Structure of presented stimuli influences gazing behavior and choice

Structure of presented stimuli influences gazing behavior and choice

Journal Pre-proofs Structure of presented stimuli influences gazing behavior and choice Attila Gere, Lukas Danner, Klaus Dürrschmid, Zoltán Kókai, Lás...

1MB Sizes 0 Downloads 3 Views

Journal Pre-proofs Structure of presented stimuli influences gazing behavior and choice Attila Gere, Lukas Danner, Klaus Dürrschmid, Zoltán Kókai, László Sipos, László Huzsvai, Sándor Kovács PII: DOI: Reference:

S0950-3293(19)30487-2 https://doi.org/10.1016/j.foodqual.2020.103915 FQAP 103915

To appear in:

Food Quality and Preference

Received Date: Revised Date: Accepted Date:

21 June 2019 21 February 2020 21 February 2020

Please cite this article as: Gere, A., Danner, L., Dürrschmid, K., Kókai, Z., Sipos, L., Huzsvai, L., Kovács, S., Structure of presented stimuli influences gazing behavior and choice, Food Quality and Preference (2020), doi: https://doi.org/10.1016/j.foodqual.2020.103915

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2020 Published by Elsevier Ltd.

Structure of presented stimuli influences gazing behavior and choice Attila Gere1, Lukas Danner2, Klaus Dürrschmid3, Zoltán Kókai1, László Sipos1, László Huzsvai4 Sándor Kovács4 1

Sensory Laboratory, Faculty of Food Science, Szent István University, Villányi út 29-43., 1118 Budapest, Hungary 2 School of Agriculture, Food and Wine, The University of Adelaide, PMB 1, Glen Osmond, South Australia 5064, Australia 3 Department of Food Sciences and Technology, University of Natural Resources and Life Sciences, Muthgasse 18, A-1190 Vienna, Austria 4Department of Research Methods and Statistics, University of Debrecen, Böszörményi út 138, H-4032 Debrecen, Hungary

* Corresponding author: E-mail: [email protected], [email protected]

Abstract Well-structured stimuli presentation is essential in eye-tracking research to test predefined hypotheses reliably and to conduct relevant gazing behavior studies. Several bottom-up factors associated with stimuli presentation (such as stimuli orientation, size etc.) can influence gazing behavior. However, only a small number of scientific papers address these factors in a sensory and consumer science context and thus provide guidance to practitioners. The two presented eyetracking studies on food images aimed at evaluating the effect of the bottom-up factors stimulus size, background of the picture, orientation of food product presentation, the evaluated products and the number of alternatives. Significant effects of product group were found in the case of all eye-movement parameters except time to first fixation and first fixation duration. In contrary, orientation significantly influenced only the time to first fixation and first fixation duration parameters. Stimulus size significantly increased fixation and dwell count, while background showed no significant effects. Furthermore, significant relationships were found between the number of presented images and eye-movement and decision time. Less time was needed in 2AFC (alternative forced choice test), 3AFC and 4AFC and significantly more time was needed to choose one alternative out of 7AFC and 8AFC. The results of the two studies show that the investigated bottom-up factors can significantly influence gazing behavior, and therefore need to be carefully considered when planning or comparing results of eye-tracking experiments.

Keywords: survival analysis; bottom-up factors; eye-tracking; decision time; food images

1. Introduction Well-structured stimuli presentation in eye-tracking research is essential to test predefined hypotheses reliably and in general to create relevant gazing behavior studies. Several bottom-up factors associated with stimuli presentation (such as stimuli size, orientation etc.) might influence gazing behavior. One of the most thorough collection of threats to the validity of eye-movement research was conducted by Orquin and Holmquist (2018), who describe internal and external threats. The authors define internal threats as inferences affecting eye-tracking studies, while external threats are referred as the dubious generalization of these inferences to new populations and stimuli. Within internal threats, inappropriate comparisons might arise when comparing highly different visual stimuli, e.g. bar plots and neuroimages. Analysis of multiple metrics might cause multicollinearity and careful data preprocessing (e.g. definition of areas of interests, AOIs) is needed. Data quality should also be checked for example by using capture rate. Total dwell time (TDT), the sum of all dwells, might cause misleading results, since TDT aggregates fixation and dwell counts, and/or durations and fixation likelihood. When choosing between fixed versus free exposure times, it is suggested not to control exposure time to eliminate time pressure, since each condition has different psychological interpretation. Assuming an eye–mind relationship might also be misleading since eye-trackers measure solely eye-movements and not the way of thinking. The connection between looking and thinking must be proven rather than using assumptions to validate their relationships. Among external threats, the authors mention undersampling of naturalistic stimuli might occur in eye-tracking studies using critical trials, e.g. in studies focusing on nutritional label use. The authors suggest 16 product variants to minimize bias (Orquin & Holmqvist, 2018). Characteristics of the group of individuals participating in the test might also influence the results. It has been shown that children can have a different gazing behavior than adults and elderly people also have a specific way of looking at pictures (Kemps, Tiggemann, & Hollitt, 2014; Spielvogel, Matthes, Naderer, & Karsay, 2018; Yasui, Tanaka, Kakudo, & Tanaka, 2019). However, it must be mentioned, that none of the cited papers focuses explicitly at the differences among various age groups. Age-related effects on visual attention and gazing behavior when viewing emotional faces have been reported by Isaacowitz et al. (2006). However, studies investigating age effects when viewing food items or choosing foods are still lacking. Also gender can be an influencing factor; it has been shown that females look differently at food compared to males (Hummel, Ehret, Zerweck,

