Scand J Med Sci Sports 2016: : –
doi: 10.1111/sms.12695
Validity and reliability of elastic resistance bands for measuring
shoulder muscle strength
L. L. Andersen1,2, J. Vinstrup1, M. D. Jakobsen1, E. Sundstrup1
1
National Research Centre for the Working Environment, Copenhagen, Denmark, 2Physical Activity and Human Performance
Group, SMI, Department of Health Science and Technology, Aalborg University, Aalborg, Denmark
Corresponding author: Prof. Lars L. Andersen, PhD, National Research Centre for the Working Environment, Lersø Parkalle
105, DK-2100, Copenhagen, Denmark. Tel.: +45 39165200, Fax: +45 39165201, E-mail: lla@nrcwe.dk
Accepted for publication 30 March 2016
Valid and reliable measurements of muscle strength are
important in sport medicine. This study assesses
concurrent validity and intrarater reliability (test–retest
reliability) of elastic resistance bands for measuring
shoulder muscle strength. Altogether, 50 healthy adults
[mean age 36.0 (SD: 11.6), 29 women and 21 men]
participated in testing and retesting 1–2 weeks later. The
maximal elastic resistance (TheraBand) that each
participant could hold for 3 s during standing bilateral
shoulder abduction to 90° was converted into torque and
validated against gold standard maximal voluntary
isometric contraction (MVC) (Vishay force transducer)
performed unilaterally while lying supine. The intrarater
reliability of both tests were high; for the MVC and
elastic band test, respectively, ICC(3,1) was 0.98 (95%
Valid and reliable measurements of shoulder muscle
strength – or maximal voluntary force production –
are an important part of clinical evaluation of status
and progression of patients undergoing resistance
training or physical rehabilitation. In our laboratory,
we have used such measurements when comparing
shoulder muscle strength of healthy individuals with
chronic pain patients (Andersen et al., 2008a, c) as
well as when documenting strength gains among
healthy workers participating in physical exercise at
the workplace aimed at preventing neck and shoul-
der pain (Andersen et al., 2008b). Researchers con-
sider isokinetic or isometric dynamometers as gold
standard in strength testing (Fleck & Kraemer, 2004;
Ratamees et al., 2009; Marmon et al., 2013), but
these devices are expensive, often large and station-
ary, time consuming to use, and require specific
expertise. Thus, clinicians need alternatives. There
are several important components in assessing the
usefulness of alternative measuring methods.
First, the measuring instrument must be valid and
reliable. Researchers have developed guidelines such
as COSMIN and GRRAS to assess the quality of
ª 2016 John Wiley & Sons A/S.
Published by John Wiley & Sons Ltd
CI: 0.97–0.99) and 0.99 (95% CI: 0.98–1.00), and
measurement error was 4.8% (95% CI: 3.7–5.9) and
4.7% (95% CI: 3.1–6.2). For concurrent validity, ICC
(3,1) was 0.96 (95% CI: 0.95–0.98) and measurement
error was 8.1% (95% CI: 6.6–9.6), and the elastic band
test explained 93% of the variance in the MVC test.
However, the elastic band test produced systematically
lower torque values than the MVC [56.5 (SD: 26.8) vs
66.5 (SD: 25.5) Nm, P < 0.01]. In conclusion, the test
for shoulder muscle strength using elastic resistance
bands has excellent validity and reliability, but produces
systematically lower torque values than MVC. The
reason for the lower torque values may be that the
elastic band test has an initial concentric phase and is
performed bilaterally and standing upright.
validity and reliability studies (Mokkink et al., 2010;
Kottner et al., 2011). These guidelines are useful not
only for assessing the quality of studies but also in
the planning phase of the study. Intrarater reliabil-
ity – also known as test–retest reliability – is used to
assess the stability of measures performed by the
same raters or same equipment using the same indi-
viduals at different time points. Concurrent criterion
validity is used to assess whether a new test measures
what it is supposed to, that is, compared to a “gold
standard” measured in the same session. Finally,
responsiveness reflects whether a test is able to detect
actual change over time (Mokkink et al., 2010;
Kottner et al., 2011).
Second, clinicians often need portable and easy-to-
use devices to test muscle strength in different set-
tings and with different clients. Handheld devices to
measure muscle strength exist, where the clinician
typically applies force manually to resist the force
developed by the patient. However, there are limita-
tions to using handheld devices as the clinicians
experience and technique influence the outcome
(Bohannon, 1986; Byl et al., 1988; Newman et al.,
1
Andersen et al.
2012). Thus, a systematic review found acceptable Table 1. Descriptives of the study population
intrarater reliability of handheld muscle strength
All Women Men
assessment only for the elbow flexors, whereas con- Mean (SD) Mean (SD) Mean (SD)
flicting results were found for assessing shoulder
muscle strength (Schrama et al., 2014). Thus, for N 50 29 21
Age (years) 36.0 (11.6) 34.5 (12.1) 38.0 (10.8)
clinical purposes the majority of such shoulder tests Weight (kg) 70.7 (13.8) 63.8 (12.2) 80.2 (9.5)
does not meet common acceptability criteria. This Height (m) 1.73 (0.09) 1.67 (0.05) 1.81 (0.06)
may in part be due to the practical difficulties in BMI (kg/m2) 23.4 (3.2) 22.7 (3.6) 24.4 (2.3)
