Museum exhibits hold great potential as places of learning (e.g., Falk and Dierking, 2000; Rainoldi et al., 2018). From a theoretical perspective, unfacilitated exhibits offer opportunities for constructivist learning, where learners build their own understanding by integrating new ideas from the exhibit with their existing cognitive resources (e.g., Allen, 2004; Bransford et al., 2000). Additionally, when accompanied by friends, family, or significant others, exhibits offer social opportunities in which learners can come to understand new ideas through engagement with both the exhibit and their companions (e.g., Ash, 2003; Blud, 1990; Oler et al., in press). However, learning through exhibits, given that such learning is self-directed, relies on substantial engagement with exhibits (Allen, 2004), which can vary across types of exhibits and learners (e.g., Shaby et al., 2017). Specifically in the context of museum exhibits, there has also been a growth in technologically advanced and interactive displays. A precursor to learners' constructing new knowledge within museum exhibits, then, relies on prolonged engagement. To afford visitors learning opportunities through exhibits, museums have the difficult task of presenting scientifically or historically accurate information to a diverse audience, many of whom spend only brief moments engaging with any given exhibit. Consequently, museums must optimize exhibit design to effectively capture visitors' attention and convey key information (Allen, 2004; Garbutt et al., 2020; Mayr et al., 2009). In Figure 1, we show a museum visitor going through an exhibit while wearing a mobile eye tracker. The mobile eye-tracking technology can be quite useful at various stages of exhibit development and maintenance.

During the initial stages of exhibit development, pilot data can provide useful information in making decisions about exhibit design. Eye trackers can indicate where and for how long museum visitors are engaging with particular sections of exhibits. Existing research on scene perception provides useful knowledge about where and how long visitors look at a section of an exhibit is linked to visual attention (Kümmerer and Bethge, 2023; Nuthmann, 2017; Nuthmann and Canas-Bajo, 2022; Williams and Castelhano, 2019). During the piloting process, this information can be valuable in ensuring that the visual behavior of visitors aligns with the plans for the exhibit. Similarly, initial eye tracking data can be collected to identify portions of the exhibit that are hard to follow or are less engaging to visitors and can be improved before the exhibit is opened to the public. This information can be tremendously useful in streamlining the finalists for the exhibit, as it can help to parse which aspects of an exhibit can foster high levels of engagement (e.g., Shaby et al., 2017). Hence, the mobile eye tracking methodology can enhance the exhibit development process that prioritizes intrinsically motivating exhibits that sustain interest and engagement among visitors, thereby supporting learning toward intended outcomes (e.g., Allen, 2004; Garbutt et al., 2020; Mayr et al., 2009).

Furthermore, mobile eye trackers can also be adopted to improve the exhibit over time to make sure additional information can be assimilated into the exhibit for continued learning. For example, applying this methodology to the climate change context, we adopted this technology to identify portions of the exhibit that were less visually engaging to museum visitors (Oler et al., in press). The exhibits team can rearrange their exhibit with some replacements or relocations and test if the engagement with key exhibit details continues to occur. This is supported by a study that examined the viewing behavior before and after the artwork was rearranged in its permanent collection at the Austrian Gallery Belvedere (Reitstätter et al., 2020). This effort demonstrated the benefits of mobile eye-tracking in examining differences in exhibit design (Reitstätter et al., 2020). Use of mobile eye tracking methodology allowed researchers to quantify the differences in the attention of individual pieces of artwork and the exhibit on the whole before and after a redesign.

Additionally, mobile guides are an increasingly common way to provide visitors with personalized content that aligns with their interests and motivations for visiting an exhibit. They increase the time visitors spend on each exhibit (Mokatren et al., 2016). Mobile guides can be improved through the implementation of eye-tracking to suggest what information should be supplemented to the user and when (Mokatren et al., 2016). For example, someone who is trying to deepen their understanding of climate change might spend more time reading signs, which could trigger a mobile guide to suggest other supplementary materials. In contrast, someone who is learning about climate change for the first time may spend more time looking at images and might benefit from a mobile guide that allows the viewer space to process the information and suggests helpful ways to regulate emotional reactions to climate change. Therefore, eye trackers can be used to better understand how most people are spending their time in an exhibit and what supplementary information might be useful in a mobile guide.