Winter, & Stroebele-Benschop, 2018; Manippa, Padulo, van der Laan, & Brancucci, 2017). It has also been observed that obese individuals gaze at food with high energy density longer and more often than normal and under weighted persons (Joechl, Danner, & Dürrschmid, 2013; Joechl & Duerrschmid, 2011; Werthmann, Jansen, & Roefs, 2015). There might also be differences in gazing behavior between individuals from different cultures, for example Chinese participants proved to be more influenced by the background of the presented food stimuli than Americans (Zhang & Seo, 2015). Mental and eating disorders were also found to have an effect on gazing behavior (Baldofski, Lüthold, Sperling, & Hilbert, 2018; Frazier et al., 2016, 2018; Harezlak & Kasprowski, 2018; Schmidt, Lüthold, Kittel, Tetzlaff, & Hilbert, 2016). Therefore, recruiting well defined participants is especially relevant for eye tracking studies. The number of participants is usually considered as “the higher the better”. For consumer sensory tests an absolute minimum of 60 participants is suggested in order to fulfill the requirements of multivariate statistical methods (Næs, Brockhoff, & Tomic, 2010). However, if small effect sizes need to be captured, large sample sizes are needed, whereas smaller sample sizes might be enough to detect large differences. Besides the participant group, environmental factors can influence the measurements. In sensory science these aspects are well described and have been discussed by several authors (Gere, Kókai, & Sipos, 2017; Motoki, Saito, Nouchi, Kawashima, & Sugiura, 2019a, 2019b). However, only a few scientific papers address these factors in a sensory and consumer science context and thus provide guidance for users. Such bottom-up factors include the number and size of images on a stimulus, the contrast differences on the stimulus, the question type, the background color etc. Although eye-tracking has been used widely in consumer research (Duerrschmid & Danner, 2018 for a review), the influencing effects of such bottom-up factors on gazing behavior and decision making is still not fully understood. Studies have shown that products and their characteristics can also cause differences in gazing behavior. Several studies report that emotional attributes showed significant effects. In a recent study, it has been proved during a target-distractor paradigm study that tastiness and not healthiness captures automatic visual attention when food is presented as distractor (Motoki, Saito, Nouchi, Kawashima, & Sugiura, 2018). The significant influencing effect of food products was also identified in a study conducted with children. The study identified that unhealthy food captured children’s visual attention in a greater extent, compared to healthy food (Spielvogel et al., 2018).

It has been shown that the color of food products affected the sensory expectations and hedonic perceptions of consumers in a cross-cultural study between Austria and Thailand (Jantathai, Danner, Joechl, & Dürrschmid, 2013). These results raised the question for other bottom-up factors, that might influence gazing behavior. Vu, Tu, & Dürrschmid (2016) investigated the effect of number of alternatives (images) on participants gazing behavior across four different evaluation types (maximum choice, minimum choice, ranking and rating). Their results showed the number of alternatives as a bottom-up effect significantly affected the consumers’ visual behavior independent of the type of evaluation. During eye-tracking studies, usually a limited number of alternatives is presented due to the limits of the presentation screen. This creates a far less realistic situation compared to choices made during regular shopping, where the number of alternatives on the shelves is significantly higher. High numbers of alternatives usually make the choice task more complicated. Usually 2, 3 or 4 alternatives are presented, and conclusions are drawn based on these simplified choice sets. Two alternatives choice sets are easy to complete, tiring is low and more decisions can be made within a study compared to more complicated tasks (Manippa, van der Laan, Brancucci, & Smeets, 2019; Tórtora, Machín, & Ares, 2018). Three alternative tests are also frequently used but the study becomes more complicated due to the positioning of the stimuli. In order to provide symmetric positioning, one sample needs to be placed in the middle making the positioning of the fixation cross difficult (Bogomolova, Oppewal, Cohen, & Yao, 2018). Studies using three-alternative choice tests report inconsistent results. Some papers report that participants tend to choose the middle options from an array of similar options (middle choice preference) (Atalay, Bodur, & Rasolofoarison, 2012; Shaw, Bergen, Brown, & Gallagher, 2000), however, this is not always conformed (Missbach & König, 2016). The size of the stimulus (the alternatives) is greatly influenced by the presenting monitor and the used resolution, as well as the applied eye-tracker. The empty space between stimuli should be set higher when the eye-tracker works with lower sampling frequency (e.g. 60 Hz) due to the lower accuracy. Usually, researchers tend to use the whole screen when presenting alternatives (Bialkova, Grunert, & van Trijp, 2020; Bogomolova et al., 2018). There are also no standards regarding the suggested backgrounds during choice tasks. Some researchers use white background (Hummel, Zerweck, Ehret, Salazar Winter, & StroebeleBenschop, 2017; Spielvogel et al., 2018), but in other cases gray is used (Manippa et al., 2019;

Mitterer-Daltoé, Queiroz, Fiszman, & Varela, 2014), while a limited number of publications use black background (Motoki et al., 2018) during eye-tracking studies. However, the use of harsh colors is avoided. In order to get a deeper understanding of bottom-up effects, this paper presents two studies to i)

investigate the effects of stimulus size, picture background, stimuli orientation and product group on eye-movements and

ii)

investigate the effects of number of alternatives on eye-movements.

By a thorough analysis of the above-mentioned bottom-up factors, we aim to understand the degree of influence of the investigated bottom-up factors and to give suggestions how to control these effects.

2. Materials and Methods In order to analyze the effects of stimulus size, background of the picture, orientation of food product presentation, the evaluated product groups and the number of alternatives two experiments were designed. The first experiment aimed at evaluating the effect of background, stimuli orientation, size and product group using a mixed model nested design. The second experiment analyzed the effect of the number of alternatives (using multiple food choice situations from 2 to 8 alternatives) on gazing behavior. In both experiments a Tobii Pro X2-60 eye-tracker (60Hz) was used to analyze the eye-movements of participants. Confounding factors, e.g. noise and smells, were minimized during the experiments, which took place in the sensory laboratory of the Department of Postharvest and Sensory Evaluation at the Szent István University, Hungary. I-VT (identification by velocity threshold) filter method was used that incorporated interpolation across gaps (75 ms), reduced noise (median), used velocity threshold at 30°/s, and merged adjacent fixations (<0.5°) between fixations (<75 ms) and discarded short fixations (<60 ms). Participants took a seat at 65 cm distance from the 17-inch monitor (Samsung SyncMaster 757 MB) and were asked to move their head as little as possible throughout the experiment. Color calibration of the eye-tracker display was done using an X-rite Eye-One pro device as sRGB (gamma=2.2, CCT=6500 K). Stimuli were shown on the eye-tracker display (17 in., 1280 × 1024 pixels resolution, 75 Hz). Prior to the recordings, a 5-point calibration was completed using Tobii Studio software (version 3.0.5, Tobii Technology AB, Sweden). Between stimuli, a fixation cross

was presented in the middle of the screen to standardize the starting point of the gaze between participants. The most widely applied stimuli presentations use numbers of alternatives, which enable symmetric positioning (Danner et al., 2016) and also higher numbers of alternatives (Fenko, Nicolaas, & Galetzka, 2018). Product-related studies (e.g. studies focusing on the identification of the best/most preferred etc. products) require strict control of the participants (socio-demographic factors, psychological state, etc.). The presented study focuses on bottom-up factors, though product types are also connected to top-down processes (e.g. by liking, healthiness, etc.). Stimuli were randomized between participants and the following eye-tracking parameters were measured: Time To First Fixation (TTFF, time elapsed between the appearance of a picture and the user first fixating his/her gaze within an area of interest in seconds); First Fixation Duration (FFD, time a user gazes at his/her first fixation point in seconds); Fixation Duration (FD, length of a fixation in seconds); Fixation Count (FC, number of fixations on a product alternative as count); Dwell Duration (DD, time elapsed between the user’s first fixation on a product alternative and the next fixation outside the product alternative in seconds) and Dwell Count (DC, number of “visits” to an Area Of Interest (AOI) as count). FD and DD values are usually expressed as total (e.g. summed up the values of all respondents) or average (e.g. mean values of all respondents). In the present study, the average FD and DD values were calculated and used in order to eliminate individual differences.

