The Accuracy of Consumer-Made Body Measurements for Women's Mail-Order Clothing

H U MA N FACT 0 RS, 1994,36(3),557-568

JONGSUK C. YOON, Yonsei University, Seoul, South Korea, and ROBERT G. RADWIN,t University of Wisconsin, Madison, Wisconsin

An anthropometric study was performed on 103 women 19 to 50 years of age. Each subject measured 19 dimensions of her own body and then performed similar measurements on a random partner using a conventional tape measure with 1/,6- inch (l.6-mm) precision. The experimenter measured all subjects using a laboratory-grade anthropometer and tape measure with 1.0-mm precision as a standard. Subjects' measurements of different body dimensions had significantly different average errors, ranging from -4.54 cm to +6.15 cm. Hip circumference, which is often utilized as a key dimension for garment sizing, was significantly undermeasured (M == -4.54 cm). Measurements having the least error included waist, bust, and neck circumferences and shoulder and waist heights. Subjects' measurements of their own bodies had greater absolute error (M == 4.10 cm, SD == 6.06 cm) than those of their partners (M == 3.34 cm, SD == 4.86 cm). Although subjects used a tape measure having 1.6-mm precision, 97% reported measurements using only 1/4-inch (6.4-mm) precision or less. Subjects overestimated their own stature by an average error of 0.68 cm (SD == 4.02 cm) and an absolute error of 2.26 cm (SD == 3.39 cm).

INTRODUCTION be labeled by the same size code, garments made by different manufacturers are often Along with the rapid growth in catalog made to fit different figure types, body mea- shopping for clothing (Smallwood and surements, and proportions (Fellingham, Wiener, 1987), there is a growing need for an 1991). Also, clothing manufacturers continu- anthropometric sizing system for women ously revise their own size code standards. (Delk and Cassill, 1989; LaBat, 1987). Most Consequently, the standard body measure- clothing manufacturers and designers are re- ments associated with specific size codes luctant to conform to a unified standard set of have varied, not only among different manu- body measurements (Belkin, 1986). Even facturers but also within individual manufac- though women's ready-to-wear clothing may turers over time. Furthermore, size labels at-

I Requests for reprints should be sent to Robert G. Rad- tached to garments do not inform consumers win, Ph.D., Department of Industrial Engineering, Univer- sity of Wisconsin, 1513 University Ave., Madison, WI of the body measurements associated with 53706. particular size codes.

Because mail-order clothing cannot be the same accuracy and (2) consumers mea- tried on in advance of purchase, catalogs as- sure their own bodies and others' bodies with sist consumers by listing exact body measure- the same accuracy. ments for key body dimensions that a particular garment is designed to fit. Most catalogs METHODS use similar key dimensions, including bust, Subjects waist, and hip circumferences, for all garment types. Different key dimensions, how- A group of 103 women in Madison, Wiscon- ever, may be more appropriate for clothing sin, took part in this study during September and styles that cover different parts of the and October 1991. Subjects were recruited body. Little attention has been given to how through advertisements and voluntary sign- accurately consumers measure key dimen- up. They were paid a small fee for participat- sions or to whether more-accurately mea- ing. Their ages ranged from 19 to 50 years (M sured dimensions could be substituted. Little = 24.7 years, SD = 7.5 years). Subjects were is known about the magnitude of errors that divided into two categories---experienced and untrained consumers make when they mea- nonexperienced-according to their reported sure specific body dimensions using a con- experience in designing or constructing cloth- ventional tape measure. Measurement error ing. Of the 103 subjects, 36 reported that they is particularly costly for mail-order houses had made a garment at least once. because it results in returned merchandise The subjects in this study represented not and dissatisfied customers. only Wisconsin natives; many were univer- The objective of this investigation was to sity students who came from other states or determine differences in the accuracy of body foreign countries. They were, however, biased measurements made by consumers for vari- in age and educational level; most (80.5%) ous body dimensions. These data were were under age 26, and many (76.7%) were needed for selecting specific body dimensions college students. as key dimensions for an anthropometric sizing system describing women's garment size Procedure for use in retail mail-order women's clothing (Yoon, 1992). Key dimensions must satisfy Subjects made all measurements using a two criteria for suitability as a mail-order re- conventional tape measure having an English tail sizing system. First, in order to select the system scale with V16-inch(1.6-mm) preci- correct garment size, consumers must be able sion. These tape measures were used because to measure the key dimensions accurately. they are commonly available in the home and Second, a key dimension must also be a good retail stores. Measurements were made and predictor of other body dimensions that are recorded in uni ts of inches because the En- related to a particular garment type in order glish system is commonly used in the gar- to provide the framework from which size ment industry in the United States. Measure- categories will be generated. This paper de- ments were later converted into their metric scribes the procedures used for selecting key equivalents for analysis. The experimenter dimensions that satisfy the first qualification, measured all subject heights and breadths us- accuracy of body measurement. The second ing a laboratory grade anthropometer (Siber criterion is treated in Yoon (1992). This study Hegner, #101) with I-mm precision. Circum- investigated the hypotheses that (1) consum- ferences and lengths were measured using a ers measure different body dimensions with I-mm-precision tape measure (Seritex, Inc.). CONSUMER-MADE BODY MEASUREMENTS September 1994--559