In addition to being helpful in setting up and improving exhibits, the eye tracking methodology can be used to inform research in applied settings (Fu et al., 2024; Hou et al., 2025; Silva et al., 2019). Eye-tracking can provide objective, real-time, and unobstructed ways of studying visual engagement and related informal learning that can continue outside formal learning (e.g., schools; Jarodzka et al., 2021). Depending on the research questions of interest, additional measures can be assessed to understand their connections to exhibit content. For instance, before, during, and after visitors experience the museum, more information can be collected around affective experiences (positive vs. negative), memory (e.g., learning outcomes), traits (e.g., personality), identity (e.g., political affiliations), attitudes and beliefs (e.g., “is climate change real?”), interest and motivation, etc. These data can collectively inform understanding of the factors linked to engagement with a particular exhibit. Instant and delayed questions can also be built into the study protocol to investigate how learning outcomes are connected to visual engagement and other individual difference variables relevant to the research questions.

On a related note, eye tracking data can be simultaneously collected with additional measures in real-time to gain a comprehensive understanding of museum visitors' experiences. For example, as shown in Figure 1, a visitor is also carrying a digital tablet in her hand and, when prompted, can report questions specific to the exhibit or personal experiences in response to the exhibit. Such an ecological momentary assessment approach can be utilized to gain insights into the diverse experiences of visitors Lohani and Blodgett, (2025). Furthermore, video and audio information is also collected in real-time and synced with fixation data. Building on this existing concept, additional measures can be synchronized with eye tracking data, including physiological measures (e.g., heart rate and skin conductance).

Mobile eye tracking technology, as compared to stationary eye tracking, has its own advantages because it allows for a much more naturalistic and objective observation of the viewers' experience. While stationary eye-tracking can provide a wealth of information that is applicable to exhibits, it still lacks compatibility with the true experience of visitors. Advanced mobile eye-tracking technology has been shown to be just as accurate when participants are dynamic (e.g., walking) as when they are stationary (Onkhar et al., 2024); however, research has also shown that a chinrest should not be used with a mobile eye tracker as it can adversely impact accuracy (Onkhar et al., 2024). Furthermore, the recent mobile trackers have better accuracy (e.g., Tobii Glasses 3 was better than the previous model due to their advanced illuminators and camera placement; Onkhar et al., 2024) for conditions like walking, which is relevant for museum settings. It can also be easily used with minimal training, making it accessible to anyone interested in adopting it in exhibit work (Bulling and Wedel, 2019). Eye tracking software has also improved substantially in recent years and has now made it much easier for users to process collected data, i.e., quickly extract usable and interpretable information (Bulling and Wedel, 2019).

A key draw of eye tracking is that it provides a way of capturing visual attention in real-time to content of interest, such as text panels, images, videos, and interactive panels. Until recently, the primary way of assessing engagement was through observations, testimonials, interviews, and questionnaires (Museums Association, 2017). With the advances in technology, mobile eye-tracking allows for real-time measurement of continuous engagement while visitors experience and exhibit, which provides a powerful way of assessing engagement and interest without utilizing self-reports from visitors (which can have limitations). Additionally, one of the key benefits of mobile eye tracking is being able to provide within-person analysis by seeing how individuals' chosen behaviors and interactions with environments change their gaze patterns (Fu et al., 2024). Several mobile eye-tracking metrics provide objective measures of visual engagement with an exhibit in real-time, enabling researchers and gallery designers to have the information they need to make decisions about specific items in the exhibit. We utilize a climate change exhibit to elaborate on this point further.