standardizing these tests. Therefore, there is still a Shoulder muscle strength
MVC (Nm) 66.4 (25.9) 49.0 (7.8) 90.5 (22.6)
need for developing valid and reliable methods for Elastic band test (Nm) 56.4 (27.2) 38.1 (6.9) 81.7 (24.4)
measuring shoulder muscle strength with portable
and easy-to-use devices. Elastic resistance bands Mean and standard deviations (SD) are from the original dataset (i.e.,
have shown accurate and repeatable force-length not the bootstrapped values).
properties in material testing studies (Hughes et al.,
1999; Patterson et al., 2001; Thomas et al., 2005). variables of this study. However, based on general guidelines
Because most physical therapists already use elastic for quality assessment of validity and reliability studies we
included 50 participants (Terwee et al., 2007; Mokkink et al.,
bands in clinical practice, it would be feasible also to
2010; Kottner et al., 2011).
use these for measuring muscle strength.
This study aims to test intrarater reliability and
concurrent validity of a new elastic resistance band Procedures
test compared with “gold standard” maximal volun- All participants engaged in two test sessions (test–retest) inter-
tary isometric contraction for measuring muscle spersed by 1–2 weeks in the same indoor laboratory during
strength. The “gold standard” used for comparison – similar conditions of temperature (21 °C), light, and verbal
MVC against a high-quality force transducer – has instructions. We chose a time period between the two test
shown to be highly reliable (ICC = 0.95) in repeated rounds long enough to avoid subsequent fatigue due to muscle
soreness and to minimize the chance of participants or
testing of isometric shoulder muscle strength with research assistants remembering test results from the first test,
10 weeks in between (Andersen et al., 2011). We while at the same time short enough to minimize the chance of
hypothesized that both reliability and validity of the participants changing their actual muscle strength due to
elastic resistance band test would be high (correla- other reasons (e.g., seasonal change in physical activity). The
tion coefficients >0.70). The reason for this expecta- same team of research assistants performed all testing and no
other persons, besides the participant, were present during
tion was that elastic bands have shown accurate and testing. One research assistant (author JV) instructed the par-
consistent linear force–length properties in prior ticipants, whereas the other research assistant (KH in
studies (Hughes et al., 1999; Patterson et al., 2001; acknowledgements) noted the results. Both research assistants
Thomas et al., 2005). had experience with similar types of testing in previous studies
at our lab and in the field. The results from the first test were
concealed from the participant as well as the research assis-
Methods tants during the follow-up retest.
Study population and ethical approval
We recruited a convenience sample of 50 healthy adults Strength test using force transducer (“gold standard”)
between October 2014 and March 2015 through personal con-
tacts, social media and email-based information to employees Muscle strength was assessed during maximal voluntary iso-
at our research center. Enrollment continued until we reached metric contraction (MVC), which is generally considered gold
50 participants. The inclusion criterion was an age between 18 standard in muscle strength testing (Fleck & Kraemer, 2004;
and 67 years. Exclusion criteria were pregnancy, life-threaten- Ratamees et al., 2009; Marmon et al., 2013). To measure
ing disease, blood pressure above 160/100 mmHg, and pain or force we used an industrial force transducer with extremely
musculo-skeletal disorders restricting shoulder abduction. All high precision and repeatability (KIS-2, 1 KN; Vishay Trans-
participants were informed about the purpose and content of ducers Systems, Karlskoga, Sweden). Figure 1 illustrates the
the study, and gave written informed consent to participate in force transducer attached to a rigid box, with the center of the
the study, which conformed to The Declaration of Helsinki, force transducer 0.17 m from the floor. Test–retest reliability
and was approved by the Local Ethical Committee (H-3-2010- of muscle strength with 10 weeks apart using this device has
062). Table 1 shows descriptive characteristics of the partici- previously shown to be excellent (ICC = 0.95) (Andersen
pants. et al., 2011). The research assistant instructed participant to
lie down supine during the test to remain completely stable, to
abduct the shoulder to 90°, and to position the middle of the
Sample size calculation wrist at the force transducer. Using visual inspection, the
research assistant ensured that the body was positioned per-
We expected a correlation coefficient of at least 0.70 (strong pendicular and the upper arm parallel to the wall behind the
association) between tests and retests as well as between the box (shoulder joint angle ~90°). The research assistant then
two different types of test, respectively. At a significance level instructed the participant to build up force over 2 s and then
of 0.05 and a power of 0.80 we would then need at least 21 exert maximal force for another 3 s in the direction of the
participants to show a significant correlation between the force transducer. Prior to the test the research assistant

