This article was downloaded by: [ ] On: 27 October 2012, At: 13:09 Publisher: Routledge Informa Ltd Registered

in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Sports Sciences

Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rjsp20

Concurrent validity and testretest reliability of a global positioning system (GPS) and timing gates to assess sprint performance variables
Mark Waldron , Paul Worsfold , Craig Twist & Kevin Lamb
a a a a a

Department of Sports and Exercise Sciences, University of Chester, Chester, UK

Version of record first published: 18 Oct 2011.

To cite this article: Mark Waldron, Paul Worsfold, Craig Twist & Kevin Lamb (2011): Concurrent validity and testretest reliability of a global positioning system (GPS) and timing gates to assess sprint performance variables, Journal of Sports Sciences, 29:15, 1613-1619 To link to this article: http://dx.doi.org/10.1080/02640414.2011.608703

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Journal of Sports Sciences, December 2011; 29(15): 16131619

Concurrent validity and testretest reliability of a global positioning system (GPS) and timing gates to assess sprint performance variables

MARK WALDRON, PAUL WORSFOLD, CRAIG TWIST, & KEVIN LAMB

Department of Sports and Exercise Sciences, University of Chester, Chester, UK (Accepted 26 July 2011)

Downloaded by [ ] at 13:09 27 October 2012

Abstract There has been no previous investigation of the concurrent validity and reliability of the current 5 Hz global positioning system (GPS) to assess sprinting speed or the reliability of integrated GPSaccelerometer technology. In the present study, we wished to determine: (1) the concurrent validity and reliability of a GPS and timing gates to measure sprinting speed or distance, and (2) the reliability of proper accelerations recorded via GPSaccelerometer integration. Nineteen elite youth rugby league players performed two over-ground sprints and were simultaneously assessed using GPS and timing gates. The GPS measurements systematically underestimated both distance and timing gate speed. The GPS measurements were reliable for all variables of distance and speed (coefcient of variation [CV] 1.62% to 2.3%), particularly peak speed (95% limits of agreement [LOA] 0.00 + 0.8 km h71; CV 0.78%). Timing gates were more reliable (CV 1% to 1.54%) than equivalent GPS measurements. Accelerometer measurements were least reliable (CV 4.69% to 5.16%), particularly for the frequency of proper accelerations (95% LOA 1.00 + 5.43; CV 14.12%). Timing gates and GPS were found to reliably assess speed and distance, although the validity of the GPS remains questionable. The error found in accelerometer measurements indicates the limits of this device for detecting changes in performance.

Keywords: Speed assessment, accelerometers, comparison of methods, GPS

Introduction Assessment of sprint performance is commonly undertaken using infra-red timing gates, which provide reliable measurements of sprint performance over short (10 m) intervals (Cronin & Templeton, 2008; Duthie, Pyne, Ross, Livingstone, & Hooper, 2006). Due to logistical practicalities, the alternative application of global positioning systems (GPS) for such a purpose remains an attractive prospect, yet recent reports have questioned the reliability and validity of GPS devices (Dufeld, Reid, Baker, & Spratford, 2010; Gray, Jenkins, Andrews, Taaffe, & Glover, 2010; Jennings, Cormack, Coutts, Boyd, & Aughey, 2010; Macleod, Morris, Nevill, & Sunderland, 2009; Petersen, Pyne, Portus, & Dawson, 2009; Portas, Rush, Barnes, & Batterham, 2007). However, the generalization of such ndings is limited, since the validity and reliability of a GPS device may be affected by both the manufacturer and the sampling rate, which often differ between studies (Dufeld et al., 2010; Petersen et al., 2009). Indeed, the reliability of the current device (GPSports,

SPI-Pro, 5 Hz, Canberra, Australia) in measuring speed is yet to be evaluated, providing limited justication for studies utilising such equipment to assess sprinting activity (McLellan, Lovell, & Gass, 2010, 2011; Waldron, Twist, Daniels, & Worsfold, 2010). Furthermore, there are no published data on the agreement (concurrent validity) between the current GPS device and timing gates for the assessment of linear sprinting speed. The relatively recent advancement in micro electrical mechanical systems (MEMS) has also supported the integration of GPS technology with in-built tri-axial accelerometers (100 Hz). The triaxial accelerometer, housed within the current GPSports models (SPI-Pro; 5 Hz, Canberra, Australia), measures a composite vector magnitude (expressed as g-force, the acceleration relative to freefall) by recording the sum of proper accelerations measured in three separate orthogonal axes (anterioposterior [x], mediolateral [y], and vertical [z]). However, to date, the testretest reliability of integrated accelerometry within current GPS devices is unknown, providing users with limited information

Correspondence: M. Waldron, Department of Sports and Exercise Sciences, University of Chester, Parkgate Road, Chester CH1 4BJ, UK. E-mail: m.waldron@chester.ac.uk ISSN 0264-0414 print/ISSN 1466-447X online 2011 Taylor & Francis http://dx.doi.org/10.1080/02640414.2011.608703