2.1

Experiment 1

Experiment 1 was conducted to test the effect of size, background and orientation on five product groups on the gazing behavior. The effects of the three bottom-up factors (size, background and orientation) on gazing behavior were measured at three, three and six levels (Table 2), thus the total number of factor combinations in our design was 54. In order to introduce the factor product group, each participant conducted the tests on five different product groups, creating a random product group factor. Hence, product group is used as a nested factor whose levels are nested within the participant. The layout among product groups was randomized within participants in order to ensure proper randomization (Table 1). In this way the obtained design is considered as a mixed model nested design. ---Table 1 about here---

Backgrounds were chosen based on the patterns found in published papers as introduced in the Introduction section. Orientation was defined as all the possible orientations of the three product alternatives. Size was defined as follows: In case of large stimuli, the vertical or horizontal arrangement should fit into the 1280 x 1024 resolution of the presenting screen. Medium stimuli were defined as 30% of the large stimulus, while small stimuli were defined as 50% of the large stimulus. The space between the alternatives was defined as 150 pixels (Figure 1). The full factorial design of experiment required 54 participants, who were university students aged between 18 and 25 (53% female, 47% male). After a warm-up task (which was excluded from the data analysis) each participant completed five randomly presented choice tasks (one from each product group) in which three product alternatives were presented. There was no predefined time limit and participants were asked to observe the presented stimuli while the mouse cursor was hidden. As soon as they decided, which alternative to choose, they clicked the left mouse button once and a choice stating screen appeared, in which they could click on the chosen product alternative using the now visible mouse cursor. Each participant completed one choice task from each product group, but the choice sets were randomized between participants.

---Table 2 about here-----Figure 1 about here---

2.2 Experiment 2 Experiment 2 was conducted to evaluate the effect of the number of presented alternatives on the gazing behavior. Food product images (starting from two alternatives increasing to eight alternatives) were presented to the participants during a decision-making task. The stimuli consisted of pictures of salads (2AFC), breads (3AFC), apples (4AFC), pizzas (5AFC), lettuces (6AF), tomato salads (7AFC) and hamburgers (8AFC) and the presentation order of the stimuli was randomized across participants (Figure 2). The food product images were randomized, and each participant was asked to choose the most appealing one. The alternatives were fixed within product groups; hence the two alternative forced choice test (2AFC) was performed on salads, while 3AFC on breads, etc. Images were placed at the margin of the screen in order to keep the first fixation in the middle using fixation cross. A total of 150 volunteers (63% female, 37% male, aged between 18 and 45) participated in the study. After screening, 35 participants were excluded

from the final data analysis, 11 participants due to insufficient recording quality and further 17 required too long time to state their decisions. This was measured by the time elapsed between the first mouse click and the decision stating. All those participants who needed more than 2000 ms to click on the chosen alternative were excluded from the data analysis. An additional seven participants were excluded because of misunderstanding of the task. The final data set consisted of 115 participants (67% female, 33% male, aged between 18 and 45). Similar to Experiment 1, participants were asked to complete a choice task, but in this experiment the number of alternatives was varied. Each participant evaluated all the sets but in different orders to avoid order-effect.

---Figure 2 about here---

2.3 Data analysis In experiment 1 the full-factorial experimental design was created, and statistical analysis was performed by using R 3.3.2 software (R Development Core Team, 2015). Treatment effects on the factors were analyzed by analysis of variance (ANOVA) testing only the main effects. Each comparison was performed at a significance level of 5%. Duncan’s new multiple range test (MRT) was applied to compare level mean differences regarding only the significant factors (Duncan, 1955). During the procedure of this test the different comparisons between level means might differ by their experiment-wise error rate (α) as it depends on the size of the subgroups. Hence, this post hoc test is less conservative than Tukey's range test (Tukey, 1949). Thus MRT can be considered as the best and top modern post hoc test for multiple comparisons as it permits the assessment of mutual differences that may exist among a set of obtained means (Ram, 1998). Data analysis of experiment 2 was performed differently due to the different data structure. The effect of the number of alternatives on eye-tracking variables was evaluated using nonparametric paired-sample Friedman test coupled with Neményi post-hoc test using XL-Stat software Version 2014.5.03 (Addinsoft, Paris, France). Survival analysis was used to compare decision times computed by StatsSoft Statistica 8.0 (Tulsa, OK, USA).

3. Results 3.1 Experiment 1

Results of the three-way ANOVA showed significant effects for several variables. Table 3 lists the p-values of the ANOVA analyses and the corresponding Duncan post-hoc tests are listed in table 4 for those variables which showed significant effects. It must be mentioned that table 3 lists the differences between product groups and not product alternatives.

Product group Results show that factor product group influences the gazing parameters time to first fixation (TTFF), fixation duration (FD), fixation count (FC), dwell duration (DD), dwell count (DC), and time to first mouse click (TTFMC). Decision time was only influenced by the product group, but not by orientation or size. The Duncan test revealed that the only significant difference occurred for the apple product group, which had the significantly longest decision time with 6.12±2.56 seconds compared to plaited loaf with 4.99±2.49, chocolate (4.59±2.59), pear (4.45±2.07) and gummy bear (4.10±1.51). Participants needed more time to choose one from the apples compared to the other product groups. When the data set was rearranged into two groups (fruits and sweets), fruits achieved significantly higher TTFMC values, which is due to the apple product group. No other significant effects were observed between the two merged groups.

Size Size influenced fixation counts and dwell counts. A significant effect was observed for the size, and the most dwell counts were registered in the case of the large stimuli. The size of the presented stimuli influenced fixation count and dwell count significantly. The bigger stimuli give more room to the viewers to jump more often into or within the stimulus. Results of Duncan post hoc test supported the previous assumptions; the smaller the stimuli, the less fixations are registered.

Background Results show that background had no significant on any of the measured gazing parameters (Table 5).