2
 A new method for measuring muscle strength

Fig. 1. Illustration of the maximal voluntary isometric contraction (MVC) (left) and elastic resistance band test (right).

Resistance in resistance (yellow, red, green, blue, black, silver, gold). Because
Resistance in Newtons at the mechanical properties of elastic bands may change during
pounds at 100% 100% stretch the initial stretch cycles, all elastic bands were pre-stretched
Level stretch length length Color of elastic bands 100 times prior to the experiment, and the same set of bands
1 was used throughout the study. The TheraBand CLX has
2 consecutive loops of elastic bands, and the loops at each end
3
functioned as handles. The research assistant instructed partic-
4
ipants to stand upright on the middle of the elastic band with
5
the feet together without shoes. Starting with the lowest elastic
6
band resistance (yellow), the research assistant instructed par-
ticipants to perform standing bilateral shoulder lateral raise to
7
90° shoulder abduction and to maintain that position for 3 s
8
(static). Thus, participants had to perform the concentric con-
9
traction phase themselves – without help from the research
10
assistant – before reaching the static test position. Using visual
11
inspection, the research assistant ensured that the upper arms
12
were horizontally positioned (shoulder joint angle ~90°). After
13
1-min rest the research assistant instructed the participant to
14
attempt the next resistance level, and so forth, until the resis-
15
tance was so high that the position could not be reached and
16
maintained for 3 s. When participants failed at any given resis-
17
tance, two more attempts to reach and maintain the desired
18
position were performed. In all cases, the research assistant felt
that it was easy to distinguish between a successful and a failed
Fig. 2. The resistance of each level used in the elastic band attempt. Figure 2 shows the progression of resistance used
test. during in test. During the test the research assistant provided
verbal encouragement. The highest resistance that the partici-
pant could maintain statically at 90° shoulder abduction for
showed the force signal on the computer screen, while the par-
3 s was used for the subsequent analyses.
ticipant performed shoulder abduction to ensure that partici-
pant understood the test. During the test the research For subsequent analyses of muscle strength, we calculated
assistant provided verbal encouragement. The participants resistance based on the color and stretch length based on
had a maximum of five attempts with each arm with 1 min of information supplied from the manufacturer. Stretch length
rest in between. For each attempt the maximal root mean of the elastic band relatively to height of the subject was deter-
square value over a 500 ms moving window was calculated. mined in a pilot sample of 10 subjects of varying height, show-
For left and right arm separately, “maximal force” was ing that relative stretch length (percentage above resting
defined as the highest value of the five attempts. length) was equal to 134.5 times height (m) of the participant
minus 111.2. The relative stretch length was then multiplied
For subsequent analyses of muscle strength, shoulder joint
by the pre-specified resistance provided on the package
torque was defined as the lever arm length, using textbook
(Fig. 2) to obtain force. Subsequently torque was calculated
anthropometrical values (Sjøgaard, 1995), multiplied by the
based on this force and biomechanical calculations using text-
average of left and right arms maximal force value. Because
book standard anthropometrical values for relative length
the direction of force is perpendicular to gravity we did not
and weight of extremities in relation to height and weight of
include weight of the arm in the equation.
the participant (Sjøgaard, 1995). At 90° shoulder abduction,
the angle between the elastic band and upper arm were 60°.
Strength test using elastic bands The perpendicular force vector (based on stretch length and
color multiplied by the sine to 60°) was multiplied by the lever
For the strength test with elastic bands, we used TheraBand arm length to obtain torque. The total torque was then calcu-
CLX (Hygenic Corporation, Akron, Ohio, USA) with a stan- lated as the torque exerted against the elastic band added to
dard length of 1.47 m, ranging from very low to very high the torque exerted to resist gravity of the arm.

