**Back to
Annotated
Bibliography**

**A MULTIVARIATE
ANTHROPOMETRIC METHOD FOR CREW STATION DESIGN: ABRIDGED (U)**

**AL-TR-1992-0164**

**Gregory F. Zehner**

**CREW SYSTEMS DIRECTORATE
HUMAN ENGINEERING DIVISION
ARMSTRONG LABORATORY**

**Richard S. Meindl
Jeffrey A. Hudson**

**DEPARTMENT OF
ANTHROPOLOGY AND
SCHOOL OF BIOMEDICAL SCIENCES
KENT STATE UNIVERSITY
KENT, OHIO**

**FINAL REPORT FOR PERIOD
JANUARY 1989 TO DECEMBER 1992**

**APRIL 1993
**

**ABSTRACT: "Body size accommodation in USAF cockpits is
still a significant problem despite all the years of experience and the many
aircraft designs that have been developed. Adequate reach to controls, body
clearances (particularly during escape) and vision (internal and external), are
all functions of pilot body size and position in the cockpit.**

**One of the roots of this problem is the way cockpit
accommodation is specified and tested. For many years the percentile pilot has
been used. This paper describes the errors inherent in the "percentile
man" approach, and presents a multivariate alternative for describing the
body size variability existing in a given flying population. A number of body
size "representative cases" are calculated which, when used properly
in specifying, designing, and testing new aircraft, should ensure the desired
level of accommodation.**

**The approach can be adapted to provide anthropometric
descriptions of body size variability for a great many designs or for computer
models of the human body by altering the measurements of interest and/or
selecting different data sets describing the anthropometry of a user
population."**

**[The Abstract is
identical to that of the original version under the same title.] **

**PREFACE: [The Preface does not, as one would
expect, contain an explanation regarding the reason for issuing an abridged
version of the original Technical Report of essentially the same title and
authored by the same investigators.]**

**INTRODUCTION: "Military personnel of every
size and shape much be able to operate complex equipment safely, effectively,
and comfortably. Personnel charged with the specification and procurement of
complex workstations and personal protective equipment are continually
challenged by the need to accommodate and fit very large numbers of an
increasingly heterogeneous population. In writing specifications, the goal is to
ensure that the body size and proportions of most of the population will be
accommodated in each item or system to be procured. Traditionally, this has been
done by using percentiles to specify the portion of the population that must be
accommodated. Typically, specifications read: "the system shall be designed
to allow safe operation by the fifth percentile female pilot through the
ninety-fifth percentile male pilot." What is not specified is how the 5th
and 95th percentile pilots are defined.**

**The purpose of this report is to point out the
drawbacks inherent in the percentile approach, and the present a more suitable
method for describing variability in body size. The proposed method is based on
the pioneering work of Bittner et al. (1986). For a detailed statistical
description of the technique, see Meindl et al. (in press)."**

**[The work by Meindl, Hudson and Zehner,
cited above as being "in press," has an Armstrong Laboratory technical report number of the 1993 series, thus
appearing to have been published later than the abridged version. This is even
more confusing since the "abridged" version, appropriately, has a
later publication date.]**

**
Percentiles: ". . . while a 5th percentile Stature value can be accurately
located, that value tells us little or nothing about the variability of other
body dimensions of individuals with 5th percentile Stature. Consider Weight, for
example. Individuals of 5th percentile Stature in the 1967 survey ranged from
125 lbs. (less than 1st percentile Weight) to 186 lbs. (74th percentile Weight).
. . . [What might appear to be a] logical next step is to consider the fifth
percentile for both measures. It is common for people to assume that the 5th
percentile for both Stature and Weight represents a "5th percentile"
person. In fact, only 1.3 percent of subjects in the 1967 survey were smaller
than the 5th percentile for both measures, while 9% were smaller for one or
the other. The problem is compounded with each additional measurement used to
specify the size of the USAF individual. Thus, at worst, use of percentiles can
mean that workspaces or equipment are not suitable for anyone. . . ."**

**"The pitfalls
attendant upon the use of multiple percentiles can be illustrated by considering
the body dimensions critical to cockpit design. . . . Sitting Height, Shoulder
Breadth, Buttock-Knee Length, Knee Height Sitting, and Functional Reach.
Generally a group of measures such as this is listed in a specification or
standard along with 5th and 95th percentile values for each. This gives the
impression that if these values are used as design criteria, 90% of the
population will be accommodated. . . . There is no difficulty in identifying the
individuals who constitute 90% of the population in Sitting Height. However, . .
. when those same individuals are screened for 5th-95th percentile Buttock-Knee
Length values, their numbers drop. With application of each additional cockpit
dimension, the group diminishes until, finally, it represents only 67% of the
population."**

**THE MULTIVARIATE
ACCOMMODATION METHOD: "The multivariate accommodation method is an
alternative to the percentile . . . It corrects the deficiencies of both while
retaining the concept of accommodating a specific percentage of the population
in the design. Briefly, the multivariate accommodation method is based on
principal component analysis, which reduces a list of variables to a small
manageable number, and then enables designers to select the desired percentage
level of a population to be accommodated. This percentage level is accommodated
in a way which takes into account not only size variance but proportional
variability as well -- i.e. not only individuals who are uniformly large or
small, but those whose measurements combine, for example, small torsos with long
limbs, or vice versa.**

**A number of examples of
the approach are given . . . beginning with a very simple two-measurement
example, building to a basic cockpit layout, and concluding with a fairly
complex 11-variable computer man-model." **

**DISCUSSION: "There
are a number of multivariate statistical techniques which could be utilized to
determine similar combinations of body size test cases. The technique described
here, however, when combined with lists of minimum and maximum values, gives a
much more accurate description of the body size and proportional variability
existing in the population and, if used in designing workspaces, will greatly
reduce the accommodation problems experienced by users. This assumes , of
course, that the seat, rudder, and other adjustable components can be adjusted
in sufficiently small increments. Without such adjustability, it may be
necessary . . . to pick many more representative cases than the numbers
suggested here to ensure the desired level of accommodation. However, for the
purposes of writing anthropometric specifications, large numbers of representative
cases may overwhelm the designer and thus, be counterproductive." **

**Back to
Annotated Bibliography**