An evaluation of methods of indicating active words in hypertext documents

• Main Author: Watkins, Rani Lea
• Other Authors: Industrial and Systems Engineering
• Format: Others
• Language: en
• Published: Virginia Tech 2014
• Subjects: LD5655.V855 1991.W375; Hypertext systems > Research
• Online Access: http://hdl.handle.net/10919/46458; http://scholar.lib.vt.edu/theses/available/etd-12302008-063654/

Three methods of indicating active (linked) words in hypertext documents were investigated. The Methods consisted of font-manipulation techniques (such as shadowed and outlined text), word-enclosure techniques (variations of a box and pseudo-brackets around the word), and "punctuation-like" icons (located directly above the first letter of an active word). Two Cues were nested within each Method, yielding a total of six Cues. Twenty-four subjects (12 males and 12 females) performed reading tasks and visual search tasks to evaluate each of the three Methods and six Cues. A hierarchical, within-subjects experimental design was used, employing a completely counterbalanced order of treatments. The study consisted of two experiments. For Experiment 1 (reading tasks), participants read Tinker (1955) passages and identified an inappropriate word in each passage. Passages contained either one of the six Cues or no Cues (control condition). The times taken to read each passage and locate the target word, as well as the number of errors made, were recorded. For Experiment 2 (visual search tasks), participants scanned text fields and located and counted the number of active words contained within each text field; again, the times taken to locate all active words and the number of errors made were recorded. For both experiments, participants rated Cueing Methods along various dimensions and selected a preferred Cue and Method for the tasks performed in each of the two experiments. For reading tasks, no single Cueing Method yielded significantly shorter reading times than any other Method. Similarly, no particular Method was preferred by significantly more participants than any other Method. However, user ratings were more favorable for Icons than for Enclosures or Character Styles. In addition, Character Styles were consistently rated as being highly distracting for reading tasks. For visual search tasks, Character Styles yielded shorter search times (<i>p</i> < 0.05) as well as significantly more favorable salience ratings. Participants also preferred the Character Styles significantly more often than either Enclosures or Icons for locating linked words. The number of errors produced for both experiments was very low (less than 2%) and there were no significant differences in errors across Methods or Cues. This finding is not unexpected considering the low level of difficulty for the reading and search tasks. Based on the results of both experiments, Icons are the recommended Method for indicating linked words in hypertext documents. Icons provide moderate perceived readability (reflected in subjective ratings, though not in reading times or preferences), whereas Enclosures were not sufficiently salient and Character Styles were perceived to have degraded text readability. A discussion of a wide variety of techniques which either are currently used in existing hypertext systems or could potentially be used is included in the Literature Review section. The issue of trade-offs between Cue salience and obtrusiveness is addressed in the Discussion. === Master of Science