3
Andersen et al.

Fig. 3. Flow of participants through the study.

Table 2. Intrarater reliability (test–retest reliability) for the MVC and elastic band test, respectively

N Test, mean Retest, Difference test–retest, CV (95% CI) ICC (95% CI) SEM (95% CI) MDC (95% CI)
(SD) mean (SD) mean (SD)

MVC (Nm)
All 50 66.5 (25.5) 66.6 (25.3) 0.1 (4.5) 6.8 (5.3–8.3) 0.98 (0.97–0.99) 4.8 (3.7–5.9) 13.3 (10.3–16.2)
Men 21 90.4 (21.9) 89.7 (22.6) 0.7 (5.0) 5.6 (3.7–7.5) 0.97 (0.95–0.99) 3.9 (2.6–5.2) 10.9 (7.2–14.5)
Women 29 49.0 (7.5) 49.6 (7.7) 0.7 (4.0) 8.1 (6.1–10.0) 0.86 (0.77–0.94) 5.7 (4.3–7.0) 15.7 (11.9–19.5)
Elastic band (Nm)
All 50 56.5 (26.8) 57.7 (27.3) 1.2 (3.8) 6.6 (4.5–8.8) 0.99 (0.98–1.00) 4.7 (3.1–6.2) 12.9 (8.7–17.2)
Men 21 81.7 (23.4) 83.0 (24.3) 1.4 (2.7) 3.2 (1.5–5.0) 0.99 (0.98–1.00) 2.3 (1.1–3.5) 6.3 (2.9–9.6)
Women 29 38.1 (6.7) 39.3 (7.1) 1.2 (4.3) 11.2 (7.2–15.2) 0.78 (0.61–0.96) 7.9 (5.1–10.8) 21.9 (14.0–29.8)

Values shown for all (n = 50) as well as for men (n = 21) and women (n = 29) separately. Means, standard deviations (SD) and 95% confidence
intervals (95% CI) were obtained by bootstrapping (5000 resamples of the original data).
CV = coefficient of variation; ICC = Intraclass correlation coefficient (3,1); SEM = Standard error of measurements (percentage); MDC = Minimally
detectable change (percentage) on the individual level.