Orientation The experimental factor of orientation influenced only TTFF and FFD significantly, due to the stimuli placed in the middle of the screen. Results of three-way analysis of variance showed that

orientation had a significant effect on time to first fixation. It was not surprising because four out of the six orientations had a product alternative in the middle of the screen. Left diagonal (0.11±0.37), vertical (0.14±0.29), lower right diagonal (0.11±0.32) and horizontal (0.07±0.55) orientations required significantly less time compared to the upper triangle (0.42±0.22) and the lower triangle (0.70±0.55) orientation. Orientation had also a significant effect on the first fixation duration. The Duncan post hoc test showed that the upper triangle had the shortest values (0.25±0.12) while right diagonal (0.35±0.17) and horizontal (0.33±0.16) orientations had the highest ones. There is a positive connection between time to first fixation (TTFF) and first fixation duration (FFD); the higher the TTFF the higher is the FFD (Table 4). Fixation duration was significantly affected by the product group. Duncan post hoc test revealed that the apples (3.72±2.21) received the longest total fixation duration which was significantly different from chocolate (2.86±1.45), pear (2.32±1.69) and gummy bears (1.91±1.29). In a similar manner longer dwell duration was registered in the case of apple (3.99±2.29) and plaited loaf (3.44±2.04) compared to chocolate (2.96±1.46), pear (2.43±1.46) and gummy bears with only 1.99±1.32 seconds. Product group had also a significant effect on gazing parameter fixation count. Significantly more fixations were registered in the case of apples (12.09±6.72) and plaited loaf (11.56±6.35) compared to chocolate (8.78±3.63), pear (7.91±4.76) and gummy bear (6.22±3.27). Results of dwell count is in accordance with FC: apple with 7.94±3.46 and plaited loaf with 7.74±3.40 counts before chocolate (6.69±2.41), pear (5.61±2.72) and gummy bear (4.78±2.29).

---Table 3 about here-----Table 4 about here--Gazing behavior and choice The relationship between the measured eye-tracking parameters and choice was also analyzed. The data set has been merged in a way that it fulfilled the requirements of running Wilcoxon-Matched pair tests to compare the chosen and not chosen product alternatives. Computations were done for all the six measured variables. The chosen product alternative received higher visual attention in forms of fixation duration (FD), fixation count (FC), dwell duration (DD) and dwell count (DC) which means that participants looked longer and more times on the chosen alternative independently form the presentation (Table 5). In the cases of time to first fixation and first fixation

duration, the results are not consistent, which is in accordance with published results (Danner et al., 2016; Orquin & Mueller Loose, 2013).

---Table 5 about here---

Chi-squared test was applied to compare the empirical frequencies of the chosen alternatives and their expected frequencies. Since three alternatives were presented in every case, we numbered the alternatives from left to right as 1, 2 and 3. The only exception is the vertical arrangement, where the numbering went from top to bottom as 1, 2 and 3 (Table 2). Expected choice frequencies of the alternatives were 33.3%; hence empirical data was compared to 1/3. Chi-squared test gave significant result, which indicates that alternatives were not chosen with equal frequency (χ2(2) = 15.46, p < 0.001). Table 6 indicates that participants tend to choose the upper and/or right alternatives (horizontal, diagonal left, diagonal right and upper triangle orientation), however, in the case of vertical and lower triangle orientation, lowest alternatives were chosen. Pearson residual test indicated that vertical orientation was the only one where the choice frequencies showed significant differences and the upper alternative was chosen significantly less times compared to the bottom and lower ones.

---Table 6 about here---

3.2 Experiment 2 In the second experiment, the effect of different numbers of alternatives on gazing parameters was tested. Shapiro-Wilk normality check proved, that the data sets do not follow normal distribution, hence paired sample Friedman test was applied which is the nonparametric equivalent of repeated measures analysis of variance. The Friedman test reported significant differences between the number of alternatives in the case of FD, DD, FC and DC as shown in Figure 3. No significant differences were found in the case of TTFF and FFD. In order to uncover pairwise differences between the alternatives, Neményi post-hoc test was applied in the case of the significant gazing parameters. Figure 3 shows the median, 1st and 4th quartile and min-max values of the variables. The main result is, that there were no significant differences found between 2AFC, 3AFC 4AFC and 5AFC in the case of duration metrics. 6AFC,

7AFC and 8AFC significantly differed from the first four group in the case of all metrics. 6AFC appeared to be the barrier between the two groups, above six alternatives, the fixation and dwell counts were longer and higher number of fixation and dwell counts were observed. ---Figure 3 about here--Survival analysis of the decision times showed slightly similar results. Survival analysis can be applied to compare time data sets which measure an action, in this case the time elapsed between the presentation of the stimulus and the decision stating. During eye-tracking decision times are usually not normally distributed. In Figure 4, the cumulative survival proportion (decision making) is plotted against the decision times which give the characteristic Kaplan-Meier survival function. These functions can be compared in a way that the highest slope means fastest decision making. Pairwise comparison of these functions was done using Gehan-Wilcoxon test which is a nonparametric test used to compare decision functions. The results are listed in the upper right corner of Figure 4, which shows that decision times between 2AFC, 3AFC and 4AFC are not significantly different, 5AFC differs from 2AFC and 4AFC. 6AFC is located in the middle and significantly differs from all the other situations while there is no difference between 7AFC and 8AFC. These results strengthen the results obtained during the analysis of gazing parameters.

---Figure 4 about here---

4. Discussion Bottom-up experimental factors can significantly affect participants’ eye-movement and food choice as it has been shown by earlier studies of the authors (Jantathai et al., 2013; Vu, Tu, & Duerrschmid, 2016). Hence these factors should thoroughly be considered when planning and designing eye-tracking experiments. In the presented study, the following influencing effects have been identified. Product group had a very strong effect on eye-movement and significantly influences all eyemovement parameters except those which are in related to the first fixations. With regard to the theoretical construction of bottom up factors, the product groups are also connected to top down processes (e.g. by liking, healthiness etc.). While orientation, size and even background color are

