Human beings come in such a variety of sizes and shapes that no single generic size or shape corresponds to more than a tiny percentage of them. As a result, in order to fit or to function properly, almost everything worn by or covering human beings--or anything intended to hold, transport, or support them--must be made in a number of different sizes and shapes. This statement represents an assumption that underlies virtually every application of human factors expertise.
For some products, manufacturers have already recognized that in order to accommodate a substantial market, sizes must be varied on each of the several shape dimensions of the product. Shoes, for example, are made in several sizes of length and width, as well as several arch shapes and toe shapes. Shirts are made in several sizes of neck diameter, sleeve length, and chest circumference. The multiplicity of sizes and shapes is necessary so that the product will fit a large proportion of the population for which it is designed. This is required whether the motive is larger sales for the maker, better fit and comfort for the consumer, or better compliance with statutory regulations.
How does a manufacturer decide which dimensions of the product to vary and which sizes to make on each of those dimensions? Whether the manufacturer intends to make only one size or several sizes, the problem is the same: Which sizes should be made? In addition, how is the trade-off of increased costs to increased fit to be determined if the number of sizes is increased? The manufacturer needs an accurate way to compute the additional cost of making multiple sizes. Furthermore, how can increased fit to the population of users be determined?
In the absence of empirical and quantitative procedures that can predict the precise increment in target market fit, the conservative approach suggests minimizing costs; this explains why many products are made in one or only a few sizes. The problem becomes critical if a proper match between the size and shape of the user and that of the product affects the user's safety. In this case, the conservative approach also increases accidents and reduces efficiency.
We have used well-documented principles of human factors theory and research to create quantitative computer programs capable of specifying the number of different-sized versions of the product needed for each target market. Those programs also specify the precise size of each sized version. The same programs are used to generate the cost-benefit ratios of increasing the number of sizes of any product.
In this article we describe how the application of these principles removes all of the guesswork from "one size fits all" failures. We have validated these procedures for the products of one large manufacturer. These principles are general and applicable to virtually any product and market. The consequence of their application is a better performing and safer product.
Applying Human Factors Principles to Shoe Sizing
At least four dimensions of feet affect the fit of shoes: (a) the length of the foot from the heel to the widest point between metatarsal and ball, (b) the circumference of the foot around this widest part, (c) the additional length added by the toes, and (d) the curvature of the arch. For a particular shoe to fit a particular foot, the sizes of these shoe dimensions must match those of the foot.
At present, only two dimensions are measured and fit in shoe stores (length from heel to toe and width at the widest part), neither of which corresponds to any of the dimensions identified by experts in shoe fitting. Furthermore, the remaining critical dimensions are rarely varied within a given style. Consequently, customers experience many compromises in manufactured shoe sizes- --compromises that often result in poor-fitting shoes.
On the reasonable assumption that a sample of people would reveal wide variation in sizes along each of these four dimensions, how can the shoe manufacturer determine how many different sizes of shoes to make to provide a well-fitting pair for most people? Ten lengths and five circumferences alone produce 50 independent size combinations, without even considering toes and arches. And will 10 shoe lengths cover the entire range of foot lengths, or are 16 or 22 necessary?
Three determinations are needed for the computation procedures that can provide the answers to these questions: (a) identification of the relevant dimensions of feet that affect fit, (b) measurement of the amount of variation of these dimensions in the target market, and (c) determination of the maximum mismatch between size of foot and size of shoe along each dimension that still gives a comfortable and functional fit.
Variables that Affect the Number of Sizes
Several variables influence the number of size combinations that must be made to fit the people in a target market. Two of these are beyond the control of the manufacturer; they depend on the sizes and shapes of people in the target market or on the nature of the products themselves. For the remaining variables, the manufacturer may be able to adjust the product or its marketing to reduce the number of sizes needed. Variables include (a) number of relevant dimensions, (b) independence among the relevant dimensions, (c) expanding mismatch tolerances, (d) relative size of the target market, and (e) manufacturing precision.
Computation Procedures
A manufacturer must be able to interrogate the different subgroups of the database (e.g., genders, ethnic groups, ages) using different assumptions until the optimal combinations of sizes of dimensions needed to produce desired market shares are determined. This requires a dedicated computer program. In addition, the results of these interrogations should ideally be interfaced to both computer-aided design systems and computer- aided manufacturing systems.
Here we describe the characteristics of the kind of program that can accomplish this. Its demands are easily handled by any 486 PC with 500 MB of memory available. We have written such a program in a Windows version with a large database broken down into 10 subgroups, all measured on 10 dimensions, so we have demonstrated that the programming can be done using Windows software.
This kind of program can be interfaced to computer-aided design and manufacturing systems so that both designers and engineers can have access to these data and can inquire during the design process, or during evaluation of quality control issues. Furthermore, marketing personnel can use the program to evaluate marketing assumptions and to make marketing decisions.
###