confidence and prediction limits. For the test of validity we

Statistical analyses
We used a range of statistical methods to assess validity and
reliability. Means and standard deviations (SD) described the
data from each test. Paired t-tests assessed possible differences
between test and retest values as well as between the two dif-
ferent types of test. The intraclass correlation coefficient
(ICC) is generally preferred to express reliability rather than
the Pearson product-moment correlation coefficient. Using
Proc Glm of SAS to obtain the sums of squares necessary for
the reliability calculations, ICC (3,1) was calculated according
to the method of Shrout–Fleiss (Shrout & Fleiss, 1979). Stan-
dard error of measurement (SEM), also called measurement
error, quantifies the precision of individual scores on a test
and was calculated as SD of the test scores divided by the
square root of 1 – ICC (Weir, 2005), and subsequently nor-
malized to percentage by dividing SEM by the mean of the
test score and multiplying by 100. The minimally detectable
change (MDC) reflects the threshold for determining “true”
individual changes beyond error and was calculated as SEM
times 1.96 times the square root of 2 (Weir, 2005), and subse-
quently normalized to percentage. Furthermore, the
coefficient of variation (CV) was calculated as SD of the
change scores from test to retest divided by the mean test and
retest score. Finally, we also performed linear regression anal-
yses (Proc Reg of SAS institute, Cary, NC, USA) between test
scores and illustrated these associations graphically with
further provided the residual plot from the linear regression
analysis to check for non-uniformity of error as recommended
by Hopkins (Hopkins, 2004). Finally, we provided the Bland
& Altman plot with average values plotted against mean dif-
ferences in the two tests along with 95% limits of agreement
(Bland & Altman, 1986).
There are several different, but no uniform, methods for
calculating confidence intervals of ICC, SEM, MDC and CV.
Using Proc Surveyselect of SAS, we therefore performed boot-
strapping – that is, we resampled the dataset and ran all statis-
tical analyses 5000 times – for all participants as well as for
men and women separately – to determine means and 95%
confidence intervals of all estimates. In this way, the method
for obtaining confidence intervals was uniform between the
different outcomes of this study.
Results
We enrolled 50 people in the study, and all these
completed both test and retest without missing data.
Table 1 shows that the men and women of this study
were comparable with regard to age, and the men’s
shoulder muscle strength was almost twice as high as
the women’s. Figure 3 shows that two persons,
besides the 50 participants, were excluded during the

4

initial examination due to pain that inhibited their
ability to perform maximal shoulder abduction.
Table 2 shows the bootstrapped mean test scores,
differences between test and retest, CV, ICC, SEM,
and MDC for the MVC and elastic band tests,
respectively. Figure 4 illustrates the x-y plot between
test and retest values for each type of test as well as
the respective 95% confidence limits and 95%
A new method for measuring muscle strength
prediction limits. Overall, Table 2 and Figure 4
show that both tests were highly reliable from test to
retest. Stratifying the analyses for gender, men had
better scores of reliability than women for both the
MVC and elastic band test (Table 2).
Table 3 shows the bootstrapped mean test scores,
differences, ICC and SEM between the MVC and
elastic band test. Figure 5 illustrates the x-y plot
between elastic band and MVC test values as well as
the respective 95% confidence limits and 95% predic-
tion limits. The elastic band test explained 93% of the
variance in the MVC test. Overall, Table 3 and Fig-
ure 5a shows that the elastic band test has excellent
validity compared with gold standard MVC, although
the torque values of the elastic band test are systemat-
ically lower than the MVC torque values (approxi-
mately 9–10 Nm) (Table 3). Based on Figure 5b, the
residuals may seem larger for the middle scores than
at the extremes, although there is no clear picture of
this as the residuals are generally small. Figure 5c, the
Bland–Altman plot, further illustrates that the two
methods should not be used interchangeably, and that
there tends to be higher errors with higher values.