typical bottom up factors, the product group seems to be a mixture between a bottom up and top down factor (stimuli driven vs. goal oriented) (Orquin & Mueller Loose, 2013). The obtained results suggest that product group factor that shows a connection to goal oriented orientation (and that is not a pure bottom up factor) is more powerful than all bottom up factors. The influence of product types on gazing behavior is derived from relatively higher-level cognition (what the food products are), but not from lower level perception (the color, size of food products), which might explain that product type did not influence first fixation. Additionally, it could also be related to the complexity of the food product (which could be again a bottom up factor). The significant effect of top down factors on consumers’ perception and acceptance of foods has also been introduced recently (White, Thomas-Danguin, Olofsson, Zucco, & Prescott, 2020). Orientation (vertical, horizontal, etc.) significantly influences only time to first fixation and first fixation duration. Zhang et al. (2019) revealed in a virtual reality experiment that not only orientation, but orientation of the stimuli itself had also significant effect on time to first fixation. The participant detected inward-pointing food product alternatives significantly faster than outward-pointing ones. Additionally, their results showed that this effect is present when an inward-pointing food (pizza slice) is placed on an outward-pointing plate (W. Zhang, Chen, Huang, & Wan, 2019). Van der Laan et al. (2015) found a significant gaze bias for the chosen alternative in two alternative choice test in terms of total fixation duration, but first fixation did not show a significant effect on participants choice (van der Laan, Hooge, De Ridder, Viergever, & Smeets, 2015). Size showed significant effects only on the count type variables only (fixation and dwell count), indicating that participants required more fixations to look at larger stimuli and switched their attention more often between stimuli. Our results are similar those obtained by others in terms of the influencing effect of the size of stimuli. Larger stimuli not only received higher visual attention but increased choice frequency, too, in a recent product label experiment (Peschel, Orquin, & Mueller Loose, 2019). The three most frequently applied background types (black, grey, white) showed no significant effects on any of the measured variables. A previous study by Zhang and Seo (2015) found that background contexts can influence visual attention towards food stimuli, however in their study they did not only changed the background color, but varied the complexity and level of details in the background (e.g. table decoration, dishes and table decoration) (B. Zhang & Seo, 2015). Our

results suggest that the most common background colors used in eye-tracking experiments, black, grey and white, do not significantly influence gazing behavior. Our findings indicate that participants tended to choose the upper and/or right alternatives more often, however, in the case of vertical and lower triangle orientation, alternatives close to the bottom were chosen more frequently. Pearson residual test indicated that vertical orientation was the only one where the upper alternative was chosen significantly less times compared to the bottom and lower ones. In the literature some studies report that participants tend to choose the middle options from an array of similar options (Atalay et al., 2012; Shaw et al., 2000). The higher choice rate of middle options was not confirmed in our study. Similarly to our results, the middle choice preference was not confirmed in a snack choice situation by Missbach and König (2016). Significant connection was found between the number of presented images and eye-movement and decision time. Our results show that an increased number of alternatives increases decision time. Three groups were identified. 2AFC, 3AFC and 4AFC formed the first group, the second group consists of 5AFC and 6AFC (showing the highest variability), and the third is formed by 7AFC and 8AFC, which required the longest time and count values to complete the choice task. These results are in accordance with the earlier results of the authors who analyzed the influence of number of images on eye movements in case of 2-6 images (Vu et al., 2016). They found that the number of images significantly influenced fixation count and visit duration but not fixation duration. Their post hoc analysis showed that two and three images significantly differ from six images but there is no significant difference among four and five images and the other number of alternatives.

5. Conclusions The results of the two studies show that the investigated bottom-up experimental factors can significantly influence gazing behavior; therefore, they need to be carefully considered when planning and/or comparing results of eye-tracking studies. Future research should further investigate bottom-up effects with higher numbers of alternatives, since they represent more realistic circumstances e.g. in a food store or in a web-shop. Even numbers are easier to handle during an eye-tracking study, since they can be presented symmetrically. Odd numbers of alternatives will always force the researchers to present asymmetrical orders. Since the factor of orientation significantly influenced time to first fixation and fixation duration, further research is needed to define, which orientation should be suggested.

Furthermore, these findings indicate that simplified studies might insufficiently represent real-life situations where individuals are confronted with large numbers of alternatives varying in appearance, size, orientation and positioning. Utilizing the rapidly improving eye-tracking technology including mobile eye-trackers and eye-tracking in virtual reality, future research should increasingly focus on more realistic scenarios and environments to improve ecological validity.

Acknowledgement AG thanks the support of Premium Postdoctoral Researcher Program of Hungarian Academy of Sciences and the National Research, Development and Innovation Office of Hungary (OTKA, contract No K119269) as well as the support of NTP-NFTÖ-18-B-0417. SK was supported by the ÚNKP-19-4-DE-5 New National Excellence Program of the Ministry for Innovation and Technology and the János Bolyai Research Scholarship of the Hungarian Academy of Sciences .

This project was supported by the European Union and co-financed by the

European Social Fund (grant agreement no. EFOP-3.6.3-VEKOP-16-2017-00005).

References Atalay, A. S., Bodur, H. O., & Rasolofoarison, D. (2012). Shining in the center: Central gaze cascade effect on product choice. Journal of Consumer Research, 39(4), 848–866. https://doi.org/10.1086/665984 Baldofski, S., Lüthold, P., Sperling, I., & Hilbert, A. (2018). Visual attention to pictorial food stimuli in individuals with night eating syndrome: An eye-tracking study. Behavior Therapy, 49(2), 262–272. https://doi.org/10.1016/j.beth.2017.07.005 Bialkova, S., Grunert, K. G., & van Trijp, H. (2020). From desktop to supermarket shelf: Eye-tracking exploration on consumer attention and choice. Food Quality and Preference, 81, 103839. https://doi.org/10.1016/j.foodqual.2019.103839 Bogomolova, S., Oppewal, H., Cohen, J., & Yao, J. (2018). How the layout of a unit price label affects eye-movements and product choice: An eye-tracking investigation. Journal of Business Research. https://doi.org/10.1016/j.jbusres.2018.10.049 Danner, L., de Antoni, N., Gere, A., Sipos, L., Kovács, S., & Duerrschmid, K. (2016). Make a choice! Visual attention and choice behavior in multialternative food choice situations. Acta Alimentaria, 45(4), 515–524. https://doi.org/10.1556/066.2016.1111 Duerrschmid, K., & Danner, L. (2018). Chapter 12 - Eye tracking in consumer research. In G. Ares & P. Varela (Eds.), Methods in Consumer Research, Volume 2 (pp. 279–318). https://doi.org/10.1016/B978-0-08-101743-2.00012-1 Duncan, D. B. (1955). Multiple range and multiple F tests. Biometrics, 11, 1–42. https://doi.org/10.2307/3001478

