There's a lot of buzz around "mouse tracking" and analytic tools that record mouse position, like the super nifty Robot Replay. It's natural to wonder if mouse tracking might offer some of the value of eye tracking at much lower cost and much greater scale. I've written about the state of understanding of mouse and eye synchronization before; this post looks at a different viewpoint, setting a maximum bound on the potential relationship between the mouse and the eye.
Fitt's law states that the time to move the mouse from one point to another is heavily influenced by both the distance of the mouse from the target and the size of the target. While it's very easy to over-apply this to site and software design, it is a solid truism of human computer interaction.
This relationship between time and distance doesn't hold in the same way for eye-movements. Given this, research from Google surprisingly shows an increased duration of "sweeps", or leftward motions like carriage returns homing the eye to the next line of text.
Beymer, D., Russell, D. M., Orton, P.Z. An Eye Tracking Study of How Pictures Influence Online Reading, INTERACT Conference, Rio de Janerio, Brazil (September, 2007) PDF-515Kb.
While the researchers do see time increase with distance, looking at the data more closely shows this is due to a large number of correction saccades, not an increase in the duration of the first saccade.
To illustrate this point in another way, when the font size changes from 10 to 14 point, the average distance of saccades increase. So, you read very similarly with different font sizes -- the increased size neither speeds or slows your saccades, despite changing the total distance.
This finding is also supported by Silbert, Jakob, et al. (2000):
Evaluation And Analysis Of Eye Gaze Interaction - Sibert ...
5 Speed and accuracy of saccadic eye movements: Characteristic. ... 2 Fitts' law and the microstructure of rapid discrete movement..
Eye movement is largely independent of distance.
However, for visual items that are not well separated by white space, like typical successive lines of text in the Beymer study, the accuracy of saccades does decrease as the distinctiveness of the target decreases.
While operating system designers, and perhaps browser designers
-- any highly used software with toolbars -- do need to pay attention to button size, the key for good design for vision is not distance. It's an elusive property
of contrast, shape, color, and even typographical semantics.
To return the the title theme of this post, the mouse simply can't keep up in many cases. The eye is a capable of moving more rapidly than the hand can move the mouse. Hence, Silbert, et all, and many other researchers have been able to get user efficiency gains from gaze based selection. The challenge of course, is distinguishing between intentional selection and simple inspection (aka the Midas Touch
So, while mouse tracking can inform on where the users attention is focused, and it certainly is a great way to visualize user activity, the mouse is simply slower than the eye and destined to reveal less of the users behavior than eye tracking.