Balance of Power and Aggression: An Experiment in Simulation

There are many theories in the field of international relations
that relate the level of aggression in the international
system to the distribution of power among nations. Unfortunately,
most such propositions are difficult to validate or refute in the
real world for which they were formulated. For this reason an
experiment was devised which attempted to east two such conflicting
positions, "balance of power", and "preponderance of power",
into testable form. The experimental paradigm employed was a modified version of the
two person mixed-motive game developed at the Systems Analysis
Corporation, Santa Monica, California. Fifty-four subjects participa
ted in the experiment. Twenty-seven subjects were placed into
each of three treatment conditions: "balance of power": "moderate-inequality
of power"; and "preponderant-inequality of power". In
the "balance of power" condition all subjects were allotted twenty
shocks. In the "moderate- inequality of power condition" one subject
received twenty- five shocks while his dyadic partner received fifteen
shocks. In the "preponderant-inequality of power" one subject
received thirty shocks while his dyadic partner received ten shocks.
All subjects played a total of fifteen games. A count was made
of each subjects responses on eight game-play variables which were
viewed as indices of cooperative or aggressive behavior. Additional data were also gathered with respect to each subjects sex,
academic aptttude and attitudinal attributes. The purpose of the
additional data was to identify variables, other than the treatment
condition, affecting game behavior. Thus, facilitating future
experimentation. The results of this research indicate that balanced situations
tend to produce more cooperative non-aggressive behavior, while imbalances
of power tend to generate less cooperative and more aggressive behavi.or. In addition, a balanced situation, when compared
to the imbalanced situations, was found to increase the likelihood
of participants never employing shock capabilities. Both this experiment
and a pilot experiment found de-escalation cycles diffficult
to instigate and unlikely to develop from natural causes.
Finally, attitudinal, sex, and SCAT data did not appear to be meaningful factors in explaining game play. However, post-experiment
sample data indiciated these factors must receive continued attention
in future research.