Fenko, A., Nicolaas, I., & Galetzka, M. (2018). Does attention to health labels predict a healthy food choice? An eye-tracking study. Food Quality and Preference, 69, 57–65. https://doi.org/10.1016/j.foodqual.2018.05.012 Frazier, T. W., Klingemier, E. W., Beukemann, M., Speer, L., Markowitz, L., Parikh, S., … Strauss, M. S. (2016). Development of an objective autism risk index using remote eye tracking. Journal of the American Academy of Child & Adolescent Psychiatry, 55(4), 301–309. https://doi.org/10.1016/j.jaac.2016.01.011 Frazier, T. W., Klingemier, E. W., Parikh, S., Speer, L., Strauss, M. S., Eng, C., … Youngstrom, E. A. (2018). Development and validation of objective and quantitative eye tracking − Based measures of autism risk and symptom levels. Journal of the American Academy of Child & Adolescent Psychiatry, 57(11), 858–866. https://doi.org/10.1016/j.jaac.2018.06.023 Gere, A., Kókai, Z., & Sipos, L. (2017). Influence of mood on gazing behavior: Preliminary evidences from an eye-tracking study. Food Quality and Preference, 61, 1–5. https://doi.org/10.1016/j.foodqual.2017.05.004 Harezlak, K., & Kasprowski, P. (2018). Application of eye tracking in medicine: A survey, research issues and challenges. Computerized Medical Imaging and Graphics, 65, 176–190. https://doi.org/10.1016/j.compmedimag.2017.04.006 Hummel, G., Ehret, J., Zerweck, I., Winter, S. S., & Stroebele-Benschop, N. (2018). How eating behavior, food stimuli and gender may affect visual attention - An eye tracking study. Eating Behaviors, 31, 60–67. https://doi.org/10.1016/j.eatbeh.2018.08.002 Hummel, G., Zerweck, I., Ehret, J., Salazar Winter, S., & Stroebele-Benschop, N. (2017). The influence of the arrangement of different food images on participants’ attention: An experimental eye-tracking study. Food Quality and Preference, 62(May), 111–119. https://doi.org/10.1016/j.foodqual.2017.07.003 Isaacowitz, D. M., Wadlinger, H. A., Goren, D., & Wilson, H. R. (2006). Is there an age-related positivity effect in visual attention? A comparison of two methodologies. Emotion, 6(3), 511–516. https://doi.org/10.1037/1528-3542.6.3.511 Jantathai, S., Danner, L., Joechl, M., & Dürrschmid, K. (2013). Gazing behavior, choice and color of food: Does gazing behavior predict choice? Food Research International, 54(2), 1621–1626. https://doi.org/10.1016/j.foodres.2013.09.050 Joechl, M., Danner, L., & Dürrschmid, K. (2013). Eye-tracking the obese - BMI has little effect on attention bias. Ernährung/Nutrition, 37, 149–158. Joechl, M., & Duerrschmid, K. (2011). Eye-tracking the gazing behaviour of morbidly obese in relation to caloric density. 9th Pangborn Sensory Science Symposium. Toronto: Elsevier. Kemps, E., Tiggemann, M., & Hollitt, S. (2014). Exposure to television food advertising primes foodrelated cognitions and triggers motivation to eat. Psychology & Health, 29(10), 1192–1205. https://doi.org/10.1080/08870446.2014.918267 Manippa, V., Padulo, C., van der Laan, L. N., & Brancucci, A. (2017). Gender differences in food choice: Effects of superior temporal sulcus stimulation. Frontiers in Human Neuroscience, 11, 1–9. https://doi.org/10.3389/fnhum.2017.00597 Manippa, V., van der Laan, L. N., Brancucci, A., & Smeets, P. A. M. (2019). Health body priming and food choice: An eye tracking study. Food Quality and Preference, 72, 116–125. https://doi.org/10.1016/j.foodqual.2018.10.006 Missbach, B., & König, J. S. (2016). Middle choice preference and snack choice: The role of selfregulatory resources to nudge healthier food choice. Food Quality and Preference, 53, 127–131. https://doi.org/10.1016/j.foodqual.2016.06.007 Mitterer-Daltoé, M. L., Queiroz, M. I., Fiszman, S., & Varela, P. (2014). Are fish products healthy? Eye tracking as a new food technology tool for a better understanding of consumer perception. LWT Food Science and Technology, 55(2), 459–465. https://doi.org/10.1016/j.lwt.2013.10.013 Motoki, K., Saito, T., Nouchi, R., Kawashima, R., & Sugiura, M. (2018). Tastiness but not healthfulness captures automatic visual attention: Preliminary evidence from an eye-tracking study. Food Quality

and Preference, 64, 148–153. https://doi.org/10.1016/j.foodqual.2017.09.014 Motoki, K., Saito, T., Nouchi, R., Kawashima, R., & Sugiura, M. (2019a). Anxiety increases visual attention to hedonic foods: A preliminary eye-tracking study on the impact of the interplay between integral and incidental affect on foods. Appetite, 137, 218–225. https://doi.org/10.1016/j.appet.2019.02.009 Motoki, K., Saito, T., Nouchi, R., Kawashima, R., & Sugiura, M. (2019b). Light colors and comfortable warmth: Crossmodal correspondences between thermal sensations and color lightness influence consumer behavior. Food Quality and Preference, 72, 45–55. https://doi.org/10.1016/j.foodqual.2018.09.004 Næs, T., Brockhoff, P. B., & Tomic, O. (2010). Statistics for sensory and consumer science. In Statistics for Sensory and Consumer Science (1st Ed.). https://doi.org/10.1002/9780470669181 Orquin, J. L., & Holmqvist, K. (2018). Threats to the validity of eye-movement research in psychology. Behavior Research Methods, 50(4), 1645–1656. https://doi.org/10.3758/s13428-017-0998-z Orquin, J. L., & Mueller Loose, S. (2013). Attention and choice: A review on eye movements in decision making. Acta Psychologica, 144(1), 190–206. https://doi.org/10.1016/j.actpsy.2013.06.003 Peschel, A. O., Orquin, J. L., & Mueller Loose, S. (2019). Increasing consumers’ attention capture and food choice through bottom-up effects. Appetite, 132, 1–7. https://doi.org/10.1016/j.appet.2018.09.015 Ram, S. (1998). Modern Teaching Methods. New Delhi, India: Sarup and Sons. Schmidt, R., Lüthold, P., Kittel, R., Tetzlaff, A., & Hilbert, A. (2016). Visual attentional bias for food in adolescents with binge-eating disorder. Journal of Psychiatric Research, 80, 22–29. https://doi.org/10.1016/j.jpsychires.2016.05.016 Shaw, J. I., Bergen, J. E., Brown, C. A., & Gallagher, M. E. (2000). Centrality preferences in choices among similar options. The Journal of General Psychology, 127(2), 157–164. https://doi.org/10.1080/00221300009598575 Spielvogel, I., Matthes, J., Naderer, B., & Karsay, K. (2018). A treat for the eyes. An eye-tracking study on children’s attention to unhealthy and healthy food cues in media content. Appetite, 125, 63–71. https://doi.org/10.1016/j.appet.2018.01.033 Tórtora, G., Machín, L., & Ares, G. (2018). Influence of nutritional warnings and other label features on consumers’ choice: Results from an eye-tracking study. Food Research International, 0–1. https://doi.org/10.1016/j.foodres.2018.10.038 Tukey, J. W. (1949). Comparing Individual Means in the Analysis of Variance. Biometrics, 5(2), 99–114. van der Laan, L. N., Hooge, I. T. C., De Ridder, D. T. D., Viergever, M. A., & Smeets, P. A. M. (2015). Do you like what you see? The role of first fixation and total fixation duration in consumer choice. Food Quality and Preference, 39, 46–55. https://doi.org/10.1016/j.foodqual.2014.06.015 Vu, T. M. H., Tu, V. P., & Duerrschmid, K. (2016). Design factors influence consumers’ gazing behaviour and decision time in an eye-tracking test: A study on food images. Food Quality and Preference, 47, 130–138. https://doi.org/10.1016/j.foodqual.2015.05.008 Werthmann, J., Jansen, A., & Roefs, A. (2015). Worry or craving? A selective review of evidence for food-related attention biases in obese individuals, eating-disorder patients, restrained eaters and healthy samples. The Proceedings of the Nutrition Society, 74(2), 99–114. https://doi.org/10.1017/S0029665114001451 White, T. L., Thomas-Danguin, T., Olofsson, J. K., Zucco, G. M., & Prescott, J. (2020). Thought for food: Cognitive influences on chemosensory perceptions and preferences. Food Quality and Preference, 79(August 2019). https://doi.org/10.1016/j.foodqual.2019.103776 Yasui, Y., Tanaka, J., Kakudo, M., & Tanaka, M. (2019). Relationship between preference and gaze in modified food using eye tracker. Journal of Prosthodontic Research, 63(2), 210–215. https://doi.org/10.1016/j.jpor.2018.11.011 Zhang, B., & Seo, H.-S. (2015). Visual attention toward food-item images can vary as a function of background saliency and culture: An eye-tracking study. Food Quality and Preference, 41, 172–179. https://doi.org/10.1016/j.foodqual.2014.12.004

