Content uploaded by Paul Swinton
Author content
All content in this area was uploaded by Paul Swinton on Oct 21, 2016
Content may be subject to copyright.
Short Communication
Reliability and validity of the running anaerobic sprint test
(RAST) in soccer players
Katherine Burgess, Thomas Holt, Steven Munro, Paul Swinton
Objectives: To investigate the validity and relative and absolute reliability of the Running Anaerobic Sprint Test (RAST) in ama-
teur soccer players.
Design: Cross-sectional experimental design with an element of repeated measures.
Methods: Twenty three males completed the RAST on two occasions and a Wingate test (WAnT) as criterion measure of anaero-
bic power.
Results: Criterion validity for the RAST was strong for peak power (r = 0.70, p < 0.001) and average power (r = 0.60, p = 0.002);
however, the RAST significantly underestimated peak power compared to WAnT. The RAST showed very good relative reli-
ability for average power, ICC = 0.88 and good relative reliability for peak power, ICC = 0.72. Assessment of absolute reliabili-
ty highlighted that although when averaged across a group, test and re-test scores will be similar, when monitoring individuals
an individual’s retest score may range between 0.81 and 1.2 times the original value for peak power and between 0.9 and 1.16
for average power.
Conclusion: The RAST is a practicable field test to estimate levels of average anaerobic power. However, the results show that
the RAST is not sensitive enough to detect strongly individual changes below 20 % and is therefore not recommended to con-
tinually monitor an individual’s anaerobic power. Also, if true measures of peak power are required the RAST test is limited.
(Journal of Trainology 2016;5:24-29)
Key words: field test anaerobic power specificity football
INTRODUCTION
During most field based team sports, athletes are required to
perform repeated sprint efforts. In soccer, sprinting bouts gen-
erally occur every 90 s and each of these sprints last approxi-
mately 2-4 s.1,2 During a game of soccer 1 % to 11 % of the dis-
tance covered is done so whilst sprinting and each outfield
player performs 1000 to 1400 high intensity short duration
activities.3 Although aerobic metabolism dominates the energy
delivery during a soccer game, the most decisive actions are
engendered by means of anaerobic metabolism 3. In addition, it
has been shown that elite professional players cover greater
cumulative distances by high-intensity running and sprinting
than professional players of a lower standard during the course
of a game.4 This critical reliance on a soccer player’s ability to
generate anaerobic power creates interest in methods that can
be used to objectively assess this ability.
The Running Anaerobic Sprint Test (RAST) was developed
in 1997 by Draper and Whyte to provide a means of determin-
ing anaerobic power, which was both inexpensive and simple
to implement and thus accessible to coaches for players of all
levels.5 In addition, the test was founded on the basis of pro-
viding a more specific assessment of power for running based
sports as it utilises flat sprinting rather than other modes of
exercise such as cycling or staircase running.6,7 The RAST
comprises performance of six 35 m sprints with 10 s rest inter-
vals. The power produced during each sprint is calculated
based on the mechanical principle that power is the product of
force and velocity. Average velocity for each 35 m sprint is cal-
culated using the sprint time and known distance (veloci-
ty = displacement/time). Acceleration is then calculated
through change in velocity/time, whereby average velocity is
used to represent the change in the quantity due to the observa-
tion that the initial velocity is equal to zero. Consequently,
force is calculated based on Newton’s second law (force =
mass × acceleration). As a result, power calculated from per-
formance in the RAST is based on assumed uniform motion of
the individual.
Since its conception in 1997, the RAST has become widely
used by sports teams despite limited research evaluating the
effectiveness of the test. Hodson and Jones were the first to
report data concerning the reliability of the RAST when they
included the test to investigate the effects of caffeine ingestion
on repeated sprint ability.8 The authors reported reliability
coefficients ranging between r = 0.92 and r = 0.97 for all test
performance measures, although they did not specify how the
reliability data were determined or the specific r value for each
of the measures. Subsequently, reliability of the test has been
reported by others9-11 through the use of test-retest correlations
or Intraclass Correlation Coefficients (ICCs). Both statistics
provide an indication of the test’s relative reliability but not its
absolute reliability. In order to present a more robust account
of the measurement error in a performance test it is recom-
mended that statistics measuring both relative and absolute
24
Received June 22, 2016; accepted August 5, 2016
From the School of Health Sciences, Robert Gordon University, Aberdeen, AB107QG, UK (K.B., T.H., S.M., P.S.)
