Continuing in the ‘How to set up an eye tracker’ series, I’ll now briefly discuss some pointers to consider before you set out buying a tracker.
Research before buying. Find some people who do what you do, and send someone over to work with them for a bit. Post-grads (grad students) and post-docs are particularly useful for this kind of task. More often than not, they’ll come back with loads of useful experience, as well as hints and tips on how to do things-these are normally the kind of hints and tips that don’t end up written into the Methods sections of papers.
Lab Space. Get ready to be extra nice to your colleagues if you need to negotiate some new lab space. Eye trackers tend not to be tiny, and need a dark room with plenty of space for several computers and chairs.
Spatial Accuracy. Make sure you get a tracker with a decent spatial accuracy, otherwise you won’t be able to be certain that the participant was looking where you thought.
Temporal accuracy. This one is very important as well. Temporal accuracy is measured in Hertz (Hz), and describes how many times per second the tracker can sample eye position. So, the higher the Hz, the better the tracker. If you want to do gaze-contingent work, don’t get a tracker that is lower than 250 Hz (i.e., 250 samples per second). The tracker I use, the Eyelink 1000, can go as high as 2000 Hz.
Eyelink 1000 Lab #1 (School of Psychology, University of Southampton)
Monitors. Get a CRT monitor for displaying your stimuli. Flat screens aren’t appropriate, because of the way they refresh the display (if you’re a decent experimental psychologist, you’ll probably know this already though). Sadly, the only type of people who want CRTs nowadays are vision scientists, so you’ll probably have to go on a massive hunt for one that’s appropriate (I recently travelled with a colleague on a 4-hour road trip to buy some monitors that were sold on Ebay–fortunately we weren’t charged for them though!). Don’t get anything smaller than 19 inches. 21 inches is much nicer, because participants will be some distance from the monitor in many cases. Make sure that the monitor goes to 100 Hz at least with a decent resolution – 140 Hz with no flicker would be better. You’ll probably have to check through a lot of CRTs before you find one that is good enough. See below:
Some dead monitors
Another dead monitor
Even more dead monitors!
Chin rests. Use chin rests, head restraints, and anything else at your disposal. This may sound a little terrifying, but it’s vital that you keep your participants as still as possible! The smallest fidget can throw the tracker off. Make sure the participants are comfortable when you do this, of course.
Checking and Support! Check with others who use the tracker that you are using. Go on the forums for the company who you are buying a tracker from. If they don’t have forums, email them and see how good they are at replying. This is crucial as well. I’ve seen so-called ‘support’ forums run by some experimental software companies where staff simply provide links to the documentation relating to someone’s question, even when the individual has stated that they have read the documentation and can’t make any sense out of it. It’s both useless and infuriating.
Though this is not an exhaustive list, it might be useful as a set of pointers to get you thinking.