1614

M. Waldron et al. coach, consisting of moderate-intensity movement, dynamic stretching procedures, and two maximal sprints. The experimental protocol consisted of two maximal sprint efforts, starting from a standing position, over a at, grass surface, separated by 3 min of passive recovery. The sprinting course was marked with a pre-measured (tape measure) straight painted line upon which timing gates (Brower Timing Systems, Draper, UT) were positioned at 10 m intervals (0 m to 30 m), at a height of 60 cm (Cronin & Templeton, 2008). Participants were instructed to sprint from 30 cm behind the rst timing gate, starting from their preferred foot. Participants were further instructed to begin sprinting from a static position, preventing pre-emptive backwards movement (Frost, Cronin, & Levin, 2008). The initiation of a sprint was determined by a continuous increase in speed from below 0 to 0.1 km h71 (Petersen et al., 2009). Split times were recorded at 10 m, 20 m, and 30 m from a wireless receiver accurate to 0.01 s. All data were analysed using Team AMS software (v.2.1, GPSports, Canberra, Australia). Concurrent validity Concurrent validity was determined by comparing mean speed (km h71) at 10 m, 20 m, and 30 m and moving speed between 10 m and 20 m measured by the timing gates with values recorded using the GPS devices (derived using Team AMS software). The time interval recorded by the timing gate was used to truncate the speed data recorded by the GPS. Criterion validity was obtained by comparing distance at 10 m, 20 m, and 30 m (quantied using a tape measure) to that recorded by the GPS devices. Estimated speed derived from the timing gates was calculated by dividing the displacement of the participant (measured distance) by the time taken to travel the given distance. Reliability Testretest reliability of both the timing gates and the GPS device to measure distance and mean speed at 10 m, 20 m and 30 m, and moving speed between 10 m to 20 m over the two repeated trials was assessed. Additional values of peak speed at 30 m, peak proper acceleration (g), and frequency of proper accelerations above 5 g over 10 m and 30 m were also assessed for testretest reliability. Peak proper acceleration was dened as the largest positive acceleration recorded during the interval in question. The value of 5 g was applied since a similar threshold value of 5 g has been used in previous studies assessing impacts during match-play in rugby union (Cunniffe, Proctor, Baker, & Davies, 2009). Furthermore, using an

regarding the sensitivity of current GPS technology to identify proper accelerations. Therefore, the aims of the current study were to determine the concurrent validity and testretest reliability of both a non-differential GPS device (5 Hz) and timing gates to measure selected variables of distance, speed, and proper acceleration during over-ground sprinting at various intervals.

Methods Participants Nineteen elite youth male rugby league players (age 14.7 + 0.45 years; stature 176.5 + 6.5 cm; body mass 72.8 + 10.7 kg) and their parent/guardian gave consent to participate in the present study. All participants were asked not to exercise on the day of testing and to follow their normal dietary guidelines. Ethical approval for the present study was received from the Faculty of Applied Health Sciences Ethics Committee. Design and procedure One week before testing, participants completed a habituation protocol, comprising three maximal sprint efforts at each sprint interval (10 m, 20 m, and 30 m), separated by 3 min of passive recovery. Suitable habituation to the test protocol was established when no systematic learning effect was evident and when participants voluntarily indicated that the procedure for testing was understood. To prevent participants from decelerating before reaching the nal timing gate (30 m), thus failing to reach or maintain true peak speed through 30 m, an additional coloured cone was placed at 35 m to which participants were instructed to consider as the nish point. Using the GPS equipment for verication, no indication of pre-emptive deceleration was demonstrated, with all participants consistently attaining peak speed after the 30 m distance. Testing took place across the centre of an open eld, free from obstruction or adjacent buildings. All GPS units (GPSports, SPI-Pro, 5 Hz, Canberra, Australia) were simultaneously activated and left for 15 min. The GPS units (size 90 6 45 6 5 mm; mass 86 g) sample at a rate of 5 Hz and are coupled with a 6 g tri-axial accelerometer sampling at 100 Hz. The typical number of available satellite signals ranged between 9 and 11 accompanied by a mean horizontal dilution of position (HDOP) of 1.2 + 0.2 throughout the test period. A GPS vest (GPSports, Canberra, Australia) was tightly tted to each participant, placing the receiver between the scapulae. Twenty minutes before exercise, players were taken for a structured warm-up led by the squad

Downloaded by [ ] at 13:09 27 October 2012

Concurrent validity and reliability of GPS accelerometer tted to the lower spinal region of treadmill runners, values of 5 g in a vertical axis have been previously demonstrated (Bhattacharya, McCutcheon, Shvartz, & Greenleaf, 1980). Statistical analyses Concurrent validity and reliability were assessed using 95% limits of agreement (95% LOA; Bland & Altman, 1986), coefcient of variation (CV; Atkinson & Nevill, 1998), and ratio limits of agreement (95% ratio LOA; Bland & Altman, 1986) for all variables. Data were initially checked for normality of differences using the Shapiro-Wilk test (P 4 0.05). Further checks for heteroscedastic errors were carried out using Pearson product moment correlation (r-value). Paired samples t-tests were used to calculate differences (biases) between means of measurement methods (validity) and to compare