After these data are mapped to the exhibit environment, they can be used to create different visualizations. Visualizations are a practical way to qualitatively learn about visual engagement patterns. These are some of the most beneficial ways to communicate results as they allow viewers to visually understand the underlying trends (Bojko, 2009). In general, most software accompanying a mobile eye tracker can typically create four common visualization types: heat maps, opacity maps, scan paths, and bee swarms. These visualizations can be created for a group of exhibit visitors (as done in Figure 3 using pilot data from 50 visitors) or for an individual visitor at a time to see their engagement with relevant portions of an exhibit (as presented in Figure 5). The heatmaps (Figure 3a) and opacity maps (Figure 3b) both provide information on where the majority of gaze has fallen, i.e., visualizations capturing the density of visual attention. The heatmap visualization provides information through a gradient of colors overlaid on the exhibit image. The regions that attracted the most attention are overlaid with colors to indicate the degree of visualization, and typically, the warmer colors mean the highest level. Inversely, the opacity map visualization blacks out most of the image, and the areas with the most gaze are the most visible. The scan path (Figure 3c) and bee swarm (Figure 3d) visualizations provide information on gaze progression in chronological order, i.e., visualizations capturing the timeline of visual engagement. The scan path visualization takes each point where the gaze rests and adds numbers indicating order. The beeswarm visualization type depicts each individual movement of the gaze chronologically. Bee swarms are usually presented as animations and are most applicable to demonstrate temporal behavior (Blaschek et al., 2017). These are best viewed in video format, as they display the chronological sequence of engagement across portions of an exhibit.

Each type of visualization has its own benefits and is appropriate for different situations. Within the museum context, density visualizations using heatmaps and opacity maps (Figures 3a, b) are best used to convey key areas of engagement of visitors. This allows exhibit designers, curators, and researchers to be able to immediately gather information about what is and is not interesting for visitors. Additionally, heatmap and opacity map visualizations can be created using different metrics; the primary types are fixation count, absolute gaze duration, relative gaze duration, and participant percentage maps (Bojko, 2009). For example, a museum researcher might choose to use a participant percentage heat map when communicating how many visitors will notice a key piece of information that they suspect to be underutilized. This allows for easy and intuitive communication of the areas that are most often viewed. Opacity maps and heatmaps are able to both communicate the same data metrics, however differ in their presentation. Opacity maps can more clearly demonstrate the areas of key relevance to viewers (Špakov and Miniotas, 2007). However, they may not be suitable for viewers unacquainted with the stimulus image, as the darkening effect can prevent first-time viewers from understanding the content of the stimulus. Whereas, heatmaps are able to show a more total view of visual behavior.

As seen in Figures 3a, b, both the heatmap and opacity map display that the majority of gaze lands on the two peaks of the graph and the text in the snapshot. Both these visualizations work well with data from a large number of people to get a sense of regions of the exhibit that received the most visual attention, and if the critical regions of the exhibit received the engagement the exhibitors would like, as this would help design effective exhibits. These visualizations can also help identify areas that are overlooked to help the designers improve engagement in line with their goals for the exhibit. Similarly, researchers could also gain an understanding of regions that museum visitors attend and engage with. However, if the interest is not in regions of engagement and rather in the temporal aspects of engagement, heatmaps and opacity maps are not relevant.

On the other hand, the scan path and beeswarm (Figures 3c, d) are timeline visualizations best used to understand what the visitors are drawn to and in what order, specifically to understand where they are drawn first in a scene. This allows curators and exhibit designers to determine if what was intended to be seen first is actually seen first. This is most important for the design of a linear exhibit when the order of visual engagement is critical. Individual scan paths and bee swarm visualizations can be created for each visitor to examine the order in which visitors interact with the whole exhibit. Typically, the scan path and beeswarm visualizations are best when a handful of participants' data are utilized. This is obvious by comparing the visualizations in Figures 3c, d created with data from 50 participants vs. Figures 5c, d that has data from only one participant. The latter is more interpretable and useful to understand the temporal path that a museum visitor took and the areas they engaged with and for how long. Therefore, the scan path and beeswarm visualizations are helpful in pilot testing how museum visitors interact with exhibits and in what order, which can be incredibly useful to make decisions about exhibit design and understanding visual behaviors. However, beeswarms are often not suitable for portrayal as static graphics because they lack the numbering system used in scan paths. They also lack the scan path's ability to demonstrate fixation duration (represented by the size of each dot). However, a key weakness of scan paths and bee swarms is that they are highly susceptible to becoming overwhelming and cluttered when overlaying data from multiple sources. Notably, to avoid creating misleading visualizations, it is best practice to include the fixation parameters (e.g., the number of milliseconds and the degrees of the visual field) and the conditions that generated the visualizations (e.g., heatmap color scale, metric depicted, and viewer task; Bojko, 2009).

