Human-Computer Interface Technology Design Lab 1:
Human Performance and Fitts' Law
Paul Fitts found a predictable relationship between the size of a target,
the distance we have to move to it, and the speed at which we can tap,
point at, click, or whatever on it. Today we'll do experiments to try to
determine this relationship. The typical Fitts' Law experiment involves
alternating tapping on two targets. Another variant involves targets appearing
randomly in a field, and tapping those.
Paper and pencil tapping task
-
Print out a number of copies of this
pdf tapping task template
-
Perform the experiment described on the template using at least 2 people
other than yourself. If you've got time, run lots of people -- the
more the better!
Paper and pencil signature task
-
Print out a number of copies of this
signature task template pdf
-
Perform the experiment described on the template using at least 2 people
other than yourself. If you've got time, run lots of people -- again,
the more the better!
"IDTest" program: evaluating pointing devices
We recommend that you run these tests in the warm, secure, and comfortable
HCI lab. If you'd like to run them at on your own machine at home,
however, here are
Microsoft Windows installation files
for the Lab 1 Java program from IBM. We make
no guarantees of merchantablility or fitness for a particular purpose,
blah, blah, blah: in other words,
you may have to thrash around with java to get
them to work.
- Run IDTest.exe from the Lab1 directory on one of the PCs in the
HCI lab.
-
Perform the tests supported by this program for at least three pointing
devices of your choice (e.g. large vs. small mouse, mouse vs. trackball,
touchpad vs. trackpoint, mice with very difference shape). Try to pick
one that you feel will be clearly superior, one that will be clearly
inferior, and one you might not know. Feel free to insert your favorite
mouse (pointing device) in here, but be sure to work with your group and
collaborate to evaluate the same three devices. Try to make sure at
least one group member is not familiar with your favorite pointing
device.
-
The program writes the data from its tests to a text file called IDTest.dat.
Save a copy of this file and import it into a spreadsheet.
-
Use this data to analyze and compare performance of the three devices,
answering the following questions:
-
Which tasks had the biggest differences between devices?
- Which tasks had little or no differences between devices?
-
Plot time vs ID and compare to Fitts' model
(intercept+slope*time).
-
Which device performed the best? (lowest mean time)
-
Which pointing device sucked?
-
Theorize as to why you think the best device worked better than the worst.
Putting it all together
-
Compare the hand/pencil results to the IDTest results,
doing whatever analysis you feel is necessary to make a valid
comparison
(Hint: don't forget the logarithm when calculating the Difficulty
Index, and be sure to use the modern definition of Fitt's law.).
-
Discuss the validity of Fitts' Law in your study.
-
Discuss the benefits of Fitts' Law.
-
Based on your results, discuss pros and cons of the pointing devices:
suggest improvements or new designs.
Further reading
The ID Test is written in Java by Barton Smith at the IBM Almaden
Research Center, partially for this class. You can download it or get
more information about running it from
Barton Smith's IDTest page.
CS436: Human-Computer Interface Technology,
Princeton University, Fall 2007
Authors: Shumin Zhai, C. Sapp, J. Hainsworth, P. Cook
Copyright © 1999-2007, Princeton University and Stanford
University