Back

How to conduct an eye tracking study

Eye tracking used to be an expensive and difficult technology, but now it can be easy, affordable, and highly beneficial to your business.

We perceive most of the world around us using our eyes. That is why the interest in studying human behaviour through the observance of eyes is anything but new. The first study based on observing the eyes’ movement dates back to the 1800s when Louis Emile Javal made direct observations and found that eyes make quick movements mixed with short pauses when reading a text.

Since his research, a lot of progress has been made in the eye tracking field and technology has allowed the methodology to become extremely precise and versatile. Eye trackers, webcams, and even the selfie camera of the smartphone can make eye tracking accessible and easy to use for all kinds of purposes. A well-conducted study can provide the researcher with relevant information regarding perception and visual attention that would not be possible to obtain using other methodologies.

However, mastering the art of eye tracking research can be challenging, especially if you are a beginner. To help you, this article guides you step by step through the process of conducting an eye tracking study, from designing a test to interpreting the data acquired.

1. Designing a test

It is important to determine the purpose of the test to get effective results (Source)

The first step is crucial for the success of the study. An eye tracking study analyses visual attention of the participants. They perform a specific task that requires them to absorb information through the eyes, such as looking at a product package, or a social media ad, finding a product in e-commerce, using a mobile application, etc.

Let’s say, for example, that you are launching ane-commerce app and want to find out if people see the sale banner placed on the main page. Based on a well-thought research question, you can formulate a hypothesis that reflects your assumptions regarding the users’ behavior, such as "banners with people in it are more attractive than those with just graphic elements". This step will enable you to test predictions and analyze the results more easily.

After defining assumptions, you select the essential metrics to track and measure whether your assumptions indeed reflect the users’ behavior. Next, you create the tasks and the visual stimuli the participants will interact with. Deciding whether there should be a time limit for each task is essential to guarantee a reliable result. If the aim is to capture what attracts immediate attention,then defining a short time limit makes a lot of sense as you only capture the immediate orientation of your participants. To test an ad in the context of social media, for example, it is ideal to limit the task to a few seconds because usually people look at an ad on social media while scrolling the screen. If you set a time limit that is too long, you always risk capturing idle gaze as people look around because they cannot proceed to the next task, rather than because they are interested in the content.

Conversely, when you use thinking aloud in parallel to the eye tracking, it is important to keep in mind that participants need time to voice their opinion, so time limits, if there are any, should be more generous than when no thinking aloud is employed. For a UX task, on the other hand, it can be counterproductive to establish a time limit because the person will need to stay on the website or app as long as necessary to finish the assignment.

It is usually extremely advantageous for web and mobile eye tracking research to conduct studies with different versions of your website, application, packaging, newsletter or advertisement, or with competitors’ websites , known as A/B testing. "A", the control, is the version that confirms the hypothesis. In our example, it would be the banner with people in it. "B", the treatment, is a modified version to challenge the assumption. By running comparative studies, you will be able to evaluate which elements work best.

The study can be conducted in-lab or remotely, moderated or unmoderated. In-lab studies demand more time and resources, and also are limited to participants who can join the study in person. All the eye tracking devices and software must be provided in the lab and the study must be monitored by researchers and facilitators.

Remote studies can be conducted with participants from all over the world, which is ideal to obtain more diversity and suits companies who have clients located in different cities or even countries. Remote studies can be done unmoderated with the help of tools that can collect and save the data. All the researcher needs to do is to send out the invitation and relax while the technology does all the work.

Once you have designed the test and implemented the study in the tool of your choice, it is time to think about the people who will join your study.

2. Recruiting participants

Recruiting the right participants to take part in your study is essential to obtain reliable data. A common mistake many companies make is to get in-house participants to join the study to save costs, but this often affects the effectiveness of the study. Only draw participants from the company if the product will be used internally. Otherwise, invest in looking for a more diverse group of participants.

For studies where researchers will compare heatmaps and quantitatively analyze the amount of attention that an Area of Interest (e.g. the logo on a package, a specific banner on website, a button on an app or newsletter, or a product in a video) receives, it is strongly recommended to recruit at least 30 participants, as these require many participants to generate more satisfactory results.

Suppose you don’t want to spend resources with so many participants using this quantitative method. In that case, it is more productive to conduct a qualitative study and watch the individual video replays while listening to what they say with the thinking aloud method. For this kind of study, you should have at least six participants joining. Reviewing the content, watching it in slow motion and taking notes can be particularly valuable to generate hypotheses, as well to observe which features work and which ones need improvement. They are more indicated as primary data than heatmaps and AOIs.

Knowing how to recruit the right number of participants for each study and the adequate demographics guarantees that your results will be more accurate and reliable. But keep in mind that regardless of the method you use,watching the video replay and taking notes is essential to understand the data better.

3. Running the test

Now that you have designed a test and recruited the participants, it is showtime! Depending on your objective and resources, you can opt for an in-lab moderated test or a remote online test.

In-lab moderated test

When the session is in-lab, the facilitator should make the environment as comfortable and genuine as possible to get more precise results. They also have to adjust the participant’s position and eye tracking calibration, which demands time from both researcher and participant. 

During the session, it is indicated to have two experimenters participating, one facilitator and one to take notes and make observations. Start the session by thanking the participant, explaining briefly what the study is about, and making sure your explanation does not influence their actions. An experienced facilitator is required to disclose information without making the participant nervous or hesitant.  

Next, guide the participant through the process. Make sure to say there are no right or wrong answers, and they should perform a task as natural as possible. Tell them to communicate as soon they finish each task to ensure the task timing and task success are right. 

After they finish the test, you can provide additional questionnaires and ask further questions about the tasks. Do not forget to get the consent form from all the participants.

Remote unmoderated test

Unmoderated tests allow testing in the participant's home (Source)

Often remote eye tracking tools have the instructions and calibration displayed automatically, and this step can be done in less than five minutes.

Before the test, communicate clearly how the test works and what participants should do to perform the tasks correctly. Some remote research tools, like Oculid, offer a feature to introduce the participant to the study and give general instructions, so the researcher can focus on the results.

After having completed these steps, just send out the invitation and wait for the results to be uploaded to the platform. After the participants finish the eye tracking tasks, you can redirect them to additional questionnaires.

4. Looking at the results

Last but not least, let’s look at the results. The metrics available in eye tracking can vary from tool to tool. Most tools offer qualitative as well as some quantitative results. What is more relevant to you highly depends on the type of study that you run. Some of the most used metrics to measure eyes’ movements quantitatively are Time to first Fixation, First Fixation Duration, Dwell Time, Revisits, amongst many others.

To learn more about the primary metrics used in eye tracking and their interpretation, check out this article that shows the data you can collect in research and what you can draw from them.

Conclusion

Eye tracking is a fascinating technology that can offer profound insights into human behavior and decision-making. It is a technology that can offer benefits for Market, UX and academic research, medical and gaming fields, amongst many others. Using it correctly can enrich your research and provide your company with information that would not be possible to get with other methods.

If you want to conduct your own eye tracking research using just the selfie camera of the smartphone, contact us today to get a demo of our platform.

Market

January 20, 2022