All measurements were made over a tightly written instructions without obtaining addi- fitted bodysuit or over underwear. tional explanation from the experimenter. Instruments developed for this experiment Separate illustrations of female body forms included instructions for making specific were provided for describing each measure- measurements, data sheets, a set of illustra- ment. The illustrations included front, back, tions of the female body form, and a set of and side views of a woman's body. Figure 1 is demographic questions. The instruction sheet a composite of all these dimensions. explained the measurement procedures and Subjects first measured a dress mannequin provided examples. So as not to influence (mannequin measurement) for 6 dimensions. performance for a particular measurement, They then measured their own bodies (self- examples in the instructions used body di- measurement) for 19 dimensions. Afterward, mensions that were not included in the study randomly paired subjects measured each (e.g., head circumference and ankle height). other for the same 19 dimensions (partner These instructions were no more detailed measurement). They then completed a demo- than the instructions typically provided in graphic questionnaire, including an estima- mail-order catalogs. Data sheets included tion of their own stature. After subjects com- specific instructions for measuring each di- pleted these procedures, the experimenter mension. In order to avoid confusion about measured each subject for the same body di- measurement units, the word inches was mensions (standard measurement). written at the end of each line. The order in The dress mannequin was a typical body which body dimensions appeared on the data form used by dressmakers for constructing sheets was counterbalanced among subjects clothing. The mannequin measurement was in order to eliminate order effects. Ten differ- included in order to determine the amount of ent sets of data sheets were distributed, each error attributed directly to the use of the tape having different sequences of body dimen- measure by having all subjects measure the sions. same quantity, without the additional vari- A pilot study was performed involving 10 ability introduced from measuring different subjects who measured 32 body dimensions bodies. All subjects, including the experi- according to the procedures described. Based menter, measured the same mannequin. Part- on the time needed to complete the pilot ner measurements were always made after study and on the comments collected, mea- self-measurements so that the partner measurements pertaining to the back, depth, and surement did not influence the self- some circumferencial measurements of the measurement. Standard measurements were lower limb were excluded, and 19 dimensions always made after subject measurements were selected for the full study (see Figure 1). were completed, in order to keep the experi- Dimension code abbreviations are contained menter from influencing the way subjects in Table 1. performed the measurements. The experi- Although questions were answered pertain- menter measurements were performed ac- ing to experimental procedures, the experi- cording to the anthropometric procedures de- menter did not explain how to accomplish veloped by Laubach, McConville, Churchill, the measurements. Each subject studied an and White (1977). illustration of the female body form showing Experimental Design each measurement, similar to that found in a mail-order catalog, and performed the exper- The experimental design employed a full fac- iment based on her understanding of the toriallinear model, including all interactions 560-September 1994 HUMAN FACTORS

Figure 1. Anthropometric dimensions and associated codes. of fixed effects with repeated measures, perimenter. Measurement error (E) was re- where the subject variable was a random- ported in units of centimeters and also as a effect-blocking variable. Independent vari- percentage, labeled percentage error (E%). ables included dimension (19 dimensions), The absolute measurement error (lEI) and ab- method (self-measurement and partner mea- solute percentage error (IE%I) were based on surement), and experience (experienced and the magnitude of the difference between inexperienced in clothing construction). subject-made measurements and the stan- Analysis of variance (ANOVA)was used for dard measurement. testing statistically significant effects for RESULTS each dependent variable. Post-hoc pairwise contrasts (Tukey's test) were used for testing A summary of the ANOVA results is pro- significant treatment effects. vided in Table 2. The results showed that (1) Measurement error was the difference be- subjects measured different body dimensions tween measurements made by subjects and with different measurement errors, (2) part- the standard measurements made by the ex- ner measurement, on the average, had CONSUMER-MADE BODY MEASUREMENTS September 1994-561

TABLE 1

Body Dimensions Included in Study and Legend of Abbreviations

Classification Body Dimension Code Abbreviation

Height measurements Shoulder height 3C 5H Waist height 6C WH Crotch height 7C CH Breadth measurements Chest breadth 20C CB Waist breadth 21 C WB Circumference measurements Neck circumference 24C NC Shoulder circumference 25C 5C Chest circumference at scye 26C CC Bust circumference 27C BC Waist circumference 29C WC Hip circumference 30C HC Upper thigh circumference 37C TC Knee circumference 38C KC Length measurements Neck-to-bust point length 49C NL Interscye, front 43C IL Shoulder length 41 C 5L Crotch length 53C CL Sleeve inseam length 51 C 51 Sleeve outseam length 52C 50

Note. Codes are as defined in Churchill et al. (1977). smaller errors than did self-measurement, Measurement Error for Different Dimensions and (3) significant interactions between dimension and measurement method indicate Mean measurement errors and standard that some dimensions were measured with deviations for each body dimension are plot- significantly smaller errors by a partner than ted in Figure 2. Significant (p < 0.05) multi- were other dimensions. Details of these find- ple contrasts using Tukey's test are indicated ings will be presented in the following sec- at the bottom of each plot. Four dimensions tions. (HC, CL, SC, and NL) were undermeasured

TABLE 2

Summary of Analysis of Variance for Measurement Error of 19 Dimensions (N = 103) E lEI %E I%EI

Source d.f·1 d.f·2 F P F P F p F P Between-subjects: Experience (Ex) 101 0.17 0.681 1.74 0.190 3.75 0.056 3.56 0.062 Within subjects: Measurer (M) 1 101 7.29 0.008' 6.10 0.015' 14.37 0.000' 7.44 0.008' Ex x M 1 101 1.66 0.201 1.07 0.303 0.61 0.437 0.020 0.888 Dimension (D) 18 1818 24.90 0.000' 9.50 0.000' 28.52 0.000' 21.97 0.000' Ex x D 18 1818 1.05 0.394 .094 0.999 1.77 0.024' 1.88 0.016' MxD 18 1818 6.26 0.000' 4.54 0.000' 6.67 0.000' 4.83 0.000' Ex x M x D 18 1818 0.82 0.679 0.63 0.875 0.62 0.885 0.40 0.988

