International Journal of Obesity (1999) 23, 904±908
ß 1999 Stockton Press All rights reserved 0307±0565/99 $12.00
http://www.stockton-press.co.uk/ijo
Comparison of measured and self-reported
weight and height in a cross-sectional sample of
young adolescents
RS Strauss1*
1
Division of Pediatric Gastroenterology and Nutrition, UMDNJ - Robert Wood Johnson School of Medicine, New Brunswick, NJ, USA
AIM: To explore the relationship between self-reported weight and height to actual weight and height in a crosssectional nationally representative sample of young adolescents.
METHODS: Weights and heights were obtained on 1932 adolescents aged 12 ± 16 y enrolled in the NHANES III study.
Self-reported weights and heights were available on 1657 of the adolescents (86%).
RESULTS: Correlation between self-reported weight and actual weight ranged between 0.87 and 0.94, depending on
gender or race. However, self-reported weights were signi®cantly lower than measured weights among girls,
compared to boys (P < 0.001). Correlation between self-reported height and actual height ranged from 0.82 ± 0.91.
There were no differences in the accuracy of self-reported heights among boys and girls or racial groups. Differences
between actual weight and self-reported weight were signi®cantly greater for obese children compared with nonobese children (P < 0.001). Nevertheless, the use of self-reported weight and height resulted in the correct classi®cation of weight status in 94% of children. As a result, small differences in self-reported weights and heights had no
impact in assessing obesity related morbidities.
CONCLUSION: In¯uences of gender and racial biases in reporting of weight and height were relatively small. Selfreported heights and weights were extremely reliable for the predicting obesity related morbidities and behaviours.
Keywords: body weight; body mass index; obese; adolescents
Introduction
In national surveys of youth behaviour, self-reported
weights and heights are often utilized in place of
actual measurement. In the United States, the Youth
Risk and Behavior Surveillance Study (YRBSS),1 the
National Longitudinal Study of Adolescent Health,2
and the National Health Interview Surveys (NHIS)3
rely on self-reported weight and height in order to
assess the impact of body weight on adolescent selfesteem, dieting, and overall health.
In adults, the correlation between actual and
reported heights and weights typically range between
0.96 and 0.99.3 ± 5 Brooks-Gunn et al reported correlations between actual and reported heights and weights
ranging between 0.77 and 0.98 in a small group of
young adolescents from a well-educated, high-income
community.6 However, self-reports may re¯ect cultural biases resulting in differing accuracies among
sexes and races. The correlation between actual and
reported heights and weights in a cross-sectional
national survey of children has not been studied.
*Correspondence Richard S. Strauss, Division of Pediatric
Gastroenterology & Nutrition, UMDMJ - Robert Wood Johnson
Medical School, One Robert Wood Johnson Place, CN-19 New
Brunswick, NJ 08903, USA.
E-mail: [email protected]
Received 18 September 1998; revised 5 March 1999; accepted
24 March 1999
In order to clarify the accuracy of self-reported
weight and height, data were analyzed from the
National Health and Nutrition Examination Survey
(cycle III), a nationally representative sample of
children and adults between 1988 and 1994. Data
were also analyzed to determine whether self-reported
weight and height resulted in differences between
predicted risk of obesity related morbidities compared
to actual weight and height.
Methods
The National Health and Nutritional Examination
Survey, Cycle III (NHANES III) is the seventh in a
series of large national health examination surveys
conducted in the United States since the 1960s.
NHANES III examined a nationally representative
sample of children and adults between 1988 and
1994. 1932 children between the ages of 12 and 16
were studied. Measured weights and heights were
available on over 99% of the children; self-reported
weights and heights were available on 1657 children
(86%).
Body weight was determined to the nearest 0.05 kg
(Toledo 2181 Scale) and height was measured to the
nearest 0.1 cm using standardized measuring equipment (Holtain Height Stadiometer). Equipment was
recalibrated between each measurement. Children
Validity of self-reported height and weight
RS Strauss
were weighed and measured in their undergarments.
