A Comparison of Skinfold and Circumference Methods in Predicting Body Composition in Weight Trained Subjects

Introduction | Body Composition Formula | Methods | Conclusion | Bibliography


Skinfold Body Composition Testing

Typically, a simple and reasonably accurate method of assessing changes in body composition have included skin fold measurements. The seven point skin fold method is used by many exercise consultants who seek a relatively accurate, simple, and inexpensive measurement of body composition. Clients who engage in exercise or dietary manipulation can receive immediate feedback so timely modifications can be made to their program. Dependent upon the expected rate of change in body composition, clients can be measured every month or so. If the rate of change is expected to be rapid, as with dieters, body composition can be sampled more frequently, e.g.. weekly. Caution should be exercised in interpreting data when tests are performed frequently, since measurement error, or variability may exceed an actual change in body composition. In which case, the general trend of the results can be noted or additional tests can supplement body composition data, such as waist circumference measurement or other girths. When performed as recommended, the whole process of the measuring body composition with the seven point skin fold method and calculating the absolute amount of fat and lean body weight can be time consuming, especially with many clients. In this study, in effort to save time and reduce test variability, two other tests that predict body composition were compared with seven point skin fold method and are discussed.

The subcutaneous fat comprises of approximately one half the total fat in the body. By the end of World War I, anthropologists realized the feasibility of measuring subcutaneous fat. By 1930, researchers developed an instrument to measure selected sites on the body with relative accuracy. (Katch and McArdle, 1983)

The thickness of a double layer of skin and the fat beneath it is measured with the special caliber that exerts a constant tension on the site. Skinfolds must be taken at precise standard locations if the results are to be reliable and used for comparative purposes. (Katch and McArdle, 1983)

Lohman and Roche made recommendations on techniques to derive skin fold readings based upon a variety of methods used in the literature throughout the years. Reliability and validity of the many sites were reported in predicting body density. (Lohman and Roche, 1998)

Katch and McArdle express that population specific equations only accurately predict body fat (BF) for a subject that is similar to the population that the formula was derived from. According to Katch and McArdle, the major drawback with the skinfold technique lies in extensive expertise, one must have in taking readings accurately and with consistency. Katch and McArdle claim that with research purposes, the investigators usually has already performed several thousand skin folds. (Katch and McArdle, 1993)

Body Composition Formulas

Jackson and Pollock describe certain limitations associated with population specific formulas and linear regression models in attempting to estimate BF from skin fold measurements. They noted that with a sample of men of different age, the slopes of the regression lines were the same, but the intercepts were significantly different. In addition, the slopes of the regression lines were not parallel between young adult men and extremely lean world class runners. Furthermore, research has shown a curvilinear relationship exists between skin fold measurements and body density (BD). Jackson and Pollock proposed that this non-linear relationship may be the reason for the differences in slopes and intercepts. They develop several formulas based upon a quadratic relation and the function of age. (Jackson and Pollock, 1978)

The following formula is one of Jackson and Pollock's that has been used widely in research:

  • BD= 1.11200000 - 0.00043499(X) + 0.00000055(X)(X) - 0.00028826 (A) X = Sum of chest, axilla, triceps, subscapular, abdomen, suprailium, and thigh skin folds (mm). A = Age in years.

This formula had a correlation of .915 when compared to hydrostatic weighing to estimate body density. The standard error of estimate was .008 expressed in body density and a standard error of estimate of 3.5 when expressed in % BF using the following equation by Siri (1961):

  • % Fat = [(4.95/BD) - 4.5] 100

The absolute amount of fat and lean body weight (LBW) can be calculated using the body weight of the individual.

Behnke and Wilmore have developed several formulas to estimate BD and LBW based upon anthropometrics they obtained from 133 college age men. The anthropometrics used in the predictive formulas included skinfolds, diameters, and circumferences. The authors noted that the correlation coefficients for estimating LBW was higher that for estimating BD, although the standard error of estimate was higher in estimating lean body weight. It seems, though, an oversight was made in their comparison of the standard error of estimates of the two different types of formulas. The formulas that were used to estimate LBW were expressed in kilograms, whereas the formulas used to estimate BD were expressed as grams per milliliter. A direct comparison needs to be made to compare the standard error of estimate for the two different types of formulas. They did, though, caution the use of these formulas until further studies could either confirm or refute their validity, since the population used to derive data was of a specific population. (Behnke and Wilmore, 1969)

The following formula offered by Behnke and Wilmore seems to offer some savings of time in the way of data collection and calculation:

  • LBW= 10.260 + 0.7927(wt) - 0.3676(s) wt = Body weight (kg) s = Abdominal skinfold (mm)

This formula had a correlation of 0.931 when compared with hydrostatic weighing. The standard error of estimate was 2.977 expressed kilograms. It must be noted, though, that misleading comparisons may be made in comparing the correlation of data derived from formulas to that of hydrostatic weighing in different studies. This is because hydrostatic weighing in itself has a standard error of estimate.

