R. Shumaker and S. Lackey (Eds.): VAMR 2014, Part I, LNCS 8525, pp. 406–417, 2014.
© Springer International Publishing Switzerland 2014
Matching Levels of Task Difficulty for Different Modes
of Presentation in a VR Table Tennis Simulation
by Using Assistance Functions and Regression Analysis
Daniel Pietschmann
1
and Stephan Rusdorf
2
1
Chemnitz University of Technology, Institute for Media Research, Chemnitz, Germany
2
Chemnitz University of Technology, Department of Computer Science, Chemnitz, Germany
daniel.pietschmann@phil.tu-chemnitz.de,
stephan.rusdorf@informatik.tu-chemnitz.de
Abstract. UX is often compared between different systems or iterations of
the same system. Especially when investigating human perception processes in
virtual tasks and associated effects, experimental manipulation allows for better
control of confounders. When manipulating modes of presentation, such as
stereoscopy or visual perspective, the quality and quantity of available sensory
cues is manipulated as well, resulting not only in different user experiences, but
also in modified task difficulty. Increased difficulty and lower user task perfor-
mance may lead to negative attributions that spill over to the evaluation of the
system as a whole (halo effect). To avoid this, the task difficulty should remain
unaltered. In highly dynamic virtual environments, the modification of difficul-
ty with Fitts’ law may prove problematic, so an alternative is presented using
curve fitting regression analyses of empirical data from a within-subjects expe-
riment in a virtual table tennis simulation to calculate equal difficulty levels.
Keywords: Virtual Reality, Performance, User Experience, Spatial Presence,
Table Tennis Simulation.
1 Introduction
The user experience (UX) of a given technology is a central research question in HCI.
For example, social and behavioral sciences are interested in cognitive and physiolog-
ical short-term and long-term effects on the user and the role that the perception has in
the user experience (UX) as a whole.
UX is often compared between systems or within different iterations of the
same system. The experimental manipulation of investigated variables within
the same system allows for better control of confounders [1] than when comparing
different systems. Still, a general drawback is an unintentional difference in task
difficulty in different experimental conditions. The more difficult the task and the less
successful the user, the more negative his subjective UX with the system is: Negative
attributions from failing the task spill over to the evaluation of the system as a whole
(halo effect). A possible solution is to create equally difficult tasks in all experimental
conditions.