Chapter 9 Deep View collaboration user study
10.3 Future work
We have identified two main directions for future work. One relates specifically to video conferencing and the other to improving interfaces for DH pair and BS pair collaboration.
10.3.1
Designing future video conference systems
Although we did not develop a novel video conferencing application, knowledge from the user study can be used to design future video conferencing systems. There are two possibilities.
First, knowledge about the DH pairs' use of gestures is useful to users regardless of disability. The user study supports the concept that video is important for making decisions. Previous studies on video conferencing involving collaborators without disabilities find limited use of video. In limited instances, however, researchers observed that collaborators use video to interpret the other person's intentions when making a difficult decision (Hudson, Helser et al. 2003).
Second, the experimenter's observations of the DH pairs indicate that the gestures are limited to a small set of acknowledgements. In the future, one could abstract the gestures and eliminate the video conferencing, which has high bandwidth requirements. The gestures could be represented as animated icons. A user could trigger the animated icon through a keystroke. With this abstracted interface, collaborators might be able to communicate as well as with video conferencing. If video conferencing proves more useful than the animated icons, however, it would suggest that other valuable information is conveyed through the video.
10.3.2
Improving collaboration between a DH pair and a BS pair
Results from our user studies indicate areas of future research to enhance user interfaces to support collaboration between DH pairs and BS pairs. In our user studies the BS and DH pairs take longer to complete the brainstorm diagram task than collaborators without disabilities. The distribution of times for the groups to complete the brainstorm diagram task are shown in Figure 10-1. Times for the groups without disabilities and deaf-hearing group were recorded in the user study investigating deaf-hearing pairs (Chapter 5). Times for the blind-sighted group were recorded in the collaborative Deep View user study (Chapter 9). The time to completion for the task, however, cannot be statistically compared for various reasons including:
• The data samples are too few to get statistically valid results.
• The tasks were completed under different conditions, for example collaborators without disabilities used a tightly coupled interface while BS pairs used a loosely coupled interface.
• In the case of DH pairs, only one pair completed the task successfully as
expected. Two pairs did not follow instructions and one pair did not complete the task successfully.
Figure 10-1. Time to complete the brainstorm diagram task (N= 9 diagrams for group without disabilities, N=12 diagrams for deaf-hearing group, N= 9 diagrams for blind-sighted group)
Although the sighted collaborators' faster time to completion is not statistically significant, it reflects a typical artifact of assistive technology: In some cases a person with a disability using assistive technology completes the task more slowly than a person without a disability. For example, blind users' access to diagrams is slower than sighted persons' because diagrams are optimized to be perceived visually. Note, however, advanced screen reader users in some cases can complete a task as fast as or faster than sighted persons who have less experience with the task. Also, taking longer to complete a task does not necessarily reflect the quality of the final product or the complexity of material covered. However, completing a task faster can increase the amount of material covered in a collaborative session.
We have two suggestions for future research for enhancing interfaces to support DH pairs and BS pairs. The first suggestion is to identify why the DH pairs and BS pairs are slower than collaborators without disabilities. Then the interfaces can be improved to
make the DH pairs and BS pairs complete the task faster. One could analyze
collaborators' sessions and categorize the situations in which collaborators spend their time. Our hypothesis is that pairs of collaborators without disabilities complete the task faster because they divide the task between themselves while DH pairs and BS pairs do less so. DH pairs and BS pairs might be able to divide the task; however, they did not do so in the user study maybe because they were becoming accustomed to the new
experience of collaborating. It seemed collaborators wanted to work together to confirm the task is completed as expected.
The second suggestion is to tailor tasks and user interfaces to direct the
collaborators' interactions. For example, DH pairs playing checkers did not have the in- depth deliberation about checkers moves that collaborators without disabilities had. The DH pairs might have benefited if they could have used chat messaging to communicate ideas. However, the difficulty having a discussion about checkers using chat messaging is that it is difficult to refer to checkers pieces, which are best referred to through pointing. In the case of BS pairs, the blind person would benefit from declaring a pause, where the blind person can examine the diagram without interruption from the sighted person.
We hope our research contributions will help researchers in the future be more successful in improving assistive technology for people with disabilities.