Skip to content
PounceFit
PTEN

Technical error of measurement (TEM): calculation and acceptable values

By PounceFit Team · Updated on

Assessor measuring a man's neck with a tape measure while another assessor watches
Photo: Paul D. Honnick, U.S. Navy (public domain), via Wikimedia Commons

Technical content for health and fitness professionals. It reproduces formulas and protocols published in the scientific literature, with the sources at the end of the page. It is not a recommendation and does not replace the judgment of the professional who runs the assessment.

The technical error of measurement (TEM) is the standard deviation between repeated measurements of the same person and shows how much the assessor varies when repeating the technique. With two measurements per person, TEM = √(Σd² ÷ 2n), where d is the difference between the 1st and 2nd measurement and n is the number of people; the relative TEM is the TEM divided by the mean of the measurement, times 100 (Perini et al., 2005). For skinfolds, the acceptable relative TEM ranges from 5% to 10%, depending on the assessor's experience and the type of analysis (Perini et al., 2005).

ISAK uses the TEM in the accreditation of anthropometrists, and it is how the assessor knows whether a difference between assessments is real change or variation in their own technique (Perini et al., 2005).

Intra-assessor and inter-assessor TEM

  • Intra-assessor: the same assessor measures the same people twice. It shows the repeatability of the person measuring.
  • Inter-assessor: different assessors measure the same people, on the same day, without seeing each other's measurement. It shows how much the assessors agree with one another. It is useful where several professionals assess the same clients (Perini et al., 2005).

The intra TEM is usually smaller than the inter, because each assessor repeats their own way of measuring (Norton, 2018).

How to calculate the TEM

According to Perini et al. (2005):

Absolute TEM (in the same unit as the measurement, mm or cm):

TEM = √(Σd² ÷ 2n)

  • d: difference between the 1st and 2nd measurement for each person.
  • Σd²: sum of the squared differences.
  • n: number of people measured.

Relative TEM (in %):

TEM (%) = TEM ÷ MVV × 100

  • MVV: mean value of the variable, the mean of each person's means.

Rules of use (Perini et al., 2005):

  • The TEM has the same unit as the measurement.
  • It holds only for that measurement and that equipment.
  • It holds only for populations similar to the one measured: calculated in athletes, it applies to athletes.
  • Use at least 20 measurements, at the same time of day.

Another way to express reliability is the reliability coefficient, R = 1 − (TEM² ÷ SD²), where SD² is the total variance between people; the closer to 1, the better, and Ulijaszek and Kerr (1999) recommend aiming for R above 0.95.

Worked example with the triceps skinfold

An assessor measured the triceps skinfold of 10 clients on two different days, in the morning, with the same caliper. With 10 people the calculation is short; in practice, use at least 20 (Perini et al., 2005).

Client 1st measurement (mm) 2nd measurement (mm) d d²
1 12.0 12.4 −0.4 0.16
2 15.5 15.0 0.5 0.25
3 9.8 10.2 −0.4 0.16
4 20.2 19.6 0.6 0.36
5 14.0 14.6 −0.6 0.36
6 11.4 11.0 0.4 0.16
7 18.6 19.4 −0.8 0.64
8 8.6 8.8 −0.2 0.04
9 16.0 15.4 0.6 0.36
10 13.2 13.6 −0.4 0.16
  1. Sum of squares: Σd² = 2.65.
  2. Absolute TEM = √(2.65 ÷ (2 × 10)) = √0.1325 = 0.36 mm.
  3. Mean of each client's means: MVV = 13.96 mm.
  4. Relative TEM = 0.36 ÷ 13.96 × 100 = 2.6%.

By the table below, 2.6% is within the acceptable range even for the experienced assessor (5%).

What is an acceptable TEM?

Relative TEM considered acceptable, according to the table by Gore et al. reproduced by Perini et al. (2005):

Analysis Measurement Novice assessor Experienced assessor
Intra-assessor Skinfolds 7.5% 5.0%
Intra-assessor Other measurements 1.5% 1.0%
Inter-assessor Skinfolds 10% 7.5%
Inter-assessor Other measurements 2.0% 1.5%

Norton (2018) gives the same targets for the assessor: up to 5% for skinfolds and 1% for other measurements within one assessor; up to 7.5% and 1.5% between assessors. Skinfolds tolerate more error because they are the least precise measurements in anthropometry, and the error tends to grow where the skinfold is larger (Perini et al., 2005).

Real change or measurement error?

Norton (2018) shows how to use the relative TEM to decide whether a skinfold really changed between two assessments:

  1. D = measurement 1 − measurement 2.
  2. SE = TEM (%) × √2.
  3. Limits = D ± 2 × SE ÷ 100 × mean of the two measurements.

If the interval between the limits includes zero, you cannot say a change occurred. If it does not, there is about 95% confidence that the change is real.

With the 2.6% TEM from the example (SE = 2.6 × 1.414 = 3.7%):

  • Triceps from 14.0 mm to 12.6 mm: D = 1.4; mean = 13.3; margin = 2 × 3.7 ÷ 100 × 13.3 = 0.98. Limits of 0.42 to 2.38 mm: they do not include zero, so the change is probably real.
  • Triceps from 14.0 mm to 13.4 mm: D = 0.6; mean = 13.7; margin = 1.01. Limits of −0.41 to 1.61 mm: they include zero, so the difference may be measurement error alone.

How to reduce your TEM

  • Practice the technique before measuring clients and compare your measurements with those of an experienced assessor (Norton, 2018). See the step by step in how to measure skinfolds.
  • Recalculate the TEM from time to time to follow how your technique is evolving (Perini et al., 2005).
  • Keep the same assessor for the same client's reassessments: differences between assessors tend to be larger than those of the same assessor (Norton, 2018).

The TEM also gives meaning to the sum of skinfolds: a drop in the sum smaller than the assessor's error is not change. The same principles apply to body circumferences. To record repeated measurements on paper, use the printable body assessment form. And to keep conditions equal between assessments, see how often to reassess.

Frequently asked questions

What is an acceptable TEM for skinfolds?

In the table used by Perini et al. (2005), the acceptable relative TEM for skinfolds is up to 7.5% for the novice assessor and up to 5% for the experienced one, for repeated measurements by the same assessor. Between different assessors, the limits are 10% and 7.5%.

How many people do I need to measure to calculate the TEM?

Perini et al. (2005) recommend at least 20 measurements, taken at the same time of day. Norton (2018) also speaks of repeated measurements on at least 20 people to establish reproducibility.

Does the TEM apply to any measurement?

No. It holds only for the measurement and the equipment it was calculated with, such as the triceps skinfold with one caliper model, and for populations similar to the one measured (Perini et al., 2005). Calculate one TEM for each skinfold and girth.

How does the TEM help in reassessment?

It tells you how much the measurement varies from technique alone. With the relative TEM, you can calculate an interval around the difference between two assessments: if the interval does not include zero, the change is probably real (Norton, 2018).

Sources

  1. Perini TA, Oliveira GL, Ornellas JS, Oliveira FP. Cálculo do erro técnico de medição em antropometria. Revista Brasileira de Medicina do Esporte. 2005;11(1):81-85. doi:10.1590/S1517-86922005000100009
  2. Norton KI. Standards for anthropometry assessment. In: Norton K, Eston R, editors. Kinanthropometry and exercise physiology. 4th ed. London: Routledge; 2018. p. 68-137. doi:10.4324/9781315385662-4
  3. Ulijaszek SJ, Kerr DA. Anthropometric measurement error and the assessment of nutritional status. British Journal of Nutrition. 1999;82(3):165-177. doi:10.1017/S0007114599001348