Notes: d.I., ~ numerator degrees of freedom; d.f'2 ~ denominator degree of freedom; 'p < 0.05 562-September 1994 HUMAN FACTORS

more than were the other body dimensions. had smaller lEIfor self-measurement than for Three dimensions (CB, IL, and WB) were partner measurement; however, the magni- overmeasured more than were the other di- tude of this difference was small (0.44 cm ~ mensions. Two dimensions (IL and WB) were lEI ~ 0.72 cm). overmeasured by an average error of 6 cm or Average errors for self-measurement, part- more. The dimensions having lEI > 5 cm in- ner measurement, and mannequin measure- cluded SC, HC, CL, IL, and WB. In terms of ment are plotted in Figure 4. Tukey's pair- percentages, four dimensions (CB, SL, IL, and wise contrasts showed that waist breadth WB) were overmeasured more than were the (WB), front interscye length (IL), and chest other overmeasured dimensions, ranging breadth (CB) were measured with signifi- from 11.9% to 26.0%. Waist breadth (WB) had cantly greater error (5.22 cm < lEI< 6.60 cm); the greatest IE%I. waist circumference (WC) and bust circum- Dimensions having the greatest measure- ference (BC)were measured with the smallest ment variation among subjects (SD > 8.00 error (lEI = 1.90 cm); p < 0.05. Waist breadth cm) included shoulder circumference (SC), (WB) was more accurately measured by part- waist breadth (WE), sleeve outseam (SO), up- ner measurement (lEI = 3.25 cm) and man- per thigh circumference (TC), and chest nequin measurement (lEI = 4.12 cm) than by breadth (CB). Dimensions having the small- self-measurement (lEI = 9.62 cm). Waist cir- est variation among subjects (SD < 4.31 em) cumference (WC) and bust circumference were knee circumference (KC), neck circum- (BC) were more accurately measured on a ference (NC), waist height (WH), neck to bust mannequin (2.24 cm < lEI< 2.57 cm) than on point length (NL), and bust circumference human subjects (0.77 cm < lEI < 0.90 cm). (BC). Dimensions having the greatest mean Self-Estimation of Stature relative error (E% > 17%) and variation among subjects (SD > 16.3%) were WB, SL, Subjects estimated their own height with IL, and CB (see Figure 2). an error ranging from - 30.0 cm to + 14.8 cm. The average lEIwas 2.26 cm (SD = 3.39 cm). Measurement Errors for Self-Measurement Average E was 0.68 cm (SD = 4.02 cm). More and Partner Measurement than 64% of the subjects overestimated their Overall, partner-measurement error 'Was stature. Almost half (48.5%) of the subjects significantly smaller than self-measurement overestimated their stature by more than 1 error for all error variables (see Table 2). Av- cm, whereas more than a third (35.9%) un- erage lEIwas 4.10 cm (SD = 6.06 cm) for self- derestimated their stature by more than measurement and 3.34 cm (SD = 4.86 cm) for 1 cm. partner measurement. Mean E for self- Precision of Subject Measurements measurement and partner measurement are plotted against 19 body dimensions in Fig- Subjects performed all measurements us- ure 3. ing a conventional tape measure having an The interaction between dimension and English system scale with V16-inch(l.6-mm) measurer was also significant for all error precision. No subjects, however, actually re- variables (see Table 2 and Figure 3). Eight ported their measurements with 1.6-mm pre- dimensions (HC, CL, BC, NC, SH, CB, IL, and cision. Only 3 subjects (3%) made their mea- WB) had smaller lEIfor partner measurement surements using VB-inch(3.2-mm) precision; than for self-measurement (0.55 cm ~ lEI ~ 46 subjects (44.6%) used 1/4-inch (6.4-mm) 6.37 cm). Three dimensions (TC, KC, and WH) precision; 48 subjects (46.6%) used V2-inch CONSUMER-MADE BODY MEASUREMENTS September 1994-563

20 MEASUREMENT ERROR (em)

10 .eE oII: II: o II: W -10

-20 HC CL 5C NL CC 50 BC 5H WC NC WH TC KC 51 SL CH CB IL WB TUKEYTEST:

ABSOLUTE MEASUREMENT ERROR (em) i:1 w 0 I :.~! ~, ; ~. ~~ ~ NC KC WH BC WC 5L NL CC 50 5H 51 TC CH CB 5C HC CL IL WB TUKEYTEST:

90 MEASUREMENTERROR(%) 60 ~ ~ 30

i 0 !-=-1- -I -1- -i I ~ ~ ~ I 0--1-H ------J -1 -30

-60 NL HC CL 5C CC SO BC 5H WC NC WH TC 51 CH KC CB 5L IL WB TUKEYTEST:

90 ~ ABSOLUTE MEASUREMENT ERROR (%) >!!. ~ 60 oII: II: 30 II: W o 5H WH BC WC CC 50 NC 5C KC CH HC TC 51 CL NL CB IL 5L WB TUKEYTEST: DIMENSION CODES Figure 2. Mean error for 19 body measurements (N = 103). Dimensions are ordered along the x-axis in ascending order of error. 564-September 1994 HUMAN FACTORS