Self-reported weights and heights were recorded to
the nearest pound and inch and converted to metric
measurements. Tanner pubertal staging was performed by trained study staff.
Normal weight was de®ned as a body mass index
(kg=m2) less than the 85th percentile for age and
gender; overweight was de®ned as a body mass
index (kg=m2) greater than the 85th percentile for
age and gender, and obesity was de®ned as a body
mass index (kg=m2) greater than the 95th percentile
for age and gender. Reference body mass index
percentiles were derived from the ®rst National
Health and Nutrition Examination Survey.7 This de®nition is in accordance with recommendations of the
Expert Committee on Clinical Guidelines for Overweight in Adolescence.8
All children underwent a standardized interview
with trained staff. Children were asked the following
questions concerning weight, height, self-perceived
weight, and weight control practices:
Ð How tall are you without shoes?
Ð How much do you weigh without clothes or
shoes?
Ð How many hours of television did you watch
yesterday?
Ð How many times each week do you exercise
enough to make yourself sweat?
Ð Have you dieted during the past year?
Ð How often do you eat breakfast?
Because the NHANES III study over-sampled blacks,
younger children, and the elderly, the data was
weighted so that all statistics re¯ected a national
representative sample of children aged 12 ± 16 years
old. Differences in continuous variables were assessed
using the independent t-test. Differences in proportions were assessed using chi-square. Correlation
coef®cients were calculated using linear regression.
Odds ratios were calculated using logistic regression.
In order to adjust for complex sample design and
clustering effects in the NHANES III sample, statistical signi®cance was assessed using the balance
repeated replication method using the software package WesVarPC (Westat Inc., Rockville, MD) as
recommended.9
Results
936 of the children enrolled were white/Hispanic
(57%), 647 were black (38%), and 74 were
other (5%) (Table 1). Approximately 16% of young
adolescents were overweight (85th percentile < BMI
< 95th percentile), and 11% were obese (BMI > 95th
percentile). In addition, there were no signi®cant
differences in age distribution between normal
weight (BMI < 85th percentile), overweight and
Table 1
905
Characteristics of study children (n ˆ 1657)
Race
White/Hispanic
Black
Other
Age
12 y
13 y
14 y
15 y
16 y
Gender
Male
Female
Overweight (BMI > 85th %)
Obese (BMI > 95th %)
57% (n ˆ 936)
39% (n ˆ 647)
5% (n ˆ 74)
15%
20%
21%
21%
24%
(n ˆ 248)
(n ˆ 323)
(n ˆ 347)
(n ˆ 345)
(n ˆ 394)
46%
54%
27%*
11%*
(n ˆ 767)
(n ˆ 890)
(n ˆ 460)
(n ˆ 188)
*Weighted to re¯ect nationally representative sample of children.
obese children. There was a small, but signi®cantly
increased prevalence of obesity among blacks compared to whites (13% vs 10%, P < 0.05), but no
difference in the prevalence of overweight among
blacks compared to whites (27% vs 25%, P ˆ 0.4).
Self-reported weight in young adolescents was
within 2 kg of actual weight in 59% of children and
within 5 kg of actual weight in 88%. Self-reported
height was within 2 cm of actual height in 57% of
children and within 4 cm of actual height in 81%. 95%
of adolescents reported a weight within 10 kg of actual
weight, and over 95% of adolescents reported a height
within 8 cm of actual height.
On average, reported weight was approximately
0.4 kg lower than actual weight ( 7 0.4 5.5 kgs)*,
and reported height was one cm lower than actual
height ( 7 1.0 5.1 cms). Females were more likely
to under-report their weight than males (1.0 5.5 kgs
vs 7 0.1 6.1 kgs, P < 0.001). There was no difference in the degree of under-reporting in females of
different races (P ˆ 0.4) or males of different races
(P ˆ 0.8). There was also no signi®cant difference
between under-reporting of height among males and
females (0.4 2.1 vs 0.5 2.1, P ˆ 0.9) or racial
groups (P ˆ 0.2) or pubertal stage (P ˆ 0.5). A small
proportion of adolescents reported completely erroneous weights and heights: Approximately 1% of
adolescents reported a weight over 22 kg different
from their actual weight, and 1% of adolescents
reported a height over 20 cm different than their
actual height.
