Your Names:

Massachusetts Institute of Technology

Department of Electrical Engineering and Computer Science

Fall Semester, 2006

Department of Electrical Engineering and Computer Science

Fall Semester, 2006

The point of this activity is to get a hands-on understanding of Fitts's Law. You need a web browser to run the experiment workbench; you'll find a link to it at the top of the 6.831 home page, or you can enter its URL directly: http://courses.csail.mit.edu/6.831/handouts/ac3/fitts.html

Using the initial settings of the workbench, try doing the task. Collect about 10 mouse clicks for the initial settings, and record the results below. Note that ID = log(D/S) and STDERR = STDEV / sqrt(N).

Size (S) |
Distance (D) |
Index of difficulty (ID) |
Count (N) |
Mean time (T) |
Standard deviation (STDEV) |
Standard error (STDERR) |

Use the workbench to run trials with targets of varying size
*or* distance. If you're in the left half of the lecture
room, from your point of view, you should vary the size. If
you're in the right half, you should vary the distance.

Use at least 3 sizes or distances, and collect about 20 mouse click times for each size or distance. The output textbox holds 20 lines, so just fill it up. Keep both targets the same size in each trial. Don't try to save time by using targets of two different sizes at once. Vary geometrically (i.e., 1,2,4,8,... not 1,2,3,4,...). Keep the distance D significantly larger than the size S. The same person should do all 3 trials.

Record the results below.
**What kind of relationship should you see between the index
of difficulty (ID) and the mean time (T)?** Plot
the points roughly in the space on the right to see if you get that relationship.

S |
D |
ID |
N |
T |
STDEV |
STDERR |

The targets we've used so far have been square. Now let's
think about rectangular targets of size *W* by *H*,
where W and H are significantly different. Think about how we
explained Fitts's Law in terms of the human information
processor model, and try to extend Fitts's Law to cover this
case. Here are some possible hypotheses:

- S = W (only the width matters)
- S = H (only the height matters)
- S = sqrt(W^2 + H^2) (the diagonal dimension matters)

Write down *your* hypothesis (which may be one of the above, or completely different), and briefly justify it.

Now devise and run a set of trials to test the relationship between T, W, and H. For example, you might compare tall skinny targets and short fat targets.

- Do at least 4 different trials.
- Collect about 20 mouse click times for each trial.
- Keep both targets the same size in each trial. Don't try to save time by using targets of two different sizes at once.
- Vary geometrically (i.e., 1,2,4,8,... not 1,2,3,4,...).
- The same person should do all 3 trials (but use a different person than you did before).
- Keep the distance D significantly larger than the size S.

Compute index of difficulty (ID) according to your hypothesized relationship from the previous problem. Plot your points roughly in the space on the right. Does your hypothesis look reasonable?

W |
H |
D |
ID |
N |
T |
STDEV |
STDERR |