10 MEASUREMENTERROR~m) 8 ----<>- PARTNER (em) 6 E ---e--- SELF (em) 8- 4 II: 0 II: 2 II: W

-2

-4

-6 HC CL SC NL CC SO BC WC SH WH NC TC KC SI SL CH CB IL WB 10 /" ~ 8 ABSOLUTE MEASUREMENT ERROR (em) E ,.' 8- 6 -.---." II: 0 II: 4 II: W 2

0 NC KC WH WC BC SL NL so CC TC SI CH SH SC HC CB CL IL WB 50 40 MEASUREMENTERROR(%) ...' 30 .· ~ ..· II: 0 20 •• II: II: W 10

0 ---~

-10 NL CL HC SC CC WC 50 BC 5H WH NC TC KC 5 I CH 5L CB I L WB

50 ABSOLUTE MEASUREMENT ERROR (%) , 40 ... ~ ,. ~ 30 :· II: I o ,.---. II: 20 " II: W 10

o WH BC SH WC CC NC KC SC SO CH HC TC 51 CL NL SL CB IL WB DIMENSION CODES Figure 3. Mean error for interaction between dimension and measurer (N = 103). Dimensions are ordered along the x-axis in ascending order of error for the self-measurement condition. CONSUMER-MADE BODY MEASUREMENTS September 1994-565

MEASUREMENT ERROR (em)

8 ~ 6 ~ oII: 4 ~ 2 w o

-2 BC WC SL CB IL WB

10 ABSOLUTE MEASUREMENT ERROR (em)

~ 6 II: ~ 4 II: w 2

If--"------oM------'" ..--~' o WC BC SL CB IL WB

50 MEASUREMENTERROR(%) 40 t 30 II: 20 0 II: II: 10 w 0

-10 WC BC SL CB IL WB

50 ABSOLUTE MEASUREMENT ERROR (%) 40

30 oa: a: a: 20 w 10

o BC we SL CB IL WB

DIMENSION CODES Figure 4_ Mean error for self-measurement. partner measurement, and mannequin measurements. Dimensions are ordered along the x-axis in ascending order of error for the self-measurement condition. 566---September 1994 HUMAN FACTORS

(12.7-mm) precISIon; and 6 subjects (5.8%) scye length (IL) were overmeasured with the made their measurements using no more greatest errors (EwB = + 6.15 cm and ElL = than I-inch (25.4-mm) precision. + 6.01 cm; see Figure 2). These results indicate that a conventional tape measure is not DISCUSSION appropriate for breadth measurements. An This study measured the precision and ac- alternative measuring tool, such as a caliper, curacy of body measurements made by con- is needed if these dimensions are used for siz- sumers for various body dimensions in order ing. Further investigation into alternative to determine appropriate key dimensions for measurement methods and instruments for mail-order garment size descriptions. The ac- consumers is therefore needed. curacy of body measurements varied widely Shoulder height (SH) and waist height for different body dimensions. The results in- (WH) were measured with the smallest lEI. lEI dicated that two key dimensions-bust cir- of waist height (WH) also had the smallest cumference (BC) and waist circumference standard deviation of all the dimensions mea- (WC}-frequently used for mail-order gar- sured. These results imply that waist height ment sizing, were measured with relatively may be a good key dimension for garments small error. Waist height (WH) and shoulder covering the lower body and that shoulder height (SH) were measured by consumers height may be useful for garments covering with similar accuracy. The results also the neck base or shoulder to the lower body. showed that hip circumference (HC) was sig- Yoon (1992) classified garments into nificantly undermeasured by consumers, groups of related body dimensions. Key di- even though that dimension is often used as a mension candidates were then selected based key dimension in mail-order catalogs. on correlation among body dimensions using Waist circumference (We) and bust cir- residual variance analysis for data from the cumference (BC) were measured with small- 1977 U.S. Army anthropometry data base est absolute error (see Figure 2). Neck circum- (Churchill, Churchill, McConville, and White, ference (NC), a measurement often used for 1977). The selection of these key dimensions men's dress shirts, was measured with a was confirmed using factor analysis. Bust cir- small average error and standard deviation. cumference (BC) or scye chest circumference Therefore, it was concluded that these three (CC) was selected as a potential key dimen- dimensions satisfy the first qualification for sion for all upper body garments. Sleeve in- key size dimensions: accuracy of body mea- seam (SI) or sleeve outseam (SO) was selected surement. The results imply that waist cir- for long-sleeve garments, and shoulder height cumference and bust circumference may be (SH) was selected as the vertical key dimen- good key dimensions for garments covering sion for short-sleeve or no-sleeve upper body those body parts. Neck circumference may be garments. Hip circumference (HC), waist cir- a suitable key dimension for a garment de- cumference (WC),hip breadth, waist breadth signed to fit around the neck (e.g., a man's (WB), or back curvature at the waist was se- dress shirt). lected as the horizontal key dimension for Hip circumference (He), which is often uti- pants. Hip circumference (HC) or waist cir- lized in the garment industry, was undermea- cumference (WC) was selected as the key di- sured (E = -4.54 cm). Further studies are mension for skirts. Crotch height (CH) or needed to identify the actual reason for the crotch length (CL)was selected as the vertical large measurement errors observed for hip key dimension for pants, and waist height circumference. Waist breadth (WB) and inter- (WH), buttock height, or knee-cap height was CONSUMER-MADE BODY MEASUREMENTS September 1994-567