Correlation between self-reported weight and actual
weight ranged between 0.87 and 0.94, depending on
gender or race (Table 2). Lower correlations were
noted in females compared to males (P < 0.001).
There were no signi®cant differences between correlations between blacks and whites (P ˆ 0.1). Correlation between self-reported height and actual height
ranged from 0.82 ± 0.91. Lower correlations were also
present among females (P < 0.001), but not racial
groups (P ˆ 0.5). Extremely poor correlations
* Mean s.d. throughout
Validity of self-reported height and weight
RS Strauss
906
between self-reported height and actual height were
observed in adolescents with reported heights less
than 150 centimeters (r ˆ 0.06, n ˆ 191) (Figure 1).
Correlations between body mass index as calculated
from actual weight and height compared to selfreported weight and self-reported height ranged
between 0.79 ± 0.89. Lower correlations in body
mass index were present in females (P < 0.01) and
blacks (P < 0.05).
Differences between actual weight and selfreported weight were signi®cantly greater for obese
children compared with non-obese children
(4.6 7.8 kgs vs 7 0.2 5.3 kgs, P < 0.001). In fact,
self-reported weight was within 5 kg of actual weight
in 92% of the normal weight children but only 59% of
the obese children (P < 0.001). Females were also
Table 2 Correlation between reported and actual weight and
height in young adolescents
Actual Weight
vs.
Reported Weight
All Children
Sex:
Male
Female
Race
White
Black
Hispanic
Actual Height
vs.
Reported Height
more likely to under-report their weight by more
than 5 kg compared to males (P < 0.01). In addition,
there was no difference in age between those children
who under-reported or over-reported their weight by
more than 5 kg Table 3.
Since obese children weighed on average 33 kg
more than non-obese children (88 23 kgs vs
55 11 kgs, P < 0.001), errors in self-reporting of
weight and height had little impact on classifying
whether an individual was obese (BMI > 95th%) or
non-obese. The use of self-reported weight and height
resulted in the correct classi®cation of obesity status
in 94% of children. Similar results were demonstrated
in both males and females (P ˆ 0.5) and among all
races (P ˆ 0.1). Consequently, odds ratios for obesity
related behaviours were similar whether obesity
(BMI > 95th percentile) was calculated using actual
weight and height or self-reported weight and height
(Figure 2).
Actual BMI
vs.
Reported BMI
Discussion
0.93
0.89
0.87
0.94
0.90***
0.91
0.82
0.89
0.84**
0.94
0.87
0.95
0.90
0.84
0.90
0.89
0.79*
0.85
*P < 0.05 blacks vs Whites/Hispanics; **P < 0.01 Females vs
Males; ***P < 0.001 Females vs Males.
Correlations of self-reported and actual weights and
heights in young adolescents ranged from approximately 0.85 ± 0.95. Although correlations between
actual weights and heights and measured weights
and heights are lower in adolescents than those
reported in adults,3 ± 5 they remain extremely reliable
Figure 1 Actual weight compared to self-reported weight in the sample of young adolescents (1A). Actual height compared to selfreported height in the sample of young adolescents (1B). Self-reported height and weight were randomly scattered by less than 1=2%
in order to visualize points that would fall on top of each other.
Table 3
Characteristics of children who under-reported or over-reported their weight by more than 5 kg
1) Over-estimate weight by more than 5 kg
2) Reported weight within 5 kg of actual weight
3) Under-estimate weight by more than 5 kg
n
Actual weight
(kg)
Actual height
(cm)
BMI
(kg/m2)
Obese
(%)
Gender
(% female)
Age
41
1429
167
57.3 15.7
60.4 13.6
77.3 20.6**
164 10
166 9
166 9
21.1 4.4
21.9 4.2
27.7 5.7**
9%
7%
47%**
54%
47%
59%
14.3 1.2
14.2 1.4
14.1 1.3
Mean s.d.
