23
Figure 21. Index of performance as a function of task difficulty.
5.2.3. Summary
To sum up, the present exploratory study confirmed our prediction that Fitts’ law can be
investigated, with the help of zooming, far beyond the classic 10-bit barrier, without altering
the essence of Fitts’ aimed-movement paradigm. Zooming started to be useful at 8 bits;
beyond that threshold zooming amplitude was linearly dependent on task difficulty. Our MT
measurements for multiscale pointing were consistent with the simplest possible version of
Fitts’ law, MT = k * ID, a form which implies a constant information outflow, in keeping with
Fitts’ (1954) expectation and as predicted by our own model of multiscale target acquisition
(see Section 4).
5.3. Experiment 2: Effect of View Size on Bandwidth
We now turn to our experimental tests of Equations 6 and 7 on the effect of view size on
movement bandwidth. Whereas in Experiment 1 the view appeared in full-screen mode, in
Experiment 2 the view appeared in the middle of a black screen as a rectangular aperture
showing the current selection of the document. View center, coincident with screen center,
was permanently marked with a colored cross-hair. The task, alternatively reaching and
clicking on two targets located a long distance away from each other, was the same as in
Experiment 1.
5.3.1. Experiment 2a: Wide Range of View Size
Independent variables and procedure
The experiment involved two independent variables, task difficulty ID and view size V. We
used two levels of difficulty: ID = 14.6 and 17.9 bits. A 17.9-bit pointing task is roughly
equivalent to a 1 cm target at a 2.5 km distance. However, since the only effect of the (initial)
ID in our data-reduction strategy is to provide a longer time series for the evaluation of the
bandwidth, as can be understood from Figure 22, we ignore this factor.
View shape per se might have influenced navigation performance, so to avoid confounding
we resolved to vary view size in 2D, keeping a constant 1.33 aspect ratio. We explored six
levels of V, defined as the view’s half-diagonal, spread over a large range: 20, 40, 80, 160,
320, and 640 pixels (full screen).
More intriguing information
1. Økonomisk teorihistorie - Overflødig information eller brugbar ballast?2. The Folklore of Sorting Algorithms
3. Population ageing, taxation, pensions and health costs, CHERE Working Paper 2007/10
4. The Mathematical Components of Engineering
5. FASTER TRAINING IN NONLINEAR ICA USING MISEP
6. Artificial neural networks as models of stimulus control*
7. Hemmnisse für die Vernetzungen von Wissenschaft und Wirtschaft abbauen
8. Public-private sector pay differentials in a devolved Scotland
9. The name is absent
10. The name is absent