selected for skirts. To be suitable in an an- variation, experimenter experience and train- thropometric size designation, it is necessary ing, instrument resolution, and accuracy. For that a selected key dimension be a good pre- instance, measurement of some body dimen- dictor of other body dimensions and that con- sions changes with body posture (e.g., waist sumers be able to measure key dimensions circumference), breathing cycle (e.g., chest accurately. circumference), or muscle tension (e.g., bi- The combined results of the current study ceps circumference; Roebuck, Kroemer, and and results by Yoon (1992) indicate that bust Thomson, 1975). Other errors occur because circumference (BC) and waist circumference of difficulties in finding body landmarks. (WC) are suitable key dimensions because Subject-made measurements were accom- they were accurately measured by consumers plished using a conventional tape measure. and they predict other body dimensions re- Error introduced by this type of instrument lated to garments covering the upper body. should be considered in establishing size tol- For the same reason, waist height (WH) is rec- erances in an anthropometric size system. ommended as a key dimension for garments A portion of measurement error in this covering the lower body. study was introduced by the difference in pre- Subjects measured their partners' bodies cision used by the subjects and the experi- more accurately than they measured them- menter. However, this error was consistent selves for the waist breadth (WB; average among all measurements. Measurement error difference = 6.37 cm) and chest breadth less than 0.64 cm (1/4inch) may have been (CB; average difference = 2.96 cm) dimen- caused by reading precision selection rather sions (see Figure 3). Therefore, if these di- than by measurement accuracy (see Figure 4). mensions are used as key dimensions, it is Another source of measurement error con- recommended that they be measured by cerns the actual reading of the measurement someone else. instruments. Although subjects performed When subjects estimated rather than mea- measurements using a conventional tape sured their own body sizes, they overesti- measure having an English system scale with mated stature by 0.68 ± 4.02 cm. Stature was 1/16-inch(1.6-mm) precision, no subjects chose used because it is a body dimension com- to report measurements with that degree of monly estimated in daily life, such as when precision. Only 3% of the subjects used l/S_ applying for a driver's license or for other inch (3.2-mm) precision, whereas 97% of the personal identification. These observations subjects used 1/4-inch(6.4 mm) or more. are supported by the results of a previous CONCLUSIONS study (Buckle, 1985). The results indicate that consumer-estimated body sizes are not very This study investigated differences in the reliable and therefore should be measured. measurement of specific body dimensions Methods often used for published anthro- made by consumers. The results of these ex- pometric surveys include either direct mea- periments lead to the following conclusions. surement (Gallwey and Fittzgibbon, 1991; Garrett, 1971; Molenbroek, 1987) or indirect 1. Measurement error varied, depending on body dimension. The dimensions measured with photographic methods (Frobin, Hierholzer, less than 2.5 cm absolute error included shoul- and Drerup, 1982; Lovesey, 1974; Mollard, der and waist heights and neck, knee, waist, Sauvignon, and Pineau, 1982; Sheldon, 1970). and bust circumferences. Hip and shoulder circumferences were undermeasured by -4.54 Measurement errors are typically made be- cm and - 3.06 cm, respectively. Waist breadth, cause of the complicated effects of subject chest breadth. and front interscye length were 568-September 1994 HUMAN FACTORS

overmeasured by 6.15 cm, 3.19 cm, and 6.01 2. The basic univariate statistics (Tech. Report NATICK! cm, respectively. TR-TII024). Natick, MA: U.S. Army Natick Research 2. Self-measurement was generally less accurate and Development Command. for waist and chest breadths and for shoulder Delk, A. E., and Cassill, N. L. (1989). Jeans sizing: Prob- lems and recommendations. Apparel Manufacturer, 1, height (ranging from 2.27 cm to 6.32 cm abso- 18-23. lute error). Hip circumference, crotch length, Fellingham, C. (1991, August). Whose body is this for, any- chest breadth, front interscye length, and way? The secrets of a perfect fit. Glamour, pp. 159-160. waist breadth were more accurately measured Frobin, W., Hierholzer, E., and Drerup, B. (1982). Mathe- by partners (ranging from 6.37 cm to 2.15 cm matical representation and shape analysis of irregular absolute error). These results indicated that body surfaces. Biostereometrics, '82,361, 132-139. partner measurement was more accurate than Gallwey, T. J., and Fittzgibbon, M. J. (1991). Some anthro- self-measurement for breadth measurements pometric measures on an Irish population. Applied Er- gonomics, 22, 9-12. or for measurements related to the waist. Garrett, 1. W. (1971). The adult human hand: Some an- These measurements should be measured by thropometric and biomechanical considerations. Hu- someone else if they are used for key dimen- man Factors, 13, 117-131. sions. LaBat, K. L. (1987). Consumer satisfaction/dissatisfaction 3. Estimation of body sizes is not reliable. Stat- with the fit of ready-to-wear clothing. Unpublished doc- ure, one of the most commonly referenced toral dissertation, University of Minnesota, St. Paul. body dimensions in daily life, was overesti- MN. mated by 64.1% ofthe subjects. More than 50% Laubach, L. L., McConville, J. T., Churchill, E., and White, R. M. (1977). Anthropometry of women of the U.S. Army- of the subjects overestimated their height by 1977, report no. 1. Methodology and survey plan (Tech. 1.1 to 14.8 cm. Report NATICKlTR-77/021). Natick, MA: U.S. Army 4. These studies indicated that waist height, Natick Research and Development Command. waist circumference, bust circumference, and Lovesey, E. J. (1974). The development of a 3-dimensional shoulder height are appropriate as key dimen- anthropometric measuring technique. Applied Ergo- sions because they are accurately measured. nomics, 5, 36-41. Hip circumference may also be used as a key Molenbroek, J. F. M. (1987). Anthropometry of elderly peo- dimension if it is measured by someone other ple in the Netherlands: Research and applications. Ap- plied Ergonomics, 18, 187-199. than the consumer. Mollard, R., Sauvignon, M., and Pineau, J.C. (1982). Bio- stereometric study of a sample of 50 young adults by ACKNOWLEDGMENTS photogrammetry. Biostereometrics, '82,361,234-240. Roebuck, J. J. A., Kraemer, K. H. E., and Thomson, W. G. The authors would like to thank Dr. Cynthia R. Jasper (1975). Engineering anthropometry methods. New York: for her helpful comments concerning this research. Sup- Wiley. port for this research has been provided by the University Sheldon, W. (1970). The varieties of human physique. of Wisconsin-Madison, School of Family Resources and Darien, CT: Hafner. Consumer Science. Smallwood, V., and Wiener, J. (1987). Light and heavy REFERENCES catalog shoppers of clothing. Clothing and Textile Re- search Journal, 5(3), 25-30. Belkin, L. (1986, March I). Taking measure of Americans: Yoon, Y. C. (1992). A methodology for devising an anthropo- Size guides get updating. New York Times, Section I, p. metric size description system for women's apparel. Un- 33. published doctoral dissertation, University of Wiscon- Buckle, P. W. (1985). Self-reported anthropometry. Ergo- sin-Madison. nomics, 28, 1575-1577. Churchill, E., Churchill, T., McConville, J., and White, M. Date received: November 28,1992 (1977). Anthropometry of the U.S. Army-1977, report no. Date accepted: July 22, 1993 H U MA N FACTO RS, 1994,36(3),569-572