Communicated by Takashi Abe, PhD
Correspondence to Dr. Katherine Burgess, School of Health Sciences, Robert Gordon University, Aberdeen, AB107QG UK. Email: k.burgess@rgu.
ac.uk
Journal of Trainology 2016;5:24-29 ©2012 The Active Aging Research Center http://trainology.org/
Burgess et al. Reliability and validity of the running anaerobic sprint test (RAST) in soccer players 25
reliability be provided12. In a recent study conducted by
Zagatto et al.11 relative and absolute reliability of outcome
measures taken from the RAST were assessed using ICCs and
Bland-Altman plots, respectively. However, the authors failed
to meaningfully interpret data from the Bland-Altman plots, in
particular with regards to the sensitivity of the test to detect
real changes that may occur due to training. Therefore, as one
of the main uses of the RAST is to monitor training progress5,
it is important that reliability statistics assist coaches to deter-
mine the likely measurement error in a test score and the mag-
nitude of change that is required in order to be confident that
different test scores reflect a change in physical status rather
than simply random variability.
In addition to assessing reliability of the RAST researches
have also investigated the validity of the test through compari-
son with a criterion measure. Although there is no universally
agreed ‘gold standard’ for measuring anaerobic power, the
Wingate test6 is widely accepted as a criterion and has been
used to investigate the validity of a range of different anaero-
bic tests13-15. Significant correlations between the WAnT and
RAST have been reported for peak power (ranging from
r = 0.46 to r = 0.90) and mean power (ranging from r = 0.53
to r = 0.975)16,11. Conversely, non-significant correlations
between these measures have also been reported.17 However,
the study by Keir et al.17 was conducted on small number of
participants (n = 8) and did not adhere to standard protocols
incorporating an unloaded acceleration phase immediately
prior to the initiation of the 30 s Wingate test. Of the four pre-
vious studies which reported significant correlations, three of
these did not provide full details of their methods16,10,18 leading
to caution over their interpretation. The fourth study11 utilised
members of the armed forces as participants. Army personnel
typically undergo physical training with a focus on aerobic and
muscular endurance and strength and rarely include intervals
of less than 100 m in their training.19,20 Conversely, soccer
players’ training (and match play) includes the performance of
repeated short sprints.1 This element of training and testing
specificity could lead to differences in the criterion validity of
these tests between these differing populations. In addition, all
four of the aforementioned studies either completed the RAST
in a controlled indoor facility or did not state the test location.
Although using controlled indoor conditions enhances the
internal validity of the studies, it could potentially limit the
applicability to outdoor sports who would conduct this test at
their outdoor training facilities.21,22 Therefore, the aim of this
study is to investigate the criterion validity, relative reliability
and absolute reliability of the RAST in soccer players when
conducted in an outdoor environment.
METHODS
Participants and Study Design
Twenty three male amateur soccer players (age 24 ± 3 years,
mass 75.4 ± 5.9 kg, and height 180 ± 5 cm) participated in the
study. All participants played/trained at least three times a
week and the group had an average of 7 ± 4 years’ experience
of playing at club level. The investigation was approved by the
University Institutional Ethics Committee, and all participants
gave their written informed consent to participate in the study.
The study is in agreement with the declaration of Helsinki of
the World Medical Association. All participants performed
three testing sessions, all a minimum of two days and a maxi-
mum of seven days apart. The participants were asked to
refrain from partaking in strenuous exercise for a minimum of
24 hours prior to testing. There were instructed to maintain
their normal diet, ensure they had eaten on the day of testing,
but were asked to refrain from eating a full meal in the 2 hours
prior to the testing sessions.
