The Effect of Direct Annotation on Speed and Satisfaction

Conclusions

Impact for practitioners

While there was no statistical significance for the completion time among the three annotation methods, there was a statistical significance in subjective preference; direct annotation was shown to be the most popular choice among the three designs. As mentioned in Shneiderman's book, the concept of direct manipulation was well liked by our subjects. Designers of photo libraries , as well as designers of other software that can apply the concept of direct manipulation, are strongly encouraged to use direct manipulation. The satisfaction of customers who use the software will likely to be higher when appropriate direct manipulation method is applied.

Suggestions for future researchers

Even with a relatively large base of 48 subjects, there is certainly a room for improvement. For example, if we had the availability of more resources, more diverse subjects can be sampled. In our experiment, the subjects were restricted to mostly computer science majors. Since our subjects were already proficient with computers, this will likely to cause some bias in our result. For future experiments, more general pool of subjects (i.e. from general public) will be appropriate.

The instruction confused users due to its lengthy nature. Perhaps the experiment can be divided into two sessions; one to get the user to be familiar with the environment and another to carry out the actual experiment.

If the subjects can be studied over a long period of time, the validity of the results will get better. A set of subjects could be given software with one of the following: direct annotation, click and type, and textbox. They would then use the given software for a certain length of time on their home computers. Periodically, they would be required to come in and participate in a short experiment session which measures how much time they took with the current version of software. This would give insight into not only which method is preferred and faster, but also gives out some idea for a learning curve.

Having a different set of photos can be another factor; if the library consists of people that the subjects don't know, then they will likely to be less interested. On another hand, if the library of their own personal photos were used, then perhaps their enthusiasm would be heightened.

Using the aforementioned ideas, there are numerous combination of experiments to be carried out. In addition, there are other methods of direct manipulation; the drag and drop concept is only one way of implementing it. Perhaps a future experiment can find the best method amongst the direct manipulation methods.

Refinement of theory

The results suggest that the textbox can be a faster way of annotating photos. However, over a longer period of use, direct annotation will likely to be the winner since the idea of typing same names over will not likely to be appealing to users. If there are many instances of a single person in a photo library (i.e. the family's newborn baby), the typing of the same name over will quickly get tiring and dull. And over time, as users get familiar with direct annotation, they will likely become faster than the textbox method; from our observations most users did not use the keyboard entry (clicking on the menu and pressing the letter of the person's last name would get you to the first instance of the last name beginning with that letter) feature that would have sped up the process. As the user begins to use the program more often over a longer period of time, they may start to use that feature, thereby speeding up the process.

Also, the issue of menu design come into play. Perhaps if the menu for performing direct annotation was done using a different menu design method, then it may speed up direct annotation more. There are many multiple-window design methods (Shneiderman, 1998) that could be used in designing another type of direct manipulation method. Maybe if there were two windows of names, organized by different categories (i.e. one window organized by last name, one by first name), the access time could be faster because the users would have two different methods to reach the desired name. For example, for a name like Bob Smith, one could use the first name category to find Bob quicker and for a name like Warren Beatty, the last name category would be faster (since the categories default starting at the beginning of the list). Another variation could be a short list of the most frequently used names that appear at the top, as it does in Microsoft Office product when changing fonts. Since empirical evidence is very limited in the area of menu design and the effects it has on speed, further research would help develop and test these theories further.

While direct annotation did not prove to be the fastest in the short run, we believe that the statistically significant high subjective satisfaction it received warrants further exploration of increasing the speed of direct annotation.

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