Fig. 4. Intrarater reliability. Test (x-axis) and retest (y-axis)
values for the MVC (a) and elastic band test (b).
Discussion
The main finding of this study is that the new test for
shoulder muscle strength using elastic resistance
bands has excellent reliability and validity, although
the torque values of the elastic band test are system-
atically lower than gold standard MVC. Thus,
although the hypothesis was confirmed, that is, both
reliability and validity of the elastic resistance band
test was high, users should be aware that different
tests – although aimed at the same joint and muscles
– produce different results. Thus, the two different
tests should not be used interchangeably.
Several factors may explain that torque values were
systematically lower during the elastic band test than
during MVC. First, the elastic band test was bilateral,
whereas the MVC test was unilateral. A bilateral defi-
cit of force exists for some muscles, but seems to be
more pronounced during dynamic rather than isomet-
ric contraction (Jakobi & Chilibeck, 2001). Second,
the MVC test was performed in a highly stable posi-
tion – lying on the back with the full weight of the

Table 3. Concurrent validity of the elastic band strength test vs “gold standard” maximal voluntary isometric contraction (MVC)

N MVC (Nm), Elastic band (Nm), Difference elastic ICC (95% CI) SEM (95% CI)
mean (SD) mean (SD) band – MVC, mean (SD)

All 50 66.5 (25.5) 56.5 (26.8) 10.0 (7.0) 0.96 (0.95–0.98) 8.1 (6.6–9.6)
Men 21 90.4 (21.9) 81.7 (23.4) 8.8 (9.4) 0.91 (0.85–0.97) 7.8 (6.0–9.5)
Women 29 49.0 (7.5) 38.1 (6.7) 10.9 (4.2) 0.81 (0.65–0.96) 8.7 (6.3–11.2)

Values are shown for all (n = 50) as well as for men (n = 21) and women (n = 29) separately. Means, standard deviations (SD) and 95% confi-
dence intervals (95% CI) were obtained by bootstrapping (5000 resamples of the original data).
ICC = Intraclass correlation coefficient (3,1); SEM = Standard error of measurements (percentage).

5
Andersen et al.
(a)
(b)
(c)
Fig. 5. Concurrent validity of the elastic band strength test
vs “gold standard” maximal voluntary isometric contraction
(MVC) assessed by linear regression (a), the corresponding
residual plot (b), as well as the Bland & Altman plot with
average values plotted against differences in the two tests
(c). The horizontal lines in (c) represent mean diff and 95%
limits of agreement.
body to support stability. In contrast, the elastic band
test was performed standing upright with feet
together, which may be more technically challenging
and require a certain amount of postural balance. We
chose the standing test position for pragmatic rea-
sons, that is, therapists would find this position more
feasible than having patients lying on the ground.
6
Third, although the elastic band test required partici-
pants to hold the top position isometrically for 3 s,
the concentric portion of the movement may be a lim-
iting factor for some participants. Thus, we did not
test whether they would be able to hold a higher resis-
tance isometrically had they received help from the
research assistant during the concentric phase of con-
traction. Fourth, for the MVC we calculated the aver-
age torque of left and right shoulder, whereas the
weakest shoulder may be determining for the maxi-
mal resistance during the elastic band test. Fifth, for
the elastic band test to be approved participants had
to hold the position for 3 s, whereas the maximal
force value from the isometric MVC test was calcu-
lated over a 500 ms moving window. Together, these
factors may explain the systematically lower torque
values (~10 Nm) obtained during the elastic band test
compared with the MVC test. In spite of this, the elas-
tic band test had excellent reliability and validity, and
could serve as an inexpensive and simple test for
shoulder muscle strength, although the two different
tests should not be used interchangeably.
Overall, the reliability and validity of the elastic
band test was higher for men than women. The larger
heterogeneity of muscle strength among men than
women – as seen from the standard deviations of mus-
cle strength for each gender – may explain this, as the
variance between individuals is part of the equation for
calculating measures of reliability and validity.
Because therapists could potentially use the elastic
band test for evaluating individual patients, we also
determined the MDC. The MDC is calculated from
the measurement error and reflects the threshold for
determining “true” individual changes beyond error.
For example, based on Table 2, if a man improves
muscle strength 15% based on elastic band test, this
would likely reflect a true improvement in muscle
strength, but for a woman this change is within the
range of measurement error and therefore not neces-
sarily a true improvement.
Future studies should test the validity and reliabil-
ity of unsupervised elastic band test for muscle
strength. Thus, if patients are able to perform the test
on their own it could serve as a cost-effective monitor-
ing tool in eHealth and telemedicine. In addition, this
would allow measurements of muscle strength to be
collected for research studies in a cost-effective way.
Limitations
Our study has both strengths and limitations. A limi-
tation is that we did not test responsiveness of the
new test (Mokkink et al., 2010), for example, by
measuring changes in test scores after a period of
strength training or physical rehabilitation known to
increase muscle strength. Test scores from the gold
standard MVC used in this study has previously

