Different user groups exhibit different searching behaviors. College students are one of the most studied groups, because they frequently use Web search engines and they are normally the convenience sample for researchers. For example, research- ers (Hawk & Wang, 1999; Wang, Hawk, & Tenopir, 2000) examined 24 graduate students’ cognitive, affective, and physical behaviors during user-Web interactions.
Based on analysis of participants’ verbalizations during searches, they identified 10
problem-solving strategies: surveying, double-checking, exploring, link following, back and forward going, shortcut seeking, engine using, loyal engine using, engine seeking, and metasearching ranging. Furthermore, they associated the problem strategies with the physical, situational, cognitive, and affective factors related to
user-Web interactions. Cognitive factors influence users in question analysis and
in the selection of search and problem-solving strategies. Cognitive styles affect
the search process; to be more specific, field-dependent users have more difficulty
in searching the Web. Simultaneously, affective and physical factors can support
or undermine an interaction. The identification of problems and problem-solving
strategies helps researchers understand user-Web interactions. However, in order to design IR systems to facilitate these interactions, further research needs to connect types of problems/situations with types of corresponding strategies.
Fidel et al. (1999) investigated eight high school students’ (grades 11 and 12) Web-searching behavior for homework assignments based on data collected from observation and interviews. They characterized students’ searching behavior as
performing focused searching in general, conducting swift and flexible searching
in the process, using landmarks, making the assumption of always starting a new search, and asking for help. They called for the need to train students and design Web search engines considering users’ behavior, for example, providing easy ac-
cess to knowledge tools, navigational tools, correction tools, filtering tools, and
visual tools. Large, Beheshti, and Rahman (2000) held four focus groups of users ranging from 10 to 13 years old to explore design criteria for Web portals. Although the purpose of the study was not to identify information-seeking strategies, they
did observe the steps taken by each group to find answers to four questions. They
found that young users’ information-seeking strategies were affected by the design of the search engines, especially their interfaces, and the search statement. Word and phrase searches were applied the most in their searching for the four questions in four search engines, mainly because it is relatively easy to extract keywords from two of the questions. Users preferred general searching to directory, and they explored directory in one search engine because the search box was not placed in an obvious location.
Cothey (2002) conducted a longitudinal study of the information-searching be- havior of high school students based on log analysis. Contradictory to the general notion that users are more systematic in their information-seeking behavior, she found that high school students adopted a more passive or browsing approach to
Web searching after they gain experience in Web searching. To be more specific,
students accessed the Web less and used link-click as opposed to active searching as they became more experienced. In addition, they also became more diverse in selecting Web hosts when their experience increased.
Bilal (2000, 2001) reported on children’s (seventh grade students) cognitive, physi- cal, and affective behavior in using the Yahooligans! Search engine on fact-based tasks and research tasks. For fact-based tasks, more children adopted the keyword searching approach (64%) than the browsing approach (36%). The children who took the keyword approach were nonconforming and certain about the keywords, while the children who applied the browsing approach were systematic and orderly. For research tasks, children browsed more than they searched by keyword. Only one child used natural language queries. The results indicated that the children browsed and searched by keyword more in fact-based tasks than in research tasks. Simultane- ously, they made more moves and took more time to accomplish fact-based tasks
than research tasks. The findings of this study uncovered the problems of the design
of Yahooligans! Bilal further suggested that a search engine for children should support children’s learning requirements and cognitive demands.
After analyzing computer trace data of 32 elementary school students’ (fourth and
fifth grade) search process, Schacter, Chung, and Dorr (1998) found that children
were interactive searchers, and they did not plan for their search tasks. They pre- ferred browsing strategies, and they did not systematically plan or use sophisticated
analytical search techniques. This finding echoes Bilal’s results. As was reported
in Bilal’s (2000, 2001) study, children exhibited more analytical behavior for well-
defined tasks than for ill-defined tasks. Boys browsed more than girls because they
either browsed documents faster or they did not read as much as the girls did. Chil- dren also liked to use full sentence requests as their queries. This study indicated
that ill-defined tasks are better suited for children because they offer more potential
answers and require less analytical techniques.
The above-cited studies revealed different searching patterns for college students, high school students, and children. College students can master more complicated search strategies or problem-solving strategies than younger group users, and their strategies are affected by cognitive factors, especially cognitive styles. Compared with high school students, children’s searching processes were less focused; they switched back and forth between keyword searching, browsing and visiting sites, and they frequently looped searches. The limited recall knowledge of children often led
to frequent looping. Children’s learning and cognitive abilities might also influence
Interactve IR n Web Search Engne Envronments
seeking behavior on the Web: difficult to select search terms in searching, less time spent in viewing information, difficult to make relevance judgments, and difficult
to express their information needs in the form of query formulations and strategies shifts. However, existing research needs to be enhanced to have a more representa- tive sample of users to uncover the information-seeking strategies applied during user-Web interactions.