Visualizations provide an informative qualitative approach and should be complemented with quantitative analyses. One of the most common metrics is fixation, which is a time period when the eyes' gaze is relatively still within an area and stays there for a predefined minimum duration, usually representing intake of visual information and engagement (Rayner, 2009; Russo, 2019). The area and minimum duration to count as a fixation are predefined before processing data; for a detailed consideration (see Hooge et al., 2025; Kümmerer and Bethge, 2023; Startsev and Zemblys, 2023). Existing eye tracking processing software from reputable companies and scholars typically provides the most common algorithms for detecting eye movements, fixations, and saccades, which is a good starting point. However, any processing choices should be made after considering the existing literature and all efforts to make this process transparent and replicable are encouraged (Kümmerer and Bethge, 2023). Once the fixations are identified, follow-up metrics like those listed in Table 1 can be extracted using the software (Bylinskii et al., 2017). One of the most common fixation metrics is the fixation count, which is how many times the visitor fixated on a particular area of interest (i.e., the number of fixations). Another important metric is the fixation durations, which is a measure of the time period of visual engagement. In addition, it can be useful to examine which areas attracted a visitor's attention, and this information can be provided by the time to first fixation (in seconds), which captures how long it takes for the visitor to look at each area of interest. With respect to the museums, the number and duration of fixations provide a way of objectively evaluating engagement with relevant areas of the exhibit, often connected with the importance of an area. Thus, an exhibit designer, curator, or researcher can examine the critical areas of interest within an exhibit and evaluate how visitors are paying attention to them.

Another relevant metric for exhibit purposes is saccade. Saccades are swift eye movements between fixations and are often associated with search patterns (Nuthmann and Canas-Bajo, 2022; Phillips and Edelman, 2008; Rayner, 2009). The duration of the saccade varies by the goals at hand. For example, scene perception has a longer saccade duration than reading (Rayner, 2009). Saccades duration and length have been utilized to study cognitive processes such as fatigue (Di Stasi et al., 2014) and memory performance (Ohl et al., 2024). In the context of exhibits, saccades can be helpful in learning about how visitors scan an area of the exhibit (e.g., what they look at first and what order they scan different aspects of the exhibit). Similar to visualizations, it can provide useful ways of understanding visitors' interest and behavior. Thus, these metrics can be easily adapted to exhibit contexts as they have been utilized in studying user experience and usability behavior (e.g., Novák et al., 2023; Shen et al., 2020). A commonly adopted method for identifying eye fixations and saccades is the Velocity-Threshold Identification (I-VT) method Salvucci and Goldberg, (2000), which is utilized commonly by custom software, including Tobii Pro Glasses 3 (2020). This method compares the velocity of eye movement to pre-defined time and area thresholds to categorize saccades and fixations (Salvucci and Goldberg, 2000).

Distinctions have also been made between direct and indirect operationalization's that can be made through eye movement data (Hooge et al., 2025). Direct operationalization involves accurately measuring eye movements, such as categorizing fixations and saccades, which are possible through research-grade eye tracking hardware and software and implementation of best practices recommended by experts in the field (Hooge et al., 2025; Kümmerer and Bethge, 2023). Indirect operationalization is more complicated as it includes utilizing eye movements (e.g., fixation location and duration and saccades) as proxies for psychological processes (e.g., perception, attention, and learning). When making such indirect interpretations, the outcome depends on the definition of operationalization that is chosen by the researcher (Hessels et al., 2024; Hooge et al., 2025) and is contingent upon the environment, thus requiring caution (Mayr et al., 2009). For instance, fixations are most often associated with learning processes and engagement; some ambiguity can remain while making these indirect interpretations (Henderson, 2003; Hessels et al., 2025; Hooge et al., 2025; Negi and Mitra, 2020). As in some contexts, fixation count has been shown to be significantly related to the level of confusion (Salminen et al., 2018).

