Statistical Concepts in Design

In seeking a manageable way to deal with variations for a large range of body sizes, it is a common practice for designers to construct drafting board manikins, three dimensional forms or computer simulations as human analogues. Often these analogues are based upon 5th, 50th or 95th percentile values. Limitations of this approach are discussed in this paper which demonstrate fallacies underlying the assumption that 1 the proportionality of various sized individuals is the same and 2 that percentiles for body dimensions are additive. Focusing on the 5th and 95th percentile body forms where deviations in size and proportionality are most severe, the report recommends an improved approach to portray the body size of these segments of the population in design problems. A statistical analysis is made of the tails of the height weight distribution to demonstrate the usefulness of subgroups or regression values. It is suggested that, for many design purposes, subgroup or regression values be used which would maintain statistical integrity in simulations and, at the same time, portray the ends of the distribution more accurately than is presently done.