Zhang, W., Chen, Z., Huang, J., & Wan, X. (2019). The influence of placing orientation on searching for food in a virtual restaurant. Food Quality and Preference, 78, 103728. https://doi.org/10.1016/j.foodqual.2019.103728

Figure captions:

Figure 1. Examples of the presented images in experiment 1. One image was presented to the participants whose task was to choose one product from the image. a) product: chocolate, background: grey, orientation: horizontal, size: small; b) product: gummy bear, background: grey, orientation: diagonal right, size: small; c) product: pear, background: grey, orientation: vertical, size: medium; d) product: chocolate, background: white, orientation: left diagonal, size: medium; e) product: apple, background: black, orientation: horizontal, size: medium; f) product: plaited loaf, background: white, orientation: lower triangle, size: small; g) product: chocolate, background: black, orientation: upper triangle, size: large. Figure 2. Examples of the presented images in experiment 2. One image was presented to the participants whose task was to choose one product from the image. Two alternative forced choice (2AFC) test with salads, 3AFC with breads 4AFC with apples and 5AFC with pizzas, 6AFC with lettuces, 7AFC with tomato salads and 8AFC with hamburgers. Figure 3. Median, 1st and 4th quartile and min-max values of the variables. Numbers denote homogenous subsets defined by Neményi post-hoc tests. Abbreviations: FD: fixation duration, FC: fixation count, DD: dwell duration, DC: dwell count. Figure 4. Survival analysis of the decision times of experiment 2. Significant differences of the applied Gehan-Wilcoxon tests between the cumulative survival curves are presented in the upper right corner of the graph, where * indicates significant differences. Abbreviation: TTFMC: time to first mouse click. Numbers after TTFMC indicate the number of alternatives the participants choose from.

TTFMC_3

TTFMC_4

TTFMC_5

TTFMC_6

TTFMC_7

TTFMC_8

1.0

TTFMC_2

0.71

-0.03

2.11*

7.21*

8.47*

7.97*

TTFMC_3

-

-0.66

1.69

7.22*

8.69*

8.10*

-

2.18*

7.32*

8.72*

8.15*

-

-5.38*

7.78*

6.94*

-

4.28*

2.98*

-

-1.30

Cumulative Proportion Surviving

0.9 0.8 0.7

TTFMC_4 TTFMC_5

0.6

TTFMC_6

0.5

TTFMC_7

0.4 0.3 0.2 0.1 0.0 -0.1 0

5

10

15

20

25

30

Time [ms]

35

40

45

50

55

TTFMC_8 TTFMC_7 TTFMC_6 TTFMC_5 TTFMC_4 TTFMC_3 TTFMC_2

Table 1. Layout of the mixed model nested design applied in Experiment 1. Each participant received five choice tasks and each choice task had different layout. For further details see Table 2. partic ipant

plaited loaf or backg si de round ze r

apple or backg de round r

B

28

W

29

G

W

H

L

S

W

L T

L

G

D L

S

B

D R

M

G

M

G

V

L

G

S

W

S

W

S

G

D R L T H

M

B

S

W

S

B

M

B

D L

L

S

W

H

M

G

V

L

M

W

H

L

B

L T

M



L

H



H

G



W

S



M

U T L T U T



W



S



D R



B

M



54

H



G

B

B



53

L T L T L T

S

M



… W



… 52

D R D L D L

H





27

G



G

L



L

S



V

H



B

W



3

U T D L

M

gummy bear or backg si de round ze r



G

si ze

chocolate or backg si de round ze r D G S R



L



V



G

M



2

V



B



L



V



W



1

siz e

pear or backg de round r U B T

H

M

G

D L

L

B

U T

S

M

B

V

S

B

H

L

M

W

L T

M

B

U T

L

L T L T

Background: W – white, G – gray, B – black Order: V – vertical, DR – diagonal right, DL – diagonal left, H – horizontal, UT – upper triangle, LT – lower triangle Size: S – small, M – medium, L – large

Table 2. The factors and their levels used in experiment 1. Product alternatives are represented by dots in orientation and size in the columns to show the differences between the factor levels. Background Orientation Size Product group Black

