Last edited 4jun97 by jestabro@new.math.uiuc.edu
Find this document at http://new.math.uiuc.edu/projects/soccer.html

Experiments on the relative value of head-tracking and stere-opsis in CAVE immersive virtual environments.

Design and execute the multivariate experiment of playing exactly one game (3 minutes, 3 levels of difficulty) under each of the four modes:

M3 head tracking + stereo display
M2 head tracking + mono display
M1 no tracking + stereo display
M0 no tracking + mono display

NB There is an ambiguity with what is meant by no tracking. Provisionally, we shall mean no head tracking. The wand must be tracked for now since that's the navigation device.

The result of a game, the score, is the numerical output. Since the user can learn the game, we shall need at least 24 naive test subjects. Nonetheless, we cannot control for native navigation and game playing skills regardless of which mode. Maybe one should calibrate the player by a different game which requires similar skills (navigation, aim etc). Or, we need multiple subjects in each of the 24 classes. I suggest that we use the 4! model first to formulate an hypothesis, and then redesign the second experiment with fewer modes but better sampling.

Or, we need multiple subjects in each of the 24 classes. I suggest that we use the 4! model first to formulate an hypothesis, and then redesign the second experiment with fewer modes but better sampling.

Or, Alex suggests, we have each subject do 4 runs but only in one mode. This would reduce the number of subjects needed.

The statistical model is a very simple one, to estimate the additive influence of the two factors on the ability to navigate and manipulate an immersive virtual environment, as measured by the game score. Second, to estimate which of the two factors is the more significant.