交互代写 | CSC 428 / CSC 2514h – Human-Computer Interaction Fall
CSC 428 / CSC 2514h – Human-Computer Interaction Fall 2020
ASSIGNMENT 2: CONTROLLED EXPERIMENT
This assignment is a hands-on exercise in designing, building, conducting, and analyzing a controlled experiment.
In this assignment, you will be investigating techniques that can facilitate the task of pointing and selection. Your
goal is to find the most efficient way for users to select on-screen targets using a computer mouse. You have
narrowed the options down to two techniques, and would also like to compare these techniques to a standard
1) Point Cursor: A standard cursor defined by a single 2D point.
1) Area Cursor: A cursor that covers a circular area and can select any target it intersects. If multiple targets are
intersected, the cursor in the middle of the area cursor is used for selection, like a traditional point cursor.
3) Bubble Cursor: A dynamically sized area cursor, that grows and shrinks so that it always selects exactly one target.
Code segments that implement these techniques, with instructions on compilation, will be available on the course
website. In addition, the program will allow you to try out the three techniques. You should familiarize yourself with
the techniques before attempting the rest of the assignment.
Once you are familiar with how the techniques work, you must design and build an experiment that will allow you
to measure and compare user performance in using these three techniques.
An important part of experimental design is to decide which tasks to evaluate. You must choose an appropriate task
for the participants. Remember the task should define the structure of each trial in the experiment.
Once you choose a task you must choose two additional independent variables for the task which you can vary.
There are many possibilities. Think about what would be most important for the experiment, and provide
justifications for your choice. To limit the size of your experiment, your independent variables should take on 3
possible values. With the addition of the 3 selection techniques, your experiment will thus have 3 total independent
variables, with a total of 3 x 3 x 3 = 27 conditions.
You then have to decide how many trials you will have for each condition, and how many blocks of trials for all the
conditions. Your experiment length should be around 30 minutes.
You also have to decide on how you will present the trials (randomization, counter balancing, etc.)
Your design should be a within-subjects full factorial design: meaning that all subjects do all conditions for all
Coding it up
Once you’ve designed the experiment, you need to write some code to present the task and conditions to your
subjects, and record their performance.
Using the framework given in the sample code and discussed in tutorial, implement your experimental design. Note
that the sample code is indeed a sample: you are expected to make changes to the code to implement your design.
This may involve modifying the actual techniques, if required for your chosen independent variables.
Be sure to allow for rest periods.
You also need to figure out what your dependent variables are, and how you will handle errors that occur.
Your software should implement all of these aspects of the experiment – once the experiment is started, you
shouldn’t need to manually intervene.
Try out your experimental setup yourself first, and ensure all bugs are ironed out, before subjecting others to it!
You should have three participants do your experiment. You should discuss in your report what a preferable number
of participants would be if you had the time/resources.
You must use the provided consent form posted with this assignment to obtain consent from participants. This study
has been approved by the U of T Research Ethics Board, and you must follow all instructions outlined in the consent
form. Please read it carefully.
Running the Study
The study should be run on a desktop or laptop. A mouse should be used for input. The study must be run remotely
due to the COVID-19 situation. You will send the study software to participants. Ask participants to complete the
study in a quiet location to minimize distractions. It is suggested that you connect with them remotely to ensure
they understand how to use the study software, but you do not need to monitor the entire study. At the end of the
study, a data file should be generated, and participants will send this back to you for analysis.