Digital Library Interfaces for Children: The Effects of

Visual Navigation on Usability

Glenda Revelle, Benjamin B. Bederson, Allison Druin, Dana Campbell, Allison Farber, Juan Pablo Hourcade, Juhyun

Lee, Yoshifumi Takayama

Teachers and students of Yorktown Elementary School

Bowie, Maryland

University of Maryland, Human-Computer Interaction Laboratory

SearchKids: A Digital Library Interface for

Children

Completing third and final year of NSF-funded Digital Libraries project

Visual interface to digital libraries for young children (early elementary)

Enables children to search a multimedia database of information about animals

University of Maryland, Human-Computer Interaction Laboratory

This Year’s Focus

Continued our collaborative design process, including children, teachers, and researchers

Currently evaluating the effectiveness of SearchKids visual interface

Compares 2nd and 3rd graders use of SearchKids to more traditional, less visual alternatives

University of Maryland, Human-Computer Interaction Laboratory

[Demo]

“Original SearchKids” “Traditional”

University of Maryland, Human-Computer Interaction Laboratory

Research Design

Goal: understand effectiveness of SearchKids interface features

140 students from Yorktown Elem. School (half-way done)

Independent Variable: Interface - traditional AND one of:

Full SearchKids SearchKids with no images SearchKids with no animation SearchKids with neither images nor animation

Grade Gender

Dependent Variables: Speed Accuracy Subjective Satisfaction

University of Maryland, Human-Computer Interaction Laboratory

Research Design (cont.)

Children asked to find as many items as possible in 15 minutes (items spoken by test administrator)

Search tasks include specific animals (e.g., dog) and categories of animals (e.g., something that lives in water)

Half used traditional interface first; half used SearchKids version first

University of Maryland, Human-Computer Interaction Laboratory

Preliminary Results

Preliminary analysis on Original SearchKids vs. Traditional (with ten 2nd graders and eight 3rd graders)

Number of kids tested too small to know about statistical significance yet, but some patterns are emerging

University of Maryland, Human-Computer Interaction Laboratory

Preliminary Results:Search Outcomes

Looking at three measures: Total # items completed during session # “right” items found # “wrong” items found

Grade effect for total # items completed: 3rd graders found 12 items 2nd graders found 9 items

University of Maryland, Human-Computer Interaction Laboratory

Preliminary Results:Search Outcomes (cont.)

For # “right” items found: 2nd graders submit more right answers with SearchKids 3rd graders submit more right answers with traditional But, this effect occurs only on category search tasks

For # “wrong” items found 2nd graders submit more wrong answers with

traditional 3rd graders submit more wrong answers with

SearchKids Again, this effect occurs on category search tasks

University of Maryland, Human-Computer Interaction Laboratory

Preliminary Results: User Satisfaction

Original SearchKids rated as being more useful than the traditional, text-based approach

Boys rate SearchKids as being less “fun to use” if it has no images, whereas girls rate it as less fun without animations

University of Maryland, Human-Computer Interaction Laboratory

(Preliminary) Conclusions

SearchKids enables young children to perform complex queries more effectively than traditional text searches

In other situations, traditional text search proves more effective

University of Maryland, Human-Computer Interaction Laboratory

(Preliminary) Conclusions (cont.)

Young children perceive a visual search interface to be more useful and more fun to use than a traditional text interface

Therefore, your choice of interface should depend on the age of the children, and their tasks.

www.cs.umd.edu/hcil/searchkids