One of the fundamental problems faced by military planners is the assessment of changes to force structure. An example is whether to replace an existing capability with an enhanced system. This can be done directly with a comparison of measures such as accuracy, lethality, survivability, etc. However this approach does not allow an assessment of the force multiplier effects of the proposed change. To gauge these effects, planners often turn to war-gaming. For many war-gaming experiments, it is expensive, both in terms of time and dollars, to generate a large number of sample observations. This puts a premium on the statistical methodology used to examine these small datasets. In this paper we compare the power of three tests to assess population differences: the Wald-Wolfowitz test, the Mann-Whitney U test, and re-sampling. We employ a series of Monte Carlo simulation experiments. Not unexpectedly, we find that the Mann-Whitney test performs better than the Wald-Wolfowitz test. Resampling is judged to perform slightly better than the Mann-Whitney test.