The Authors

DANIEL GOPHER, MAYA WElL. and TAL lationship between human cognitive performance BAREKET (Transfer of Skill from a Computer Game and operability in submarine command and con- Trainer to Flight) trol. She received her Ph.D. in experimental psy' DANIEL GOPHER is a professor of human fac- chology from the University of Rhode Island in tors engineering in the Faculty of Industrial Engi- 1985. Currently she is the principal investigator on neering and Management at the Technion-Israel In- a program to improve decision aids by making stitute of Technology and the director of the them better meet the cognitive needs of expert Research Center for Work Safety and Human En- decision makers. Other research interests include gineering. He received his Ph.D. in experimental expertise in the decision.making process, human- psychology in 1972 from the Hebrew University of computer interactions, and the effects of informa- Jerusalem in Israel. From 1970 to 1980, he was a tion presentation variables on human memory and senior scientist at the Human Factors Branch of the learning. She has served as visiting scientist in the Israeli Air Force and the head of this branch from Human·Computer-Interaction Lab at the Naval 1975 to 1980. He held visiting positions in aca- Research Laboratory in Washington, D.C. and in demic and research institutes in the United States the Cognitive and Neural Sciences Division at the and Europe and participated in several interna- Office of Naval Research in Arlington, Virginia. tional research programs. His research work and JAMES E. ARRUDA is a Ph.D. candidate in ex- scientific publications are on the topics of attention perimental psychology at the University of Rhode processes, mental workload, training of complex Island. His research interests include physiological skills, human error, safety at work, and human fac- indicators of attention, cognitive and neural mod- tors in aviation systems. els of attention, and the relationship between neu- MAYAWElL is a research fellow in the Research rophysiology and behavior. Center for Work Safety and Human Engineering at the Technion-Israel Institute of Technology. She re- DOUGLAS J. GILLAN and ROBERT LEWIS (A ceived a B.Sc. in mathematics and physics in 1965 Componential Model Human Interaction with from the University of Minnesota and completed of Graphs: J. Linear Regression Modeling) her studies for a master's degree in statistics in DOUGLASJ. GILLAN received a B.A. in psychol- 1967. In 1978, she received an M.Sc. in manage· ogy from Macalester College and a Ph.D. in psy- ment sciences from the Faculty of Industrial Engi- chology from the University of Texas at Austin in neering and Management at the Technion-Israel In- 1978. He has been an NSF Postdoctoral Fellow at stitute of Technology. Her research work and Yale University and a Sloan Foundation Cognitive publications focus on problems of safety at work and training of complex skills. Science Fellow at the University of Pennsylvania. He has worked at the General Foods Technical Cen- TAL BAREKET is a captain at the Human Fac- ter, Lockheed Engineering and Sciences Company, tors Branch of the Israeli Navy. He received his Rice University (as a visiting faculty member), the B.A. in psychology from Tel-Aviv University in University of Idaho, and New Mexico State Univer- 1987 and an M.Sc. in behavioral sciences in 1991 sity. His research has focused on perception and from the Faculty of Industrial Engineering and cognition and their application to human-computer Management at the Technion-Israel Institute of interaction. Technology. His present work involves application ROBERT LEWIS received a B.A. in psychology and research of human factors in naval systems. from Claremont McKenna College and a Ph.D. in psychology from Rice University in 1989. His re- SUSAN S. KIRSCHENBAUM and JAMES E. AR- search interests are in industrial psychology with a RUDA (Effects of Graphic and Verbal Probability In- focus on work design. formation on Command Decision Making) SUSAN S. KIRSCHENBAUM is an engineering psychologist in the Combat Systems Department of DAVIDG. PAYNE, LESLIE 1. PETERS, DEBO- the Naval Undersea Warfare Center Division, New- RAH P. BIRKMIRE, MARIVIC A. BONTO, JEF- port, Rhode Island, where she investigates the re- FREY S. ANASTASI, and MICHAEL J. WENGER 570-September 1994 HUMAN FACTORS

