Stochastic dominance and truncated sample data

The purpose of this paper is to examine whether the chance of making erroneous investment decisions can be reduced by applying stochastic dominance rules to truncated, rather than complete, sample data of rates of return. A simulation approach is used that assumes the rates of return follow the symmetric stable probability distribution. Using a variety of relationships between probability distributions of rates of return, it is demonstrated that sample truncation has the potential of significantly reducing sampling errors in the selection between alternative investments.