The difficulty of interpreting eye movements remains a continued issue within the field of eye-tracking research today (for details, see Hessels et al., 2025; Hooge et al., 2025). However, a thoughtful research process can be quite useful in making careful interpretations. For instance, there are opposing effects of mental workload that are due to perceptual load (that can lead to shorter fixation durations and more frequent fixations) vs. cognitive load (that can lead to longer fixation durations and less frequent fixations; Liu et al., 2022), and can help make interpretations about how visitors may experience workload at different stages of processing. Furthermore, recent reviews of existing literature on utilizing eye tracking methodology in the context of education to make interpretations about cognitive processes (Liu and Cui, 2025) are also a helpful resource for newer studies to learn about relevant findings to make informed indirect operationalizations. Improvements in the interpretation of cognitive processes tied to mobile eye tracking remain an important direction for applied mobile eye tracking research.

Above, we discussed limitations around the interpretability of eye tracking metrics. Transparency in sharing decisions made for computing eye-tracking parameters in every study remains essential as this will promote comparison of findings across studies (Hooge et al., 2025). While research in museum exhibits via eye tracking methodology is limited, much foundational work can get established with additional research. Even though one advantage of ambulatory eye-tracking studies is that they are naturalistic and provide high external validity, considerations of proper internal validity are necessary as well. For instance, while some eye trackers maintain stable gaze estimates despite slippage, jostling, or bumping during movement, others experience a decline in accuracy under these high activity conditions (Niehorster et al., 2020). It is important to consider the data quality metrics reports shared by the mobile eye tracker company to get a sense of their accuracy (e.g., Tobii AB, n.d.). Similarly, the accuracy of automated mapping of gaze data can be inaccurate and ineffective, especially with dynamic and open environments (Hahn and Klein, 2023) present in museum exhibits. Even though automatic mapping can help reduce manual mapping, that is time-consuming and effortful, it is worth noting that a manual examination is still necessary for quality control and minimal inaccuracies. Notably, mobile eye-trackers may be unsuitable for studies involving small AOIs or very few fixations, as the automatic gaze mapping performed by mobile glasses software often lacks the precision needed (Hahn and Klein, 2023). For instance, they may be unable to detect word-level differences in fixation duration. Additionally, small changes in environmental stimuli can have large implications on visual attention and behavior (Kiefer et al., 2017), but it can be nearly impossible to control every aspect of the environment in an exhibit setting. Other visitors, noise level, and exhibit busyness can play a role in how a person chooses to interact with an exhibit, but can be difficult to control, especially when the interest is in naturalistic museum settings. If possible, perhaps less busy times of the day could be targeted for data collection. Researchers should be encouraged to manage such environmental variables when possible, but to otherwise accept this limitation in order to allow for the naturalistic benefits of eye-tracking.

When designing ambulatory eye-tracking studies, it is crucial to consider several limitations. First, mobile eye-tracking devices are designed to be worn by individuals with normal vision and without any obstructions to their eyes. Participants who wear corrective lenses are often excluded from eye-tracking studies because eyewear lenses reflect the infrared light used for measuring eye position, potentially interfering with the accurate tracking of the pupil (Carter and Luke, 2020; Stuart et al., 2016). However, some mobile eye-trackers have recently introduced attachable lenses (sold for an additional cost) that correct for common forms of near and farsightedness (Tobii AB, 2020). However, for those museum visitors with complex prescriptions, this limitation may currently still exist as the attachable lenses cover generic prescriptions only. For similar reasons, drooping eyelids, long eyelashes, dark makeup, or cataracts can also result in increased eye-tracking errors (Carter and Luke, 2020; Holmqvist et al., 2011). Since older adults are more likely to wear corrective glasses or to have other eye abnormalities, excluding people according to these criteria could compromise the randomness of the sample imposing significant barriers to research focused on older adults, and it remains an important consideration for researchers.