During two of the testing sessions participants completed the
RAST and during the other participants completed the WAnT.
The two RAST conditions were always carried out in consecu-
tive sessions but the first test (WAnT or RAST) was ran-
domised. For each participant all testing sessions were carried
out at the same time of day following a standardised warm up.
The warm up included five minutes of pulse raising activities
(jogging, high knees, heel flicks and lunges) and two practice
sprints at 75 % perceived maximal efforts (35 m running
sprints prior to the RAST test and 5 s cycle sprints prior to the
WAnT). At the beginning of all testing sessions clothed body
mass was measured using digital scales and standing height
was recorded prior to the first test only. Prior to the testing ses-
sions participants completed a familiarisation session in which
they performed both RAST and WAnT tests once.
Running Anaerobic Sprint Test
To complete the RAST participants were required to per-
form six maximal 35 m sprints on an AstroTurf pitch with 10 s
rest periods between each sprint. Players were instructed to
wear their normal training footwear (this was moulded football
boots in all cases). The time for each sprint was recorded using
a Brower timing gate system (Brower Timing Systems, USA)
with photocells positioned 35 m apart at approximately waist
height. The participant started each sprint 0.3 m behind the
timing gate23 (see Figure 1) and performed repeated sprints in
alternate directions. The 10 s rest periods were timed using a
stop watch and a tester gave the participant a 3 s count down
prior to each sprint. Weather conditions during testing were
dry and cold (3-6°C) with little wind.
The power produced during each sprint was determined by
the following formula: Power = (Body Mass × Distance2)/
Time3. Peak power was defined as the power obtained during
the fastest sprint and average power (for all six sprints) was
calculated by taking the mean.
Figure 1 Diagramatical representation of RAST test set up.
Journal of Trainology 2016;5:24-2926
The Wingate Test
The WAnT test required participants to cycle at maximum
cadence on a Monark cycle ergometer (Monark 894E ergo-
medic peak bike) for 30 s against a resistance equivalent to
7.5 % of their body mass. Pedal revolution rate and conse-
quently power output was measured using Monark Anaerobic
test software. Participants accelerated up to maximum cadence
against zero resistance before the load was applied and the test
began. Power was determined over 1 s time intervals with peak
power measured as the maximum value obtained and mean
power calculated over the full test (30 s).
Statistical Analysis
Criterion validity was assessed with Pearson correlation
coefficients to quantify the relationship between power values
measured during the RAST (mean of tests 1 & 2) and WAnT,
whilst paired t-tests were used to compare differences in the
magnitude of power values calculated. Correlation coefficients
ranging from 0.4 to 0.59 were categorised as indicating a mod-
erate linear relationship, 0.6 to 0.79 were categorised as
strong, and 0.8+ were categorised as very strong.24 Relative
reliability of test and retest scores of outcome variables mea-
sured from the RAST were assessed by intra-class correlation
coefficients (ICC2,1) using a 2-way random model with abso-
lute agreement and 95 % CIs. ICC values were interpreted
using the following guidance: 0.41 to 0.60 as moderate reli-
ability, 0.61 to 0.80 as good reliability and 0.81 + as very good
reliability.25 Absolute reliability of the same data was quanti-
fied using the 95 % limits of agreement (LOA) method origi-
nally described by Bland and Altman26. Firstly, tests of system-
atic bias between test and re-test scores were assessed using
paired t-tests. No evidence of systematic bias was found for
any of the comparisons made. However, as recommended by
Atkinson and Nevil21 the 95 % LOA were still to be expressed
as X
–
diff ± (1.96 × Sdiff) where X
–
diff is the difference between the
average of the test and re-test scores and Sdiff is the standard
deviation of the difference scores. Expressed in this way, the
95 % LOA provide a measure of total error (bias ± random
error) where the bias is only slight. Prior to presentation of
these results, occurrence of heteroscedasticity in the data was
investigated for each dependent measure by calculating the
Pearson correlation coefficient between the mean of partici-
pants tests scores and the absolute value of the differences.
