Mean values of performance time and oculomotor measures for the text search task are shown in Table 2.
Statistical analysis did not reveal any significant difference in performance time. At the same time, oculomotor indicators of increased cognitive load – increase in fixation duration, decrease in saccadic amplitude and saccadic peak velocity – showed statistically significant differences in the traditional and flat series. This type of combination of measures (long fixations and short saccades) is characteristic of focal visual information processing: i.e. a conscious analysis of information, precise identification of objects and events, which are implemented when the visual search tasks increase in complexity . In the text search with traditional fonts, subjects had more opportunities to switch to “semi-automatic” information processing associated with a lower cognitive load (which is indicated by shorter fixations and longer saccades). Also the lower values of saccadic peak velocity provide evidence in favour of associating the text search in the flat series with a higher cognitive load.
Mean values of performance time and oculomotor measures for the icon search task are shown in Table 3.
A significant difference was found in the mean values of the icon search time: almost twice as high for flat as for realistic icons. Unlike in the previous task, a comparison of oculomotor activity in the graphic objects search did not reveal any significant difference in mean fixation duration. Nevertheless, a difference in mean saccadic amplitude and saccadic peak velocity remained, just as in the text search task. Values were less in the flat series, which may indicate the higher complexity of the task and a higher cognitive load in the flat icon search.
Monitoring the performance process of this task allowed us to assume that many participants in the flat series could not find the target icon during the initial “fast” slide scanning. Later in the search these participants tended to show more care in scanning the images, enabling them to find the target object. This, however, led to a significant increase in search time.
Mean values of performance time and oculomotor measures, as well as rates of ‘miss’ and ‘false alarm’ errors are shown in Table 4.
As expected, total task performance time on traditional sites was higher, as information density on the screen was considerably higher than on flat screenshots. On traditional sites there were 110 clickable and 64 unclickable screen areas (total: 174), while on flat sites there were 78 clickable and 54 unclickable screen areas (total: 132). For this reason, the mean performance time for a single screen area was calculated (for both traditional and flat sites). The results demonstrated that the average processing time for a screen area (including making a decision on the objects’ clickability and clicking the clickable objects) was significantly higher for flat websites.
An analysis of ‘miss’ and ‘false alarm’ error types revealed a significant difference between traditional and flat sites: errors of both types were significantly more frequent on flat sites. It is noteworthy that the percentage of false alarms on flat sites in our experiment (28 %) almost exactly corresponds with the figure for false alarms (29 %) in the research conducted by Idler .
It should be noted that in the web search task oculomotor effects were revealed, which were the reverse of those found in the text and icon searches: a search for clickable objects on the page with flat design was characterized by a higher saccadic amplitude and saccadic peak velocity. However, we are not inclined to interpret these results as evidence in favour of a higher cognitive load when working with traditional sites. In our opinion, a key role here is played by the difference in the characteristics of the stimulus material. These effects may be associated with fundamental differences in the design of traditional and flat sites, which force subjects to use different scanning strategies. Thus, in our experiment on sites with traditional design the number and density of graphic objects was higher and interface control tools were more distinct. After initial orientation this allowed the user to develop a systematic search strategy – seen in the combination of longer fixations and shorter saccadic duration. By contrast, the flat design sites initially contained less graphic and text information, which normally facilitate the search for interface control tools. This made subjects repeatedly perform search activity and return to viewing certain areas of web pages several times, shown by a decrease in fixation duration and increase in the amplitude and velocity characteristics of saccades. Thus, the search on flat sites was more “chaotic”, which had a negative impact on time and accuracy parameters of task performance.