*P < 0.01 compared to group 2; **P < 0.001 compared to group 2.
Validity of self-reported height and weight
RS Strauss
Figure 2 Odds ratios for obesity related behaviours with 95%
con®dence interval. Obesity de®ned as a body mass index
greater than the 95th percentile for age and gender using
either measured height and weight or self-reported height and
weight. Frequent exercise de®ned as exercise enough to produce a sweat ®ve or more times per week. Children were also
categorized as breakfast skippers if they reported never eating
breakfast or only eating breakfast on weekends.
for predicting obesity related morbidities and behaviours. Since obese adolescents weighed approximately 33 kg more than the non-obese adolescents,
even moderate errors in self-reported height and
weight had minimal in¯uence on misclassi®cation of
obesity status. As a result, the odds ratios for weight
related behaviours (ie, dieting, television watching,
exercise, etc) were virtually identical whether weight
status was calculated using self-reported or actual
weights and heights.
Since self-reported weights and heights are easily
obtained, they are appropriate to utilize in surveys in
order to provide a reliable assessment of obesity
related behaviours. However, self-reported heights
and weights are not reliable for tracking weight
changes in longitudinal surveys where more precise
measurements are necessary. In addition, self-reported
heights and weights are not recommended for use in
studies of obesity related morbidities such as hypertension or hypercholesterolaemia where it is clearly
more appropriate to measure anthropometrics directly.
Finally, self-reported heights and weights are not
recommended for identifying an individual as `overweight' since a fairly narrow height and weight range
is required for accurate classi®cation (85th percentile < BMI < 95th percentile).
These data support the work of Epstein and colleagues who have also demonstrated that in children the
use of self-reported heights and weights results in the
correct classi®cation of obesity in 91% of children
aged 6 ± 12 y.10 Because children tend to under-report
both height and weight, calculations of body mass
index are less inaccurate than those obtained from
large studies in adults, where self-reported height is
over-estimated and self-reported weight is under-estimated. As a result, Kuskowska-Wolk et al have
demonstrated that over 30% of adults who were
overweight by measured weight and height would be
misclassi®ed by self-reported weight and height11
compared to the 6% misclassi®cation observed in
this study.
Although the correlations between self-reported
weight and measured weight were high, there existed
signi®cant gender and racial differences in reporting.
In particular, females are likely to under-report their
weight to a greater extent than males. Similar results
have been previously described in both children and
adults. Dwyer et al showed that teenage boys report
their weight more accurately than teenage girls.12
Discrepancies in self-reported weight and actual
weight were highest in heavy girls, similar to what
was observed in the current study.13 In adults,
women's self-reported weights are found to be consistently lower than their actual weights across the
weight spectrum; in contrast, only overweight men
tend to under-estimate their actual weight.14 ± 16
Gender differences in self-reported weight and
weight status may indicate a general desire to weigh
less.18 Gustafson-Larson et al 17 and Maloney et al 18
have also reported that pre-teen and adolescent girls
were almost twice as likely as similarly aged boys to
report a desire to be thinner.
There were no signi®cant differences in self-reported
height and weight between blacks and whites, although a
small difference in correlation between actual and
calculated BMI occurred between whites and blacks.
This is particularly surprising since signi®cant racial
differences exist in perception of weight status (for
example, `are you overweight?') between blacks and
whites. Neff and colleagues reported that white adolescent girls were almost twice as likely to perceive
themselves as overweight compared to black adolescent
girls.19 Racial differences in weight status perception
have also been reported in the Youth Risk and Behavior
Surveillance Survey (YRBSS),1 the 1985 National
Health Interview Survey (NHIS),3 and the Kids'
Eating Disorders Survey (KEDS).20
In conclusion, the in¯uences of gender and racial
biases in reporting of weight and height were relatively small. Biases in reporting weight and height
were much larger in obese individuals than non-obese
individuals, but remained inconsequential in a relationship with obesity related behaviours. Small differences in self-reported weight in females may be
related to a general desire to be thinner. However,
racial differences between self-reported weight and
actual weight were not observed despite previous
evidence that whites are more likely to perceive
themselves as overweight.