(Effects of Speech Intelligibility Level on Concurrent human factors research with IBM Systems Integra- Visual Task Performance) tion Division and Thomson Consumer Electronics. DAVID G. PAYNE received a B.S. and an M.S. in His research interests involve human memory, in- experimental psychology from State University of cluding eyewitness memory and hypermnesia as New York College at Cortland and a Ph.D. in cog- well as human-computer interaction. nitive psychology from Purdue University. He is an MICHAEL J. WENGER holds a B.S. (cum laude) associate professor in the Department of Psychol- in psychology and an M.S. in engineering (techni- ogy at the State University of New York at Bing- cal communication), both from the University of hamton. He has held research fellowships at NASA Washington. He has worked as a human factors en- Langley Research Center and Wright-Patterson Air gineer for mM Corporation and is currently pursu- Force Base and has conducted human factors re- ing a Ph.D. in experimental psychology at the State search with CAE-Link Corporation, GTE Laborato- University of New York at Binghamton. His re- ries, IBM Systems Integration Division, and Thom- search interests are cognitive skill and the use of son Consumer Electronics, He is also a member of memory in skilled performance, memory processes the Center for Cognitive and Psycholinguistic Sci- in reading comprehension, and human-computer ences at the State University of New York at Bing- in teraction. hamton. His research interests are in the areas of human-computer interaction; development of skill, JOANNE M. WOOD and ROD TROUTBECK (Ef- memory, attention, and auditory-visual interac- fect of Visual Impairment on Driving) tions; and assessment of performance deficits. JOANNE M. WOOD received a B.Sc. (Hons) de- LESLIE J. PETERS received a B.A. in psychology gree in optometry from the University of Aston in and an M.A. in audiology in 1970 from the Univer- Birmingham in 1982 and for two years worked as sity of Florida and a Ph.D. in industrial audiology an optometrist in private practice. She received a from the Pennsylvania State University in 1986. Ph.D. in vision science from the University of Aston His research interests are in the assessment of au- in Birmingham in 1987 and worked for a year as a ditory workload and central auditory processing research fellow in the Visual Science Unit at Ox- phenomena. ford. Since 1989, she has been a tenured faculty DEBORAH P. BIRKMIRE is a research psychol- member of the Centre for Eye Research at the ogist in the U.S. Army Human Research Labora- Queensland University of Technology in Brisbane, tory, Human Engineering Directorate, Aberdeen Australia. Her research interests are concerned Proving Ground, Maryland. She received a B.A. with the importance of vision and visual impair- from Loyola College in 1970 and an M.A. and a ment in driving performance, particularly with Ph.D. from the University of Delaware in 1979 and respect to the elderly, and with the functional role 1982. Her research interests include visual search of peripheral vision. and target acquisition, effects of visual-auditory in- ROD TROUTBECK received a B.E. (Civil) and an teractions on performance, and the structure and M.Eng.Sc. from the University of Melbourne. He function of procedural text. then worked for a road authority for 2 years before MARIVIC A. BONTO received an A.B. in psychol- joining the Australian Road Research Board where ogy in 1981 from the Ateneo de Manila University he worked for 15 years. He obtained his Ph.D in and an M.A. in experimental psychology from the transportation from the University of Queensland State University of New York at Binghamton in in 1989. Since 1989 he has been a tenured faculty 1990. She has held human factors internships with member of the Physical Infrastructure Centre at mM Corporation in North Carolina and New York. the Queensland University of Technology. His re- She is currently pursuing her Ph.D. in experimen- search interests include the performance of drivers tal psychology at the State University of New York on urban roads, in particular their ability to nego- at Binghamton. Her research interests are in tiate intersections. human-computer interaction and the effects of speech intelligibility on performance in multi task WILLIAM R. UTTAL, TODD BARUCH, and environments. LINDA ALLEN (Psychophysical Foundations of a JEFFREY S. ANASTASI holds a B.A. in psychol- Model of Amplified Night Vision in Target Detection ogy from the State University of New York at Bing- Tasks) hamton. He is pursuing a master's degree in exper- WILLIAM R. UITAL is professor of industrial imental psychology at the State University of New and management systems engineering and of com- York at Binghamton. He has been involved with puter science and engineering. He is principal in- THE AUTHORS September 1994-571