Vertical

Small

White

Horizontal

Medium

Grey

Diagonal left

Large

Apple

Chocolate

Gummy bear

Diagonal right

Pear

Upper triangle

Plaited loaf

Lower triangle

Table 3. F-values of the ANOVA Tables (Type III tests) with effects sizes (eta-squared). Factor Df TTFF FFD FD FC DD DC TTFMC Product group (4,265) 0.11 1.45 9.57 13.11 10.55 12.37 6.19 Effect size 0.08 0.36 0.42 0.38 0.41 0.28 Product group* (1,259) 0.09 0.02 2.18 2.87 3.04 0.97 8.25 Effect size 0.00 0.00 0.07 0.08 0.09 0.00 0.17 Order (5,264) 1.67 1.66 1.75 2.03 0.59 23.78 2.26 Effect size 0.65 0.15 0.11 0.11 0.12 0.14 0.00 Size (2,267) 0.74 1.66 1.18 1.50 0.82 5.12 4.39 Effect size 0.00 0.10 0.05 0.25 0.09 0.22 0.00 Background (2,267) 2.22 0.21 1.40 1.02 1.50 0.05 0.30 Effect size 0.13 0.00 0.08 0.02 0.09 0.00 0.00 Bold indicates significant difference on α = 0.05 level. Abbreviations: Df: degrees of freedom, TTFF: time to first fixation, FFD: first fixation duration, FD: mean fixation duration, FC: fixation count, DD: mean dwell duration, DC: dwell count, TTFMC: time to first mouse click. Product group means the analysis of the five product groups (chocolates, gummy bears, plaited loaves, apples and pears). *means product group, when all products were classified as sweet products (chocolates, gummy bears and plaited loaves) and fruits (apples and pears)

Table 4. Means and results of the Duncan post-hoc comparison for gazing parameters significantly influenced by product, order and size. Plaited Gummy Factor: Product Apple Chocolate Lettuce Fruits Sweets loaf bear FD (s) 3.72a 3.22ab 2.85bc 2.31cd 1.91d 3.02a 2.68a FC (n) 12.09a 11.56a 8.78b 7.91bc 6.22c 10.00a 8.85a DD (s) 3.99a 3.44ab 2.96bc 2.43cd 1.99d 3.21a 2.79a DC (n) 7.94a 7.74ab 6.69bc 5.61cd 4.78d 6.78a 6.40a TTFMC (s) 6.12a 4.99b 4.59b 4.45b 4.1b 5.36a 4.52b Right Left Lower Upper Factor: Order Horizontal Vertical diagonal diagonal triangle triangle FFD (s) 0.35a 0.33a 0.31ab 0.29ab 0.28ab 0.25b TTFF (s) 0.70a 0.42b 0.14c 0.11c 0.11c 0.07c Factor: Size

Large

Medium

Small

FC (n) DC (n)

10.44a 7.18a

9.43ab 6.56ab

8.06b 5.92b

Abbreviations: TTFF: time to first fixation, FFD: first fixation duration, FD: mean fixation duration, FC: fixation count, DD: mean dwell duration, DC: dwell count, TTFMC: time to first mouse click. Different

small letters indicate different means (Duncan, α = 0.05). Fruits contain the aggregated results of apple and pear product groups, while sweets contain the aggregated results of chocolate, plaited loaf and gummy bear products. In the case of Fruits and Sweets, small letters indicate the difference between Fruits and Sweets.

Table 5. Relationships between the chosen product and eye-tracking variables. Variable

Chosen Apple

TTFF (s)

Yes No

Effect size

FFD (s)

Yes No

Effect size

FD (s)

Yes No

Effect size

DC (s)

Yes No

Effect size

FC (s)

Yes No

Effect size

DD (s) Effect size

Yes No

Chocolate Plaited loaf

Gummy bear

Pear

1.19±0.77 0.67±0.83 1.17±1.14 0.90±0.99 1.03±0.86 0.73±0.85

1.15±1.22 0.91±0.72

0.94±0.98 1.09±0.87 0.98±1.08 1.19±1.29 1.02±1.13 0.86±0.83

0.26

0.09

0.09

0.32±0.30 0.37±0.18 0.30±0.14 0.27±0.16 0.30±0.15 0.25±0.15

0.33±0.23 0.29±0.23

0.35±0.24 0.33±0.28 0.33±0.19 0.28±0.19 0.27±0.18 0.28±0.17

0.04

0.18

0.22

0.37±0.21 0.36±0.16 0.32±0.13 0.27±0.13 0.29±0.15 0.26±0.11

0.38±0.22 0.31±0.23

0.33±0.20 0.35±0.20 0.35±0.17 0.31±0.24 0.29±0.18 0.28±0.16

0.04

0.34

0.28

3.68±1.66 3.70±1.44 3.55±1.59 2.43±1.59 1.91±1.08 2.67±1.36

2.96±1.46 1.65±0.95

3.22±1.67 3.49±1.67 3.47±1.53 1.97±1.07 2.25±1.42 2.16±1.25

0.52

0.66

0.54

6.30±3.59 4.98±2.22 5.69±3.27 3.29±2.72 2.40±1.55 3.80±2.13

4.31±2.37 1.92±1.26

5.09±3.38 5.81±3.55 5.12±2.76 2.57±1.62 3.00±2.36 2.85±1.92

0.66

0.69

0.60

0.65±0.32 0.5±0.24 0.52±0.21 0.35±0.18 0.38±0.23 0.42±0.3

0.56±0.35 0.37±0.29

0.52±0.23 0.59±0.30 0.52±0.26 0.39±0.28 0.37±0.23 0.39±0.27

0.77

0.62

0.43

0.28

0.30

0.41

0.72

0.71

0.47

0.34

0.23

0.49

0.43

0.54

0.38

Fruits

0.12

0.12

0.35

0.53

0.64

0.65

Sweets

0.05

0.24

0.42

0.60

0.64

0.47

Abbreviations: TTFF: time to first fixation, FFD: first fixation duration, FD: mean fixation duration, FC: fixation count, DD: mean dwell duration, DC: dwell count, TTFMC: time to first mouse click. Bold means significant differences between chosen (Yes) and not chosen (No) alternatives at α=0.05 level. Fruits contain the aggregated results of apple and pear product groups, while sweets contain the aggregated results of chocolate, plaited loaf and gummy bear products. Effect size is measured by eta-squared.

Table 6. Effect of order and position on the chosen samples (χ2(2) = 15.46, p < 0.001). Position Order left middle right Vertical* 4 21 21 Horizontal 9 15 20 Diagonal left 14 11 19 Diagonal right 7 16 22 Upper triangle 16 10 18 Lower triangle 11 20 13 *in the case of vertical order, the positions bottom, middle and top were analyzed

Attila Gere: Conceptualization, Methodology, Data analysis, Writing- Reviewing and Editing; Lukas Danner: Conceptualization, Methodology, Writing- Reviewing and Editing; Klaus Dürrschmid: Conceptualization, Methodology, Writing- Reviewing and Editing; Zoltán Kókai: Writing- Reviewing and Editing; László Sipos: Data Collection; László Huzsvai: Data analysis, Writing; Sándor Kovács: Data analysis, Writing- Reviewing and Editing;

Highlights    

Product types affected the majority of the eye-movement variables Orientation of stimuli influenced time to first fixation and first fixation duration Stimulus size influenced fixation and dwell count Decision time increased over 5 and 6 alternatives