Dealing with Dynamic Systems:... 37
version is 0.38 and 0.14 for 2 and 4 goals, for the graphical version it is 0.08
and 0.36, F(i,72) = 13.36*). Given numerical presentation and only 2 goals the
causal analysis of the system was better than for the graphical version. With
4 goals the effect was opposite, producing better identification under graphi-
cal than under numerical presentation.
A final hypothesis specified that a 2:1 relation of exogenous to endo-
genous variables (condition I, high degree of controllability) would result in
better control than the reverse relation (condition III, low degree of control-
lability). Contrary to this expectation, subjects showed best system control
under condition III, medium control under II, and worst control under I (mean
QSC for I, II, and III are 4.75, 6.98, and 8.13; F(2,72) = 5.26*, respectively).
Discussion. It is not surprising that the match or mismatch between prior
knowledge and implemented system structure shows strong effects on knowl-
edge acquisition and control performance. This illustrates an effect which
could implicitly occur in studies where subjects’ prior knowledge remains
uncontrolled. In addition to the classical methods of analyzing effects of prior
knowledge by comparing novices and experts (Reither, 1981) or by compar-
ing a semantically embedded system with an abstract one (Hesse, 1982), the
method of constructing two semantically equivalent systems which corre-
spond differently to prior knowledge proves to be useful.
Presentation format per se was not a critical factor. However, it is obvious
that depending on the nature of the task, differential effects occur: In order
to cope with the more complex task the graphical presentation which is less
precise in presenting system information yields better results. It remains
unclear why different degrees of controllability did not have their expected
effect on control performance: Selective motivational effects (the task which
appears more difficult motivates subjects to work harder on it during the
control phase) could be one possible explanation for this surprising result.
4. General Discussion
The main conclusions of the five experiments presented above are as follows.
First, a research program of manipulating formal system attributes experi-
mentally contributes to an exploration of the differential influence of these
attributes on knowledge acquisition and knowledge application. Second, the
dependent variables for the amount of causal knowledge and precision of
system control seem useful indicators in studies designed to analyze acquisi-
tion and application of knowledge. Third, the distinction between different
kinds of relations within a system (endogenous and exogenous effects,
Eigendynamik, side effects) should be further explored because QSI and QSC
seem to be influenced by that kind. Fourth, the formal approach is not
restricted to systems with “artificial” semantic embedding but is also useful
for semantically rich systems. Fifth, besides variations of system attributes