The Effect of Direct Annotation on Speed and Satisfaction

Discussion

The results supported our hypothesis that direct annotation would significantly receive the highest subjective preference. The low subjective preference of the text box annotation method could be due to the fact that there was no association between names and people in the photos. Meanwhile, in direct annotation, subjects could drag a name next to a person in the photo. Similarly, click/type allowed subjects to click on the person in the photo and type in their name. For the click and type method, most likely reason that it came in second in subjective preference was that it had association between names and people but required typing names. Most likely reason that direct annotaion had highest subjective preference was that direct annotation is categorized under the concept of direct manipulation. This likely gave the subjects a feeling of high degree of control, which is a very important factor for a computer user.

The results did not support our hypothesis that direct annotation would significantly produce faster performance time over the other two treatments. Text box annotation resulted in the fastest performance time. A one possible explanation is that subjects only needed to type the names listed in the experiment packet; the packet gave the names of the people in each photo and the task of annotation came down to typing in the names listed in the packet onto the text box. This sometime lead to a subject typing in names without looking at a photo. As for click and type's having the slowest performance time, the alternating use of the keyboard and mouse was the likely reason. Due to the swithing between a keyboard and a mouse, a subject will need a extra time to do their task. Finally for direct annotation, the reason that it was faster than click/type was that only a mouse was necessary to use. The explanation that the direct annotation method is slightly slower than text box could be more amount of time is required for searching for the names than typing straight from a given list. As the list of names grows very large, the time may start to become worse for direct annotation. However, it was also a problem that our subjects did not seem to use the interface to the full extent; they could have typed in the first letter of a name to scroll down the list more quickly. We don't know whether this was caused by a lack of emphasis on using the feature or simply subjects didn't followed their instruction.

In terms of anomalies, the way the experiment is set up gives text box an advantage in completion time than the other two annotation methods. As mentioned before, the packet provided to each subject listed the names of the people in the photo. Therefore, the task of annotation for text box becomes a task of typing names from a list, which could be done without looking at the photo on the screen. Also, another anomaly is that for text box and click/type, subjects who knew how to type had an advantage over subjects who had to look at the keyboard to type. Finally, for direct annotation and click/type, we observed that some subjects knew that clicking anywhere on the photo or dragging names anywhere close to the person in the photo suffice while other subjects made sure to click onto the person in the photo precisely. This gave the subjects who dragged or clicked anywhere close on the photo a performance time advantage.

Department of Computer Science: Direct questions and comments to the student editorial team