shown to be responsive 10 weeks of strength training
(Andersen et al., 2011). Because the new test using
elastic bands does not have an analog scale partici-
pants would have to improve at least one step, for
example, from the blue to the black elastic band, to
improve their test score. This would have relatively
greater importance for those with lower muscle
strength, for example, for women compared with
men, because the relative difference between resis-
tance levels becomes smaller as the resistance level
increases. Thus, responsiveness should be tested in
future intervention studies. Aside from responsive-
ness – which would require a longitudinal study – we
followed guidelines for reliability and validity studies
(Mokkink et al., 2010; Kottner et al., 2011) in both
the preparation and reporting of the study.
Conclusion
In conclusion, the test for shoulder muscle strength
using elastic resistance bands has excellent validity
and reliability. This provides physical therapists with
a feasible and inexpensive test of shoulder muscle
strength. However, users of the test should be aware
that the torque values of the elastic band test are sys-
tematically lower than gold standard MVC.
Perspectives
While valid methods for measuring shoulder muscle
strength exist, they are expensive and require certain
A new method for measuring muscle strength
expertise. The newly develop test for shoulder muscle
strength using elastic resistance bands has excellent
validity and reliability. This provides therapist, clini-
cians and coaches in the field of sports medicine with a
feasible and inexpensive test of bilateral shoulder mus-
cle strength. However, because we tested only healthy
subjects, the test should also be validated among shoul-
der patients. Furthermore, similar tests could be devel-
oped and validated for other muscles groups. Finally,
it should be mentioned that the test requires a full set
of colors of elastic bands. From a practical point of
view, it may be easier to use just a single color and vary
the resistance by the length of the elastic band.
Key words: Reliability, validity, sports medicine,
muscle strength, shoulder, elastic resistance, test–
retest, ICC.
Acknowledgements
The authors are grateful to Klaus Hansen for valuable practi-
cal help during the project. Thanks to the Hygenic Corpora-
tion for supporting this study.
Contributors
LLA designed the study, performed the statistical analyses
and wrote the draft of the manuscript. JOV, MDJ, and ESU
collected the data, and provided critical feedback on the
manuscript. All authors contributed significantly to the inter-
pretation of the results, to the final version and approved the
submission.