Another barrier that requires consideration is the ethical challenge at play when a person is wearing an ambulatory eye tracker in a public space. Appropriate information should be shared with those who are not a part of the study for them to decide if they prefer not to be around an eye tracker while it is recording data. The requirements set by the institutional review board will dictate these conditions too. Researchers could collect data in off-hours when others are not around to bypass this issue as well. However, additional considerations would be needed if the visitor wearing the eye tracker may have to interact with other visitors who are not wearing an eye tracker or are not involved in the research (Mayr et al., 2009).

An additional consideration for utilizing mobile eye trackers is the environment. Mobile eye tracking's greatest strength is its ability to allow researchers to investigate non-laboratory environments and produce data in applied settings. However, some environments present additional struggles because of light and overheating. The issues with light are twofold. Firstly, with significant levels of glare, participants will begin to squint, lowering data quality (Armato et al., 2013). Additionally, sunlight can lower data quality because of competing infrared light, worsening the efficacy of the embedded illuminators (Imabuchi et al., 2016). However, some new generation eye trackers are more prepared to deal with this challenge. For example, the Tobii Pro Glasses 3 can be outfitted with protective lenses that allow for protection from glare. Work is also currently being done to improve the ability for eye trackers to be used outdoors in high-light conditions (Rusnak et al., 2025). Though current generation eye trackers have approved operating temperatures up to 113°F/45 °C, at high temperatures, overheating becomes more common, especially when exposed to direct sunlight (Tobii manual, 2020).

In the current mobile eye-tracking landscape, there is a trade-off between cost and user-friendliness. The easiest-to-use eye trackers provide software that transforms enormous data files of pupil position and cornea reflection into interpretable gaze paths and fixation points, even providing helpful visualizations. However, these eye trackers can be very expensive. On the other hand, less expensive options require the researcher to analyze and interpret the raw data themselves, which requires advanced computer coding skills to create a process that is automated by custom software. However, with some consideration, it will be possible to use existing custom software with cheaper eye trackers as well. In fact, some open-source software are already available (Nasrabadi and Alonso, 2022). At the same time, with advancing technology, ongoing and future work may make ambulatory eye-tracking more cost-effective by using participants' smartphones (Gunawardena et al., 2024; Valliappan et al., 2020). Additional guidelines on tools available to conduct eye tracking are also available (see Niehorster et al., 2025).

While determining the mobile eye tracker, the nature of the tasks participants would be doing would matter. If the dynamic nature of the task includes standing and walking, the mobile trackers are about to accurately measure visual gaze; however, with tasks like skipping and jumping, the accuracy can drop (Hooge et al., 2023). Mobile eye-trackers like Tobii Glasses 3 have been found to be suitable for walking conditions, but the accuracy can drop for high movement conditions such as jumping and skipping (Hooge et al., 2023). Experimenters are encouraged to adopt eye trackers with a higher sampling rate if higher movement is expected, as it is found to handle fast movement better (Hooge et al., 2023).

Despite the busy environment of an exhibit, planning ahead on how to use mobile tracking can make its adoption highly feasible. While the data collection will happen in the exhibits section, the setup can be done in a separate room. If possible, finding a less noisy room can be helpful. The team conducting the study may benefit from having a checklist for the before, during, and after steps for getting all done seamlessly. Before data collection, it is important to have the mobile tracker and laptop fully charged. If informed consent is needed (depending on the study and institution), the required procedures can be completed before starting the setup. Depending on the eye-tracker, having a breathable bag to keep the small recording unit (as shown in Figure 1) is helpful. Also, it helps to inform the participant of certain behaviors that may disrupt data collection quality (e.g., refrain from placing fingers on the lens of the eye tracker).

