The Determinants of Web Page Viewing The Determinants of Web Page Viewing Behavior: An Eye- -Tracking Study Tracking Study Behavior: An Eye Bing Pan, Helene Hembrooke, Geri Gay, Laura Granka, Matt Feusner, Jill Newman HCI Group Information Science Program Department of Communication Cornell University http://www.hci.cornell.edu
Introduction Introduction – Web: a ubiquitous information source – Web pages involve textual, pictorial, and multimedia information – Different cognitive processes:ocular-motor to semantic – Very limited research
Background Background Determinants of viewing behavior — Three processes during an eye fixation: encoding, sampling, and planning [Viviani 1990]: — Eyes are attracted to most informative areas [Rayner, 1998] — Fixation frequency -> importance; Fixation duration -> complexity and difficulty [Fitts et al. 1950] — Instructions influences fixation sequence [Buswell 1935] — Intent influence eye movement [Yarbus 1967] — Complexity influences fixation duration [Pelz et al. 2000]
Web Page Viewing Behavior: Previous Research Web Page Viewing Behavior: Previous Research – Texts were first entry points for online news readers [Stanford Poynter Project 1998] – Viewers spent more time on pictures than texts in print advertisements [Rayner 1998] – Subjects have habitually preferred scanpath and features of site and familiarity influence scanpaths [Josephson and Holmes 2002] – Research on a web portal page: more horizontal movement; headers are not always viewed before the body; searches are not more directed along time [Goldberg et al. 2002]
Methodology Methodology – 30 Subjects (1 African American, 4 Asian Americans, and 25 Caucasians, 13 females and 17 maless) – 22 web pages from 11 popular web sites from four categories (news, shopping, search, and business); – Half of the subjects were given instructions to remember as much as they could about the content; – 30 seconds of viewings; – Demographic data and recall were measured.
Stimuli Stimuli
Indices Indices Dependent variables: – Mean Fixation Duration (average fixation duration): indication of information complexity and task difficulty [Rayner 1998] – Gazing Time (rate of gazing vs. fixation): negatively related to task difficulty [Nakayama et al. 2002] – Saccade Rate (number of saccades per second): negatively related to task difficulty and mental load [Nakayama et al. 2002]
Eye Tracking Equipment: ASL 504 Eye Tracking Equipment: ASL 504
Analysis Analysis A mixed model [Little et al. 1996] – 2 (Gender) X 2 (Task Condition) X 2 (Page Order) X 4 (Site Type) – Fixed factors: gender, task condition, page order, site type – Random factors: subject factor, random web site selection from each of four categories of sites
Results Results Table 2. Mixed Model on Mean Fixation Duration Table 3. Different Values of Mean Fixation Duration (in Seconds) Numerator Denominator Signific Effect F Value Group DF DF Mean Fixation Duration ance Gender 1 98.8 5.22 .02 Male .374 Page 1 104 16.66 .00 Female .357 Order Types First Page .377 of Sites 3 104 5.95 .00 Second Page .353 * Page Order
Results Results Table 4. Mixed Model on Gazing Time Numerator Denominator F Table 5. Different Means of Gazing Time (Percentage) Source Significance DF DF Value Group Mean Gazing Time Page 1 104 4.3 .04 Order First Page 66.7 Type of Second Page 64.3 Sites * 3 104 10.04 .00 Page Order
Results Results Table 6. Fixed Model on Saccade Rate Numerator Denominator F Source Significance DF DF Value Type of Sites * 3 104 5.530 .00 Page Order
Results Results Interaction between Types of Web Sites and Page Order 0.41 Shopping 0.4 Search 0.39 New s Business 0.38 0.37 0.36 0.35 0.34 0.33 First Page Second Page Figure 2. Interaction Effect on Mean Fixation Duration
Interaction Effects Interaction Effects 72 Shopping Search 70 News Business 68 66 64 62 60 58 First Page Second Page Figure 3. Interaction Effect on Gazing Time
Interaction Effects Interaction Effects 1.95 1.9 1.85 Shopping 1.8 Search News 1.75 Business 1.7 1.65 First Page Second Page Figure 4. Interaction Effect on Saccade Rate
Overall Results Overall Results Page Order Mean Fixation Duration Gazing Time Site Type Saccade Rate Gender Figure 5. Overall Results
Scanpath Analysis Scanpath Analysis 1.4 1.2 1 0.8 0.6 0.4 0.2 0 1 F F F F F 1 F 1 1 F 1 1 1 1 F F F F 1 1 1 - - - - - - - - - - - - - - - - - - - - - E R T E T N R N S S - Y O N N N O O O N Y 3 3 E E L T W W L N S N T S T T A R O R O O O A G N G N E E B - C C M C C - M B B Z B H H Z O C O C N N E O A A A O A E A A O O M M M Y Y M J J G G A A Figure 6. Variance of Scanpaths for Different Web Pages
Variance of Web Sites Variance of Web Sites Figure 8. Two Web Pages with the Largest Scanpath Variances Figure 7. Two Web Pages with the Smallest Scanpath Variances
Conclusions Conclusions � Gender of subjects, the order of web pages being viewed, and the interaction between site types and the order of the pages influence web viewing behavior; � A possible relationship between scanpath variability and the structural/visual complexity of the web page; � Individual characteristics and the stimuli contribute to viewers’ eye movement behavior [Rayner 1998]. � The problem with mental workload
Conclusions Conclusions � Males exhibited significantly longer mean fixation durations than females while females scan more of the pages – males and females reacted differently to images, and affected subsequent memory performance [Jones, Stanaland, and Gelb, 1998]:. – females often engage in comprehensive processing of all the available information, while males tend to focus on a fewer number of areas [Meyers-Levy and Maheswaran, 1991]:. � Complex interaction effects of page order and site type – Eye movement behavior changes over time, more variability on first pages – The change in direction and magnitude were influenced by the type of web sites – “Context Cueing” and adaptation [Chun, 2000]
Conclusions Conclusions – No significant effects for other individual variables and task instruction – Mental workload is a multi-dimensional construct. – know “what” and “how” not “why” – Future research: combined with verbal protocol, videos of facial expression, and post experiment interviews.
Thank you! Comments and questions…
Ten Things You Don’t about Women Ten Things You Don’t about Women … 8. We can scan a room faster than you can. Within sixty seconds we can spot every girl to watch out for and all the men we’d like to meet. … – Electra, C. (2004). Ten things you don’t know about women. Esquire, 141(2): 68.
Recommend
More recommend