References
Andersen LL, Holtermann A,
Jorgensen MB, Sjogaard G. Rapid
muscle activation and force capacity
in conditions of chronic
musculoskeletal pain. Clin Biomech
2008a: 23 (10): 1237–1242.
Andersen LL, Jorgensen MB,
Blangsted AK, Pedersen MT,
Hansen EA, Sjogaard G. A
randomized controlled intervention
trial to relieve and prevent neck/
shoulder pain. Med Sci Sports
Exerc 2008b: 40 (6): 983–990.
Andersen LL, Nielsen PK, Sogaard K,
Andersen CH, Skotte J, Sjogaard G.
Torque-EMG-velocity relationship in
female workers with chronic neck
muscle pain. J Biomech 2008c: 41
(9): 2029–2035.
Andersen LL, Saervoll CA, Mortensen
OS, Poulsen OM, Hannerz H, Zebis
MK. Effectiveness of small daily
amounts of progressive resistance
training for frequent neck/shoulder
pain: randomised controlled trial.
Pain 2011: 152 (2): 440–446.
Bland JM, Altman DG. Statistical
methods for assessing agreement
between two methods of clinical
measurement. The Lancet 1986: 1
(8476): 307–310.
Bohannon RW. Test-retest reliability of
hand-held dynamometry during a
single session of strength assessment.
Phys Ther 1986: 66 (2): 206–209.
Byl NN, Richards S, Asturias J.
Intrarater and interrater reliability of
strength measurements of the biceps
and deltoid using a hand held
dynamometer. J Orthop Sports Phys
Ther 1988: 9 (12): 395–398.
Fleck SJ, Kraemer WJ. Designing
resistance training programs.
Champaign, IL: Human Kinetics, 2004.
Hopkins WG. Bias in Bland-Altman
but not regression validity analyses.
Sports Sci 2004: 8: 42–46.
Hughes CJ, Hurd K, Jones A, Sprigle S.
Resistance properties of Thera-Band
tubing during shoulder abduction
exercise. J Orthop Sports Phys Ther
1999: 29 (7): 413–420.
Jakobi JM, Chilibeck PD. Bilateral and
unilateral contractions: possible
differences in maximal voluntary
force. Can J Appl Physiol 2001: 26
(1): 12–33.
Kottner J, Audige L, Brorson S,
Donner A, Gajewski BJ,
Hrobjartsson A, Roberts C, Shoukri
M, Streiner DL. Guidelines for
Reporting Reliability and Agreement
Studies (GRRAS) were proposed. J
Clin Epidemiol 2011: 64 (1): 96–106.
Marmon AR, Pozzi F, Alnahdi AH,
Zeni JA. The validity of
plantarflexor strength measures
obtained through hand-held
dynamometry measurements of
force. Int J Sports Phys Ther 2013: 8
(6): 820–827.
Mokkink LB, Terwee CB, Patrick DL,
Alonso J, Stratford PW, Knol DL,
Bouter LM, de Vet HC. The
COSMIN checklist for assessing the
methodological quality of studies on
measurement properties of health
status measurement instruments: an

7
Andersen et al.
international Delphi study. Qual Life
Res 2010: 19 (4): 539–549.
Newman BL, Pollock CL, Hunt MA.
Reliability of measurement of
maximal isometric lateral trunk-
flexion strength in athletes using
handheld dynamometry. J Sport
Rehabil 2012 Technical Notes (6).
Patterson RM, Stegink Jansen CW,
Hogan HA, Nassif MD. Material
properties of Thera-Band Tubing.
Phys Ther 2001: 81 (8): 1437–1445.
Ratamees NA, Alvar BA, Evetoch TK,
Housh TJ, Kibler B, Kraemer WJ,
Triplett NT. American College of
Sports Medicine position stand.
Progression models in resistance
training for healthy adults. Med Sci
Sports Exerc 2009: 41 (3): 687–708.
Schrama PP, Stenneberg MS, Lucas C,
van Trijffel E. Intraexaminer
reliability of hand-held dynamometry
in the upper extremity: a systematic
review. Arch Phys Med Rehabil
2014: 95 (12): 2444–2469.
Shrout PE, Fleiss JL. Intraclass
correlations: uses in assessing rater
reliability. Psychol Bull 1979: 86 (2):
420–428.
Sjøgaard G. Basisbog i arbejdsfysiologi,
bind I+II. København:
Arbejdsmiljøinstituttet, 1995.
Terwee CB, Bot SD, de Boer MR, van
der Windt DA, Knol DL, Dekker J,
Bouter LM, de Vet HC. Quality
criteria were proposed for
measurement properties of health
status questionnaires. J Clin
Epidemiol 2007: 60 (1): 34–42.
Thomas M, Muller T, Busse MW.
Quantification of tension in Thera-
Band and Cando tubing at different
strains and starting lengths. J Sports
Med Phys Fitness 2005: 45 (2):
188–198.
Weir JP. Quantifying test-retest
reliability using the intraclass
correlation coefficient and the SEM.
J Strength Cond Res 2005: 19 (1):
231–240.