Relatively large positive correlation coefficients were obtained
for all variables (r = 0.31 to 0.55) indicating that the amount of
random error increased as the measured values increased (i.e.
data were heteroscedastic). As a result of these findings, the
original test data were log-transformed using the natural loga-
rithm and the LOA procedure was performed using the trans-
formed data.26 Dimensionless ratios were calculated by taking
the antilog of the bias exp{X
–
diff(ln)}; where X
–
diff(ln) is the differ-
ence between the average of the log-transformed test and re-
test scores and the antilog of the random error component
exp{1.96 × Sdiff(ln)}; where Sdiff(ln) is the standard deviation of
the difference of the log-transformed scores.26 As a result,
95 % of the ratios of test scores (i.e. test /retest) should lie
between the antilog of the bias multiplied and divided by the
antilog of the random error component. Finally, the minimum
difference (MD) statistic which can be used as a guide for the
required change in the RAST test after a period of training to
detect a ‘real’ change was computed. MD was calculated by
multiplying the standard error of the mean (SEM) by 1.96 and
2
. The SEM was calculated using the following equation:
SEMICC1,21
=-
, where SD is the standard deviation of all
scores from the test.27
RESULTS
Validity
Criterion validity was strong for peak power (r = 0.70,
p < 0.001) and average power (r = 0.60, p = 0.002). Paired
t-tests revealed the average value for peak power was signifi-
cantly greater in the WANT compared with the RAST
(t(22) = 11.570, p < 0.001). Conversely, the average value
obtained when measuring average power was not significantly
different (t(22) = 0.565, p = 0.578) between tests (Figure 2).
Bland and Altman plots illustrating the distribution of the dif-
ference scores between tests for peak and average power are
illustrated in Figure 3.
Figure 2 a) Average power and b) Peak power recorded during the RAST (mean
of RAST 1&2) and WANT. Data are mean ± SD.
Burgess et al. Reliability and validity of the running anaerobic sprint test (RAST) in soccer players 27
Reliability
Relative reliability
The RAST showed very good relative reliability for average
power, ICC = 0.88 (0.74 - 0.95: 95% CI) and good relative reli-
ability for peak power, ICC = 0.72 (0.44 - 0.87: 95% CI) (see
Table 1 for power values and Figure 4 for sprint times).
Absolute reliability
For 95% LOA analyses evidence of heteroscedasticity was
obtained for peak power and average power (r = 0.55, p =
0.007; r = 0.32, p = 0.137, respectively). Log-transformation of
test and retest data substantially reduced estimates of het-
eroscedasticity (r = 0.12, p = 0.650; r = 0.08, p = 0.717 for peak
power and average power, respectively). The 95% LOA for the
log-transformed data are displayed in Table 2. Minimum dif-
ference for identification of a ‘real’ change for average power
was calculated as 83 W for average power and 160 W for peak
power.
Figure 3 Bland and Altman plots illustrating the distribution of the difference scores between tests for peak and average power
Figure 4 Time to complete each of the six RAST sprints
during test 1 and test 2. Data are mean ± SD.
Table 1 Peak power and average power obtained during
test 1 and test 2.
RAST 1 RAST 2
Peak Power (W) 766 ± 106 776 ± 114
Average Power (W) 596 ± 86 584 ± 87
Data are mean ± SD
Table 2 Outcome of limits of agreement analyses for heteroscedastic data.
Variable Bias Lower,
upper 95% LOA
Antilog of
bias
Antilog of lower,
upper 95% LOA
Interpretation of antilog values
Bias Lower, upper 95% LOA
Peak
Power
-0.01 -0.21,
0.18
0.99 0.81,
1.20
Averaged across a group,
test and re-test scores will
be very similar
For an individual a retest score
may range between 0.81 and
1.2 times the original value
Average
Power
0.02 -0.11,
0.15
1.02 0.90,
1.16
Averaged across a group,
test and re-test scores will
be very similar
For an individual a retest score
may range between 0.9 and
1.16 times the original value
Bias = mean of differences in test and re-test scores. The antilog values are dimensionless ratios, where 1 represents equality.