vestigator at the Perception Laboratory at Arizona an M.A.and Ph.D. in experimental psychology from State University. He received his Ph.D. from Ohio the University of Southern California. From 1984 to State University in 1957 and has done research at 1989 he was a senior human factors engineer at International Business Machines Corporation, the McDonnell Douglas Corporation in St. Louis. He is University of Michigan, and the Naval Ocean Sys- currently an assistant professor of psychology at tems Laboratory in Hawaii. He is the author of 14 Wright State University. His research interests are books on various topics in sensory psychology, neu- in the cognitive psychophysiology of attention and rophysiology, and computer science, as well as nu- performance, particularly central and autonomic merous articles and chapters. nervous system interaction in stressful environ- TODD BARUCH received his B.S. in psychology ments. in 1992. He also conducted experimental research ARTHUR M. RYAN completed his undergradu- in visual perception at Armstrong Laboratory, ate work at Wright State University, where he re- Williams Air Force Base, Arizona, from 1987 until ceived a B.S. degree in psychology with a human 1992, when he joined the Perception Laboratory at factors concentration. He is currently a student in Arizona State University. the human factors graduate program in the Depart- LINDA ALLEN is a graduate student in the De- ment of Psychology at Wright State. His primary partment of Industrial and Management Systems research interest concerns psychophysiological Engineering, Arizona State University. She re- measures of central and autonomic nervous system ceived her B.S. in mathematics from Arizona State functioning during performance in dynamic envi- University in 1978. ronments. GLENN F. WILSON received his Ph.D. in psy- PAMELAS. TSANG and MICHAELA. VIDULICH chology from the University of Arizona. He is a re- (The Roles of Immediacy and Redundancy in Relative search psychologist at the Performance Assessment Subjective Workload Assessment) Branch of the Armstrong Laboratory at Wright- PAMELA S. TSANG is an associate professor at Patterson Air Force Base, Ohio. His research inter- Wright State University. Previously she was a Na- ests include cognitive workload, in-flight workload tional Research Council postdoctoral research as- evaluation, and evaluation methodology. sociate at NASA-Ames Research Center and a research associate at the Institute of Aviation, J. PAUL FRANTZ (Effect of Location and Proce- University of Illinois. She received her A.B. from dural Explicitness on User Processing of and Compli- Mount Holyoke College and her M.A. and Ph.D. de- ance with Product Warnings) received a B.S.E. in grees in engineering psychology from the Univer- human factors engineering from Wright State Uni- sity of Illinois at Urbana-Champaign. Her research versity and a Ph.D. in industrial and operations en- interests are in time-sharing performance, mental gineering from the University of Michigan in 1992. workload, aviation psychology, and aging. He is currently a senior research engineer with MICHAEL A. VIDULICH received his M.A. in ex- Miller Engineering, Inc., in Ann Arbor, Michigan. perimental psychology from Ohio State University His research interests and areas of specialization in 1979 and his Ph.D. in engineering psychology include human factors applied to consumer and in- from the University of Illinois in 1983. From 1983 dustrial products, product safety, and occupa- to 1987 he was a member of the Human Workload tionallhome accident prevention with a particular and Performance Group at NASA-Ames Research emphasis on safety-related communications such Center. Currently he works at the Performance As- as warnings, instructions, and other product- sessment and Interface Technology Branch of the accompanying information. Human Engineering Division of the U.S. Air Force's Armstrong Laboratory. His main interests are in MICHAEL S. WOGALTER, STEPHEN W. JAR- the role of attention in human performance and de- RARD, and S. NOEL SIMPSON (Influence of Warn- veloping workload and situation awareness metrics. ing Label Signal Words on Perceived Hazard Level) MICHAEL S. WOGALTER received a B.A. from RICHARD W. BACKS, ARTHUR M. RYAN, and the University of Virginia, an M.A. from the Uni- GLENN F. WILSON (Psychophysiological Measures versity of South Florida, and a Ph.D. from Rice Uni- of Workload during Continuous Manual Perfor- versity. Before joining the psychology faculty at mance) North Carolina State University, he held appoint- RICHARD W. BACKS received a B.A. in psychol- ments at the University of Richmond and Rensse- ogy from the University of Missouri-St. Louis and laer Polytechnic Institute. His research interests in- 572-September 1994 HUMAN FACTORS

clude safety, warnings and instructions, hazard JONGSUK C. YOON has B.S. and M.S. degrees in and risk perception, human-computer interaction, clothing and textile science from Yonsei University and human cognition. in Seoul. South Korea, where she is now an assis- STEPHEN W. JARRARD is a major in the U.S. tant professor of clothing and textiles. She received Army and is currently an assistant professor at the a Ph.D. in psychology from the University of U.S. Military Academy at West Point. He has 12 Wisconsin-Madison in 1992. Her research interests years of experience in Army units in the United include international sizing systems for women's States and overseas and has worked in various ready-to-wear garments. command and staff positions from platoon to divi- ROBERT G. RADWINis an associate professor at sion level. He received a B.S. degree from the U.S. the University of Wisconsin-Madison in the Depart- Military Academy in 1982 and an M.S. degree in ment of Industrial Engineering, where he conducts human factors from Rensselaer Polytechnic Insti- research and teaches in the areas of ergonomics tute in 1992. His research interests include the and human factors engineering. He has a B.S. in study of learning techniques to improve recogni- electrical engineering from the Polytechnic Insti- tion memory, warning applications, and human tute of New York and M.S. degrees in electrical cognition. engineering and bioengineering from the Univer- S. NOEL SIMPSON is currently enrolled in the sity of Michigan. He earned his Ph.D. in industrial human factors graduate program at Rensselaer and operations engineering at the University of Polytechnic Institute in Troy, New York. She re- Michigan and was a research fellow at the Center ceived a B.A. in psychology from the University of for Ergonomics. He is actively developing elec- Toledo. Her research interests include human factors tronic instruments and analytical methods for in transportation, warnings, and decision science. measuring human performance and physical stress in the workplace. He is also establishing ergonomic JONGSUK C. YOON and ROBERT G. RADWIN guidelines for hand-held equipment and tools, and (The Accuracy of Consumer-Made Body Measure- he is investigating deficits associated with physical ments for Women's Mail-Order Clothing) stress and cumulative trauma disorders.