Experiment Design

Assignment #7: Experiment Design

Assigned: March 19th, 2009
Due: April 15th, 2009

Objective: In this assignment, you will develop a plan for testing your user interface. This includes deciding the tasks to support, the users to test, where to test, the method of testing and the metrics used to evaluate success.

What to do:

1) Decide 3-4 tasks that you would like to support during your user testing. The tasks should provide the user with a complete and realistic feel for how the eventual application will work (or, for large projects, how some aspects of the future application will work). This should include both high-level user tasks (decide where you want to eat dinner using this interface), as well as specific, low-level tasks (use the interface to search for a restaurant open at 10PM).

2) Decide the context where you will test each of these tasks. The location should be as realistic as possible, while allowing for controlled measurement and evaluation.

3) Decide how you will recruit participants for your final user study. Aim for testing each task with between 5-15 users. Each task should be tested by appropriate users fitting the demographic characteristics outlined in your persona descriptions.

4) Determine whether it is possible to include a control condition for the testing. For example, if you have designed a new interface for web search, a likely control condition would be using Google search. This allows you to measure how your new interface compares with current alternatives. Ideally, you should test each task using both your own prototype and at least one control condition. Those without a suitable control condition should compensate by either including more users and/or more performance measures (see below).

5) Decide the methods you will use to evaluate performance for each task you want to test. This could include quantitative methods for metrics such as efficiency and error rate, as well as qualitative methods for metrics such as subjective satisfaction. Ideally, you will test each task using at least two quantitative and one qualitative method. Decide how you will measure user performance (for example, task execution time or number of errors), behavior and/or responses (for example, using audio, video, screen recorders and/or detailed notes) .

6) Decide how you will address any learning and/or fatigue effects - either between tasks, or between the control and experimental conditions.

7) Decide the conditions of success. For experiments with a control, this may be some relative improvement over the control condition. For experiments without a control, this may be some absolute measure of efficiency, accuracy, subjective satisfaction, etc.

8) Decide the materials you will need to conduct the usability test. This includes the script you will use to conduct the test, the prototype itself, and any questionnaires you will administer before and/or after testing. This should include a demographic questionnaire capturing important user details (age, gender, education, experience with your proposed kind of technology and/or application, etc), as well as informed consent and records release forms allowing you to conduct the test with the user and/or capture audio, images and video. (Example forms are provided here and here).

What to turn in:

The preferred method of turn-in is to create a page for your submission on the course Wiki, linked from your final project page. Create a dedicated page for your submission, including each of the following components:

1) Cover sheet including yourself and your partners' names, and your chosen focus. Note the time, duration and attendance of each brainstorming session. Include a paragraph describing what each person contributed to the assignment [1/2 page].

2) A list of tasks that you intend to test during your usability testing [1 page].

3) Description of your control condition(s) and experimental variants, if any (it is highly recommended to include a control if possible) [1/2 page].

4) Details about how many users that you will test for each task, and how you plan to recruit participants. If the same persona applies to multiple tasks, it is fine if the same user is tested for all of them, making sure to account for any learning / fatigue effects [1 page].

5) Description of the methods that you will use to evaluate the usability of each task, and how you will measure document user behavior and responses. It is recommended to evaluate each task using at least two quantitative and one qualitative methods. Mention how you will address any learning and/or fatigue effects - either between tasks, or between the control and experimental conditions. Provide the location where you will test each task [2-3 pages].

6) A list of conditions of success for each of the tasks. [1/2 page]

7) A description of the supporting materials that you will need for conducting your user test - including forms, testing scripts, questionnaires, props, and the prototype itself. [1 page].

The total length of your report should be less than 8 pages. Brevity, clarity and focus on the goals of the assignment will be rewarded.

E-mail a link to your group's project page to the professor and TA before 11:59PM on Wednesday, April 15th.

Please contact the professor or the class TA using the course mailing list if you have any questions with this assignment.