CMSC 434 - Fall 2007
Prof Guimbretiere

Introduction to Human-Computer Interaction
Home Contact Syllabus Schedule HW & Project Links

Homework #7: Quantitative evaluation
(Due 12/03/07)

1) (50 points) KLM model

Bellow is a simple calculator interface. To perform and operation one might either use the mouse and click on each key, or one need first to set the focus to the calculator by clicking on the text field, and then use the keyboard to enter digit as well as simple operations (/, *, -, +, =) and use the mouse to access other operations. You might assume that the sqrt, % 1/x operation take place as soon as the key is pressed.

Using the KLM model, predict the time it will take a user to:

  1. (15 points) Compute the square root of 348*34 (using the sequence 3 4 8 * 3 4 = sqrt) using the mouse;

  2. (15 points) Compute the square root of 348*34 (using the sequence 3 4 8 * 3 4 = sqrt) using the keyboard;

  3. (20 points) Do you think this estimates are realistic? Justify your answer by measuring the time it takes you to perform these operations on your computer. To limit the influence of measuring errors, please measure your performance at least 5 times. Be sure to describe how you conducted your experiment.

In all case, you can assume that the user is starting the interaction with her hands on the keyboard. You should consider setting the focus as an operation selecting a command. Be careful to explain your derivation (including which rule you are using) for partial credits. You can ignore the system response time and assume the following values for the KLM parameters: tK = .15s, tP =1s, tH = .3s, tD = (.9*n + .16*l)s, tM = 1.2s.

 

2) (100 points) Measuring Fitts' law using a mouse

For this homework you will build a simple interface to measure Fitts' law parameters. For this simplified experiment, you will measure how fast a user can point between 2 squares shown on the screen for different value of the index of difficulty (ID). To do so, we are asking you to design a simple interface to measure Fitt's law parameters.  The interface should: 

  • for each trial, present a start-up screen so that the user can get ready,

  • for each trial, present a pairs of squares on the screen and let the user first click inside the left most square then click back and forth between the 2 squares as fast as they can for at least 10 times (users need to click inside the square each time),

  • for each trial, record the time it took the user to click between the 2 square (10 samples are recorded). The timing method should let you measure sub-second time (a 50ms minimum is a good target).

  • for each trial, present a completion screen,

  • present a series of trials spanning the index of difficulty from 2 to 6. For each index of difficulty, your program should measure at least 3 possible width/distance combinations. Since you are recording 10 action per condition, your system should at least record 150 trials per users,

  • present a completion screen at the end of the test.

For this project, you have to choose the specifics of how to implement your project within certain limits. All projects must be able to run under Windows XP, or 2000 (or NT). You can expect that the machine you will be running your project on will have a Java 2 virtual machine, and can execute Flash or JavaScript web pages. Finally you can expect the computer to have at least a CD-ROM drive. You are free to pick any programming languages (C++, Virtual Basic, C#, Java, Flash, Director...) and development environment you wish as long as the resulting program runs on the configuration described under the Platform section. A simple JavaScript example illustrating the key program components needed can be found here.

Deliverables

Your homework should contain: 

  • (40 points) The executable code (in running order). Please, specify accurately your platform, operating system, language used, needed software, which file to run. For this question grading will consider how close your interface is from the specifications described above.

  • (40 points) A clear an orderly presentation of your experimental results from data you collected from at least 6 different users. (If you are using fellow students as subjects you don't need to apply for a human subject protocol review). You report should include:

    • a graph showing for each ID the average "connection" time for each users,

    • a graph showing for each ID the average "connection" time across all users as well as the trend line,

    • a description of any problem you might have encountered during the experiment

  • (20 points) An explanation of the discrepancies between predicted and measured times.

No statistical analysis is necessary for this question.