HAL Id: hal-01986519
https://hal.archives-ouvertes.fr/hal-01986519
Submitted on 18 Jan 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
What scroll can teach us about web users ?
Alexandre Milisavljevic, Karine Doré-Mazars, Bernard Gosselin, Matei
Mancas, Coralie Petermann
To cite this version:
Alexandre Milisavljevic, Karine Doré-Mazars, Bernard Gosselin, Matei Mancas, Coralie Petermann.
What scroll can teach us about web users ?. European Conference on Visual Perception, Aug 2017,
Berlin, Germany. 2017. �hal-01986519�
Users adapted their eyes position before scrolling.
Results
What scroll can teach us about web users ?
1 Psychology institute, Vision Action Cognition EA 7326 lab - Paris Descartes University, Boulogne-Billancourt, France
2 Numediart institute, University of Mons, Mons, Belgium 3 Research and Development department, Sublime Skinz, Paris, France
Introduction
Methods
▶ Participants and apparatus
• 5 participants.
• Normal or corrected-to-normal vision. • 4♂; 1♀; 24±2 y.o.
• Monocular recording with a FaceLab 5 eye-tracker. • Google Chrome maximized at 1920x955.
Conclusion
40
thEuropean Conference on Visual Perception, Berlin, Germany, 2017, August 27
th– August 31
st▶ Procedure
• Click on the bookmark situated in the browser’s top bar request to the server the next link to load (cf Figure 1). • Then website or instruction is displayed.
• When the user finishes reading or doing the task, he/she clicks on the bookmark again (cf Figure 1). • Etc.
Tasks Description Free
viewing Browse the website by visiting at least two other pagesBrowse two articles of your choice Target
finding
Browse the following pages: calendar, team and news Buy the specific given item
Text
reading Read the two first paragraphs
Alexandre Milisavljevic
1,2,3, Karine Doré-Mazars
1, Bernard Gosselin
2, Matei Mancas
2, Coralie Petermann
3▶ Tasks
• At most 2 webpage visits. • No time limit.
• Full scroll possibilities.
▶ Amplitude
▶ Speed
• For our analyses we divided the screen in 2 or 6 areas
numbered as follows to have two-levels accuracy:
▶ Definitions
Mouse movement: Physical mouse shift resulting in a
change of cursor position on the screen.
Scroll: use of the mouse’s wheel or equivalent to scroll
up or down
Scroll session: set of continuous scrolling events on
the same webpage ended with a mouse movement
Scroll
speed positionEyes
Scroll down slow
fast
Scroll
speed positionEyes
Scroll up slow fast
Acknowledgment
We thank French Research and Technology Association and Sublime Skinz for supporting this work and Kevin Hamard for his help on this work. • We consider Free Viewing and Target Finding tasks because Reading task was too
localized and did not require scroll.
Mean speed with anticipation Mean speed without anticipation Target Finding 1154px/s 1038px/s Free Viewing 996px/s 783px/s
▶ Anticipation
However we found that:
•scroll is faster when eyes anticipates the scroll.
Analyses showed no significant effect of scroll anticipation by the eyes for both types of task in six-areas and two-areas configurations.
• when the user begin to scroll, there is a higher probability that his/her eye position was on the same half of the screen than the direction of his/her scroll.
We observed that participants adapted their eyes position according to their scrolling speed (cf Figures 5, 6 and 7).
Users adapt their eyes position according to their intent:
•Before scroll, eyes position are a clue to guess how fast and how far the user will scroll
•While scrolling, users position their eyes according to where they think
the information could be. While scrolling fast, they position their eyes at the opposite direction to be able to detect bottom-up characteristics through peripheral vision. Furthermore, when users are looking for a specific information or one that need more attention, they scroll more slowly and position their eyes in the center of the screen or in the same direction as the scroll.
• When the eyes were located on the top screen area before scrolling down the scroll amplitude was much higher.
References
Guo, Q., & Agichtein, E. (2010). International Conference Extended Abstracts on Human Factors in Computing Systems Huang, J., & White, R. (2012). Conference on Human Factors in Computing Systems
Huang, J., White, R. W., Buscher, G., & Wang, K. (2012). Conference on Research and Development in Information Retrieval Navalpakkam, V. et al. (2013). Conference on World Wide Web
Rodden, K., & Fu, X. (2007). Conference on Research and Development in Information Retrieval Sharmin, S. et al. (2013). Conference on Eye Tracking South Africa
Yarbus AL. (1967). Plenum Press
While looking for specific information, users scroll slower to be able to differentiate elements (text, blocs, titles, etc).
While looking for a more generic information
like an image, a colored box surrounding a website category, etc, users positioned their eyes on the opposite side of the scroll direction. They used their peripheral vision to detect bottom-up elements.
• When the eyes were located on the bottom screen area before scrolling up the scroll amplitude increased.
1 2 3 4 5 6 1 2
Figure 2: Screen areas
anti-correlation of -0.77 correlation of 0.94
Browse two articles of your choice Bookmark to go to next page
task or website Website instruction Mouse-Tracking plug-in ~50ms sampling Time
Figure 1: Study’s browser setup and procedure Buy the item n°X
Figure 3: Scroll up amplitude according to pre-scroll eyes position Figure 4: Scroll down amplitude according to pre-scroll eyes position
Figure 5: Scroll down speed according to eyes position while scrolling
Figure 6: Illustration of eyes position according to scroll speed while scrolling down
Figure 7: Illustration of eyes position according to scroll speed while scrolling up Table 1: Tasks list by category
Table 2: Speed in pixels per seconds with and without anticipation and its effects on scroll speed
▶ Set-up
• Two types of webpage to display : instruction or website bound to a task. (cf Figure 1) • Instructions and websites links were stored on a local server.
• All instructions were stored locally and websites were online.
• time
• event type (click, movement or scroll) • mouse’s coordinates • offset induced by the
scroll • URL
• screen size (width x height)
• eyes position
Collected data
Understanding why a user is on a webpage is a good way to deduce his or her interest in the content. To measure this interest, Eye-tracking is a precise tool that allows to estimate goal impact on user’s eye-gaze (Yarbus, 1967). However, this method is hard to scale up.
▶ State of the art
Thus, mouse-tracking models emerged as an efficient proxy to determine user’s attention (Rodden, 2007; Navalpakkam, 2013; Guo, 2010; Huang & White, 2012; Huang et al. 2012). These same models mainly use mouse movements, mouse clicks and hovered page elements while considering scrolling as a simple model feature. In addition to these analyses, other studies focused on how the eye behave during onscreen reading using scroll (Sharmin, 2013).
