmatism (blurred or "double vision" resulting from a defect in the lens or cornea) and vergence dysfunction (which requires a prism in one or both lenses to align the images from both eyes). Bifocals, as the name suggests, contain two different corrections, usually in the top and bottom half of the glasses, and typically have a quite noticeable split in the lens where the two have been glued together. Varifocals are the younger cousin of bifocals and whilst appearing to be a single lens the have a variable prescription from top to bottom. Corrective lenses also come with a variety of filters and coatings these day as well as photochromic reactive lenses (Transitions and Reactolite being the most famous brands) which adjust the amount of light making it through to the eye and can be an issue for vision research of any kind if you've taken time to control your stimuli!

Desktop (remote) infra-red eye trackers typically cope pretty well with single prescription lenses and only usually have a problem with them if: there are internal reflections being caused by lighting in the room, the glasses themselves keep moving on the participant, the lens correction is very strong (+/- 6 or more), or the frames somehow occlude the eye-image in the camera (a simple adjustment of the angle of the tracker can usually remedy this). They typically don't cope with bi- or varifocal lenses and this is because the lens distorts the shape of the pupil which causes detection errors in the eye tracking software. So, when recruiting for remote eye tracking studies, it's important that you always include questions about glasses in the screener and ensure participants are wearing the appropriate pair for the task BEFORE calibrating. Oh, and yes, offer them a lens cloth before you start calibration if the glasses look a bit grubby!

Mobile (or head-mounted) eye trackers are often designed to look like glasses which can make it difficult to use over the top of corrective glasses. If you're likely to be researching a cohort (such as the elderly) with an increased prevalence of corrective lenses, it's worth investing in the magnetic clip-on lenses with +/- corrections that come with some eye tracking glasses.

Oh, and by the way, contact lenses also come in bi-focal/varifocal form these days. So, whilst contact lenses are usually OK with all kinds of eye-trackers, the multi-prescription lenses CAN cause problems, so it's always worth checking!

Given a significant portion of the population requires some level of vision correction, it's a factor that needs to be considered. I read plenty of papers with the statement "corrected to normal vision" in them, but just how many of them have explicitly tested that assumption before starting data collection?

Tips for facilitating eye tracking studies with people who wear corrective lenses – and even those who don't... it's still possible they need them.

• If you see your participant squinting, it's almost certainly going to cause issues for the tracker and it suggests their vision is not corrected, so keep an eye on your participant during data collection.
• If you have access to one, use a Snellen or LogMAR chart at the distance of your stimuli to test your participant's vision before you start. You're not an optician and you're not going to diagnose their condition, but this is a great way to check their vision is as good as they claim it to be and so they should be wearing their lenses whilst doing this!
• Show them some example images/text on screen and get them to describe them to you or read out loud and you'll soon know if you're going to have a problem – because they'll squint, lean in to the screen or simply not be able to do it. If you can, have the eye tracker on for this to test how well the eyes are being detected.
• If you're testing print media, again give them an example and make sure they can read it at an appropriate distance for the scene camera you are probably using to record the text they are looking at.
• If you're using a desktop tracker, have a range of cheap reading glasses, covering different corrections – ideally from -6 to +6 diopters. When the inevitable happens and someone forgets their glasses you can offer them as a temporary solution and avoid having to discard the participant. These are typically available from large supermarkets or chemists.

Quickly, I just want to mention non-infra-red eye tracking. Here I'm talking about webcam-based systems particularly. Because these are highly susceptible to screen reflections on the glasses, they typically do not work very well at all with ANY kind of corrective lenses except for contacts. I don't talk much about these types of eye trackers, because they are typically not as accurate as infra-red trackers which more involved research requires, but they are used quite extensively these days and form a part of many home-testing panel studies for market research.

When screening subjects for market research, check color vision as well as oculomotor disorders.

Now that we've covered corrective lenses, there's another subject that warrants consideration, and although it has nothing to do with the eye tracking itself, it has a lot to do with the potential research questions being tackled using eye tracking. I'm talking about colorblindness and if participants have full color vision.

It's a tricky one to ask because plenty of people with a color deficiency either aren't aware of it, or don't want to admit it. Although this typically means losing around 10% of the population, it's especially important to screen out for market research or usability (user experience) studies where color is being used to make something stand out, convey warnings, or evoke emotional responses etc. Reading that back it sounds deeply prejudiced against those with color deficiency, but unfortunately the majority of design is in fact, for the majority, and so for most commercial users of eye tracking this is quite important. For those of us who strive for a more inclusive world though, people with color deficiency struggle with all kinds of user interfaces (ranging from Google maps, ATMs to websites) because color cues are favored over luminance or contrast cues which can often work just as well; so if you want to give your product designs a real workout maybe you should be deliberately screening for people with a color deficiency. Having piloted a lot of my PhD experiments on my partner who is red-green colorblind, I know for a fact that it can radically alter the results!