Katch and McArdle recognize that circumferences may be taken with ease, achieving accuracy with little practice. They recommend to record measurements in centimeters for added precision. In addition, an average of two measurements at the same site was advised. They also noted that pulling the tape too tight would compress the underlining soft tissue making the measurement inaccurately smaller. (Katch and McArdle, 1983)

A simple formula using an abdominal circumference is offered by Behnke and Wilmore:

  • LBW= 44.636 + 1.0817(wt) - 0.7396(c) wt = Body weight (Kg) c = Circumferences of Umbilicus Abdomen (cm)

This formula had a correlation of .938 when compared with hydrostatic weighing. The standard error of estimate was 2.815 expressed in kilograms.

Methods, Data, and Statistical Analysis

To examine the new formulas explored above, seven male subjects were included for testing. All subject desire LBW gains and a reduction of BF. All but one subject engaged in weight training and supported a larger than average musculature. The anthropometrics necessary to complete all three formulas discussed above were collected and calculated. Only two subjects have more than one set of data.

Subjects Percent Body Fat
Seven Site Abdominal Skinfold Abdominal Circumference
WY 22.68 25.17 17.18
24.62 19.55 11.65
24.62 29.02 18.57
10.31 15.62 12.78
8.94 15.50 12.48
6.93 13.81 11.52
AR 14.92 19.89 15.47
BM 7.93 14.10 12.77
CT 25.17 28.91 19.52
22.55 27.50 18.08
AF 23.6 30.50 21.50
B 15.59 21.60 19.89
KM 7.73 14.58 13.78

The data were compiled and analyzed using both T-test (LSD) and Tukey's Studentized Range (HSD). These tests were used to determine the relationship between all three formulas. Both tests indicated a significant difference between the single abdomen skin fold and the other two methods. No significant difference was found between the seven site skin fold method and the abdominal circumference method.


Critical Value of T= 2.04
Least Significant Difference= 3.8147

Tukey's Studentized Range Test

Critical Value of Studentized Range= 3.486
Minimum Significant Difference= 4.6047

T-test & Tukey's Studentized Range Test

Alpha= 0.05
df= 30
* Means with same letter are not significantly different.
Method T Grouping Tukey Mean N
AB SF A A 21.212 13
7 Site B B 16.583 13
AB CR B B 16.168 13


A closer analysis of individual data shows a general trend. The abdominal skin fold method appears to overestimate BF when compared to the seven point method. On the other hand, the abdominal circumference method underestimates BF when the subject is above average BF according to the seven site method. This "cut off" seemed to be somewhere within 15.59 % BF and 22.68 % BF. When BF was 15.59 % and lower, according to the seven site method, the abdominal circumference method overestimated BF. It can be seen how a greater than normal musculature of the midsection could falsely decrease the LBW calculation, therefore increasing the % BF calculation. The overestimation of BF on some calculations and an underestimation of BF on others is most likely responsible for the seemingly high correlation with the seven point method shown above. Despite the ambiguity of the abdominal circumference method, it was thought to be more reliable than the seven site skinfold with those individuals above 22.68 % BF. It is known that more obese individuals pose a greater threat to the reliability of skinfold measurements. (Lohman and Roche)

It became evident only after the first few calculations that the trial formulas were not applicable to the population tested in this study. As mentioned before, Behnke and Wilmore's formula's were derived from average age college students. The subjects used in this study were, probably at best, average weight-trained college students with greater than average muscle mass (ie: amature bodybuilders).


Behnke, A.R., Wilmore, J.H. An anthropometric estimation of body density and lean body weight in young men. Journal of Applied Physiology. 27: 25-31, 1969.

Jackson, A.S., Pollock, M.L. Generalized equations for predicting body density. British Journal of Nutrition. 40: 497-504, 1978.

Katch, F.I., McArdle, W.D., Nutrition, Weight Control, and Exercise . Lea & Febiger: Philadelphia. PA, 1983.

Lohman, T.G., Roche, A.F. & Martorell, R. (eds.). Anthropometric Standardization Reference Manual . Human Kinetics Books: Champaign, IL ISBN O-87322-121-4. 1998.

Related Articles