References
1 Serdula MK, Collins E, Williamson DF, Anda RF, Pamuk E,
Byers BE. Weight control practices of US Adolescents and
Adults. Ann Intern Med 1993; 119: 667 ± 671.
2 Resnick MD, Bearman PS, Blum RW, Bauman KE, Harris
KM, Jones J, et al. Protecting adolescents from harm.
Findings from the National Longitudinal Study on Adolescent
Health. JAMA 1997; 278: 823 ± 832.
907
Validity of self-reported height and weight
RS Strauss
908
3 Stunkard AJ, Albaum JM. The accuracy of self-reported
weights. Am J Clin Nutr 1981; 34: 1593 ± 1599.
4 Stewart AL. The reliability and validity of self-reported
weight and height. J Chron Dis 1982; 35: 295 ± 309.
5 Stevens J, Keil JE, Waid LR, Gazes PC. Accuracy of current,
4-year, and 28-year self-reported body weight in an elderly
population. Am J Epidemiol 1990; 132: 1156 ± 1163.
6 Brooks-Gunn J, Warren MP, Rsso J, Gargiulo J. Validity of
self-report measures of girls' pubertal status. Child Develop
1987; 58: 829 ± 841.
7 Must A, Dallal GE, Dietz WH. Reference data for obesity:
85th and 95th percentiles of body mass index and triceps
skinfold thickness. Am J Clin Nutr 1991; 53: 839 ± 846.
Erratum: Am J Clin Nutr 1991; 54: 773.
8 Himes JH, Dietz WH. Guidelines for overweight in adolescent
preventive services: Recommendations from an expert committee. Am J Clin Nutr 1994; 59: 307 ± 316.
9 Mohadjer L, Montaquila J, Waksberg J, Bell B, James P,
Flores-Cervantes I, Montes M. National Health and Nutrition
Examination Survey III: Weighting and estimation methadology. National Center for Health Statistics. Vital Health Stat
1(32). 1994.
10 Espstein LH, Mccurley J, Murdock RC. Estimation of
percent overweight within families. Add Behav 1991; 16:
369 ± 375.
11 Kuskowska-Wolk A, Karlsson P, Stolt M, Rossner S. The
predictive value of body mass index based on self-reported
weight and height. Int J Obesity 1989; 13: 441 ± 453.
12 Dwyer JR, Feldman JJ, Mayer J. The social psychology of
dieting. J Health Soc Behav 1970; 11: 269.
13 Dwyer J, Feldman JJ, Seltzer CC, Mayer J. Body image in
adolescents: Attitudes towards weight and perception of
appearance. J Nutr Educ 1969; 1: 14.
14 Pirie P, Jacobs D, Jeffery R, Hannan P. Distortion in selfreported height and weight data. J Am Diet Assoc 1981; 78:
601 ± 606.
15 Cash TF, Grant JR, Shovlin JM, Lewis RJ. Are inaccuracies in
self-reported weight motivated distortions? Percept Mot Skills
1992; 74: 209 ± 210.
16 Paltra M, Prineas RJ, Berman R, Hannan P. Comparison of
self-reported and measured height and weight. Am J Epidemiol
1982; 115: 223 ± 230.
17 Gustafson-Larson AM, Terry RD. Weight-related behaviors
and concerns of fourth-grade children. J Am Diet Assoc 1992;
92: 818 ± 822.
18 Maloney M, McGuire J, Daniels S, Specker B. Dieting
behavior and eating attitudes in children. Pediatrics 1989;
84: 482 ± 487.
19 Neff LJ, Sargent RG, McKeown RE, Jackson KL, Valois RF.
Black-white differences in body size perceptions and weight
management practices among adolescent females. J Adol
Health 1997; 20: 459 ± 465.
20 Childress AC, Brewerton TD, Hdges EL, Jarrell MP. The
Kids' Eating Disorders Survey (KEDS): A study of middle
school students. J Am Acad Child Adolesc Psychiatry 1993;
32: 843 ± 850.