Journal of Trainology 2016;5:24-2928
DISCUSSION
The aim of this study was to investigate the criterion validi-
ty, relative reliability and absolute reliability of the RAST in
amateur soccer players when conducted in an outdoor environ-
ment. The results have shown that criterion validity of the
RAST was strong for peak power (r = 0.70, p < 0.001) and
average power (r = 0.60, p = 0.002); however, the RAST signif-
icantly underestimated peak power in comparison to the
WAnT. The RAST showed very good relative reliability for
average power, ICC = 0.88 (0.74 - 0.95: 95% CI) and good rel-
ative reliability for peak power, ICC = 0.72 (0.44 - 0.87: 95%
CI). Assessment of absolute reliability highlighted that
although when averaged across a whole group test and re-test
scores will be very similar (for both average and peak power),
when used to monitor individuals an individual’s retest score
for peak power may range between 0.81 and 1.2 times the
original value and an individual’s retest score for average
power may range between 0.9 and 1.16 times the original
value.
Previously reported individual scores from the RAST have
ranged from 367-1092 W for peak power and 319-927 W for
average power.9-11,17,27 In addition, previous research has
reported large ranges in group peak power (599-810 W) and
average power (451-665 W) values.9-11,17,28 This current study’s
group means for peak and mean power lie within these ranges
(781 ± 121 W and 591 ± 85 W). Similarly, the values obtained
here for soccer player’s peak and average power during the
WAnT are comparable to values previously reported. Typical
values of group means for soccer players measured during a
WAnT are in the order of 740-860 W for peak power and 350-
700 W for average power10,17,29,30, with individual values for
both variables varying from 400-1434 W10 and 218-900 W10,29.
The values obtained for the RAST test are in line with those
previously reported from indoor environments. Although ‘out-
door’ testing is often considered to be less reliable due to
potentially fluctuating conditions the conditions remained rela-
tively stable throughout this studies testing period and hence
resulted in relative reliability scores which fell within the
range of those previously reported. This would support the
idea that as long as the environment remains relatively stable
outdoor testing can be just as reliable as indoor testing.
The significant correlations for both peak and mean power
(r = 0.70, p < 0.001; r = 0.60, p = 0.002) reported in the present
study between the RAST and WAnT correspond with correla-
tion values reported elsewhere. Previous studies have reported
significant r values of 0.46-0.90 for peak power and 0.53-0.98
for average. To the authors’ knowledge the only previous study
conducted with soccer players performing the RAST in an out-
door environment is that by Keir et al.17 who found no signifi-
cant correlations between peak and mean power in the RAST
and WAnT tests which contradicts the findings reported here.
However, the study by Keir et al.17 was conducted on only
eight participants and they did not include an unloaded accel-
eration phase immediately prior to the initiation of the 30 s
Wingate tests which is contrary to the standard protocol6. The
results presented here also show that whilst power produced
during the RAST is related to power production during the
WAnT, comparatively the RAST significantly underestimates
peak power values. This finding is similar to that reported by
Zagatto et al.11 who also found that the RAST produced signif-
icantly lower peak power scores than the WAnT. A potential
explanation of this finding is that peak power in the WAnT is a
more instantaneous measure of power (1 s average) which usu-
ally occurs in the first three seconds of the test following the
unloaded acceleration phase. Conversely, for the RAST peak
power is determined from the fastest sprint which equates to
an average power for an approximately 5-6s time period. The
RAST also includes an element of acceleration with the stand-
ing start and therefore an average value will be reduced by this
initial low velocity period. However, when this acceleration
phase was also removed by Keir et al.17 during the WAnT they
still found the WAnT test to produce significantly higher peak
power than the RAST.