Before the data collection starts, it is important to calibrate the trackers (which allow the tracker to learn where the person is looking). After calibration, it is also a good practice to make sure that the tracker is actually doing a good job tracking the participants' eyes by asking them to look at exact locations and checking on the software that the real-time recording is accurate. Such validations can be very helpful in collecting accurate and valid data. Also, it is helpful to have a reminder to hit record (often, that is another step); otherwise, offline data won't be available. Any other instructions (where to go in the exhibit, etc.) can also be shared before starting the data collection because it is best to make it as least disruptive as possible. Before starting with the main exhibit region, it is helpful to have some “practice” time built in so that participants get used to the setup and are more naturally interacting with the exhibits of interest.

The researcher should also consider their interest and need to be able to monitor the live feed generated by the mobile eye-tracker. Generally, the recording does continue even when the live feed may be discontinued due to the eye tracker being out of range because the recording unit is with the participant along with the eye tracker. The Wireless Local Area Network (WLAN) will limit the maximum distance the researcher can view the live feed on the computer receiving the wireless signals from the mobile tracker. The maximum distance to be able to receive the live feed is dependent on the environment and is informed by factors such as high radio traffic. The range is more outside than indoors. Additionally, the live view can reconnect when the eye tracker comes back into range allowing researchers to stop recordings when data collection is finished.

After eye tracking is complete, the participants could be brought back to a quiet space where the setup was initiated for post-exhibit data collection (e.g., interviews and surveys) and debriefing purposes. Improving the protocol with multiple pilot participants before collecting the main data is also quite useful. As expected, all used products should be cleaned before being used by another participant. Additional technical information is available for adopting eye-tracking in research and analysis (for more information, see Carter and Luke, 2020; Hessels et al., 2025; Kasneci et al., 2024; Rayner, 2009).

The eye-tracker calibration process can present additional limitations. Mobile eye trackers require calibration each time a new wearer uses them to ensure accurate measurements. This calibration process aligns eye movements with the visual field, enabling the device to map gaze to perceived objects (Slone et al., 2018). While this process is normally straightforward, it is helpful to know how to execute the calibration process properly (Nyström et al., 2013). In general, calibration errors can be reduced by presenting the calibration target at the participant's eye level and making sure they are at a proper distance away from the calibration target. If using eye-tracking software, it can sometimes be helpful to restart the software prior to calibration if multiple error messages occur. This is because after sitting idle, the communication between the glasses and the laptop unit can be disrupted, and restarting the system fixes this issue by re-establishing the connection. Furthermore, additional suggestions are available to collect high-quality eye-tracking data by paying attention to factors such as corrected vision, eye characteristics, and makeup (Nyström et al., 2013). Another recommendation is to conduct re-calibration, especially when the study is long or participants are constantly moving around (Shen et al., 2020). This can be done in between sessions by asking participants to look at specific locations to verify that the eye tracker is accurately detecting the gaze of the wearer in real time.

It is also helpful to consider data analysis plans before starting the main data collection. Significant progress has been made in properly analyzing eye-tracking data, and custom software is available to assist in this process. The visualization features provide rich information to exhibit curators, designers, and researchers to understand how visitors interact with the exhibits. Offline indices can also be created, but they may need the support of a research collaborator familiar with these additional features. However, most eye-tracking companies have been providing additional resources to make this viable.

At the same time, certain aspects of the analytical process still need further consideration. A primary consideration is which areas of the exhibit are of most interest for the project at hand. In order to maximize generalizability, a two-step process is effective. First, the team should determine which areas in the exhibit are of interest to examine, i.e., the AOIs. These are informed by the rationale for collecting data and which regions matter the most. Second, as a follow-up process, feedback can be collected from others to evaluate if the chosen areas meet the needs or if additional regions should be edited. If there is room to make this process more objective, ratings can be collected for each of the regions on the dimensions of interest, such as relevance or interest. While designing the areas of interest, the software often allows one to load the same template across participants (so that variability in drawing them does not add noise). Failing to follow best practices for the creation of AOIs could lead to misinterpretation of each AOI (Orquin et al., 2016; Hooge et al., 2025). It is a good practice to conduct frequent quality checks using scan path visualizations to look for issues such as missing data (suspiciously few scan paths; Holmqvist et al., 2011). Finally, pilot studies are an effective way to identify many data collection and analysis issues and are highly encouraged (Hessels et al., 2025).