The relative reliability of the RAST reported here lie within
the range previously reported for both average power (ICCs
between 0.72 and 0.97) and peak power (ICCs between 0.58
and 0.92).8-11 The major novel finding of this study is the abso-
lute reliability statistics associated with this test demonstrating
the high random variability that exists and that the variation is
heteroscedastic. Heteroscedasticity is a relatively common fea-
ture of test-retest scores in sport and exercise science but has
implications in detecting real changes in an athlete’s fitness,
particularly those with already well developed attributes and
high test scores. In the current study heteroscedasticity was
reduced by applying a log-transformation and once the anti-
logs were taken, to return values to the original scale, the 95%
LOA represented ratio limits of agreement. That is, instead of
the potential change in a retest score being calculated by add-
ing and subtracting a total error value, the potential change is
found by multiplying by an error value. For example, If a new
athlete representative of the population studied here were to
perform the RAST and obtain an average power score of
440 W, we expect with approximate 95% probability that the
second score will be between the range of 396-510 W (e.g. 440
× 0.9 = 396 W and 440 × 1.16 = 510 W, with error values taken
from the antilog of lower, upper 95% LOA in Table 2). The
range in these possible retest values may be considered by
some already to be too large and therefore unacceptable.
However, for an athlete with higher power production the 95%
LOA for a retest value will be even wider due to the heterosce-
dastic nature of the data. If, for example, the second athlete
obtained an average power score of 680 W, there is an approxi-
mate 95% probability that the second score will be between
the range of 612-788 W. For peak power the potential range of
scores that could be expected on a retest are shown here to be
even greater than average power, with changes of up to
approximately 20% for peak power within the 95% probability
level. It is clear that with the above criteria the RAST would
be of limited use in detecting changes in the majority of pre-
post training designs used by coaches. One potential strategy
to mitigate this problem is to incorporate narrower LOA, and
therefore, the magnitude required of any change in test score
to be considered indicative of altered ability will be lower.
However, with this approach the probability that a different
Burgess et al. Reliability and validity of the running anaerobic sprint test (RAST) in soccer players 29
test score will incorrectly be considered a true change is
increased. A less abstract and more readily understood practice
is to calculate the minimum difference statistic which provides
a single minimum value for all players which a test score must
change by after a period of training to reflect a ‘real’ change.
The results from the study suggest that individuals would have
to increase their average power by at least 83 W and their peak
power by at least 160 W in the RAST test to be confident that
the difference reflected a training related increase in muscular
power. These values equate to approximately 14% and 27% of
the average values produced, further illustrating the low sensi-
tivity of the RAST test to detect training related changes.
CONCLUSIONS
In conclusion the findings of this study have shown that the
RAST is a relatively reliable, practicable field based test which
can be used by coaches to estimate their soccer player’s level
of average anaerobic power. However the test is not sensitive
enough to detect individual changes below approximately 15
to 20% and is therefore not recommended to be used to contin-
ually monitor individual performance. The test can, however,
be used to monitor the anaerobic power of a team as a whole.
In addition if true peak power measures are required this test
also has limitations.
ACKNOWLEDGEMENTS
To the authors knowledge there are no conflicts of interest
concerning this manuscript.
REFERENCES
1. Bangsbo J, Mohr M, Krustrup P. Physical and metabolic demands of
training and match-play in the elite football player. J Sports Sci
2006;24:665-674.
2. Reilly T, Thomas V. A motion analysis of work-rate in different positional
roles in professional football match-play. J Hum Mov Stud 1976;2:87-97.
3. Stølen T, Chamari K, Castagnaand C et al. Physiology of Soccer An
Update. Sports Med 2005;35:501-536.
4. Mohr M, Krustrup P, Bangsbo J. Match performance of high-standard
soccer players with special reference to development of fatigue. J Sports
Sci 2003;21:519–528.
5. Draper N, Whyte G. Anaerobic performance testing. Peak Perform
1997;97:3-5.
6. Bar-Or O. The Wingate anaerobic test: An update on methodology,
reliability and validity. Sports Med, 1987;4:381-394.
7. Margaria R, Aghemo P, Rovelli E. Measurement of muscular power
(anaerobic) in man. J Appl Physiol 1966;221:1662-1664.
8. Hodson B, Jones MA. The effect of caffeine ingestion on repeated sprint
performance. J Sports Sci 2004;22:281.
9. BalčiūnasM,StonkusS,AbrantesCetal.LongTermEffectsof Different
Training Modalities on Power, Speed, Skill and Anaerobic Capacity in
Young Male Basketball Players. J Sports Sci Med 2006;5:163-170.
10. Reza AB, Rastegar M. Correlation between Running-based Anaerobic
Sprint Test (RAST) field tests, Sargent jump and 300 yard shuttle run tests
with laboratory anaerobic Wingate test in evaluation of indoor soccer
player’s anaerobic readiness. Ann Bio Res 2012;3:377-384.
11. Zagatto AM, Beck WR, Gobatto CA. Validity of the running anaerobic
sprint test for assessing anaerobic power and predicting short-distance
performances. J Strength Cond Res 2009;23:1820–1827.
12. Atkinson G, Nevill A. Selected issues in the design and analysis of sport
performance research. J Sports Sci, 2001;19:811-827.
13. Cooper SM, Baker JS, Eaton ZE et al. A simple multistage field test for the
prediction of anaerobic capacity in female games players. Br J Sports Med
2004;38:784-789.
14. Haj-Sassi R, Dardouri W, Gharbi Z et al. Reliability and validity of a new
repeated agility test as a measure of anaerobic and explosive power. J
Strength Cond Res 2011;25:472-480.
15. Nummela A, Alberts M, Rijntjes RP et al. Reliability and validity of the
maximal anaerobic running test. Int J Sports Med 1996;17:S97-S102.
16. Abbasian S, Golzar S, Onvani V et al. The predict of RAST test from
WANT test in elite athletes. Res J Recent Sci 2012;1:72-75.
17. Keir DA, Theriault F, Serresse O. Evaluation of the Running-Based
Anaerobic Sprint Test as a Measure of Repeated Sprint Ability in
Collegiate-Level Soccer Players. J Strength Cond Res 2013;27:1671-1677.
18. Zacharogiannis E, Paradisis G, and Tziortzis S. An Evaluation of Tests of
Anaerobic Power and Capacity. Med Sci Sports Exerc 2004;36:S116.
19. British Army (2013). Physical training and tests. Retrieved from http://
www.army.mod.uk
20. US army (2013). Army Physical Readiness Training (PRT) (FM 3-22.20)
Retrieved from http://www.army.mil/
21. Atkinson G, Nevill A. Statistical methods for assessing measurement error
(reliability) in variables relevant to sports medicine. Sports Med
1998;26:217–238.
22. Svensson M, Drust B. Testing soccer players J Sports Sci 2005;23:601-
618.
23. Glaister M, Howatson G, Pattison JR et al. The reliability and validity of
fatigue measures during multiple-sprint work: an issue revisited. J Strength
Cond Res 2008;22:1597–1601.
24. Evans JD. Straightforward statistics for the behavioural sciences.
California, Brooks/Cole Publishing, 1996.
25. Altman DG. Inter-rater agreement, Practical statistics for medical research,
London, Chapman and Hall, 1991.
26. Bland J, Altman D. Statistical methods for assessing agreement between
two methods of clinical measurements. Lancet 1986;327:307–310.
27. Weir JP. Quantifying test-retest reliability using the intraclass correlation
coefficinet and the SEM. J Strength Cond Res 2005;19:232-240.
28. Kaminagakura EI, Zagatto AM, Redkva PE et al. Can the Running-Based
Anaerobic Sprint Test be used to Predict Anaerobic Capacity? J Exerc
Physiol online, 2012;15:90-99.
29. Miller DK, Kieffer HS, Kemp HE et al. Off-season physiological profiles
of elite National Collegiate Athletic Association Division III male soccer
players. J Strength Cond Res 2011;25:1508-1513.
30. Popadic Gacesa JZ, Barak OF, Grujic NG. Maximal anaerobic power test
in athletes of different sport disciplines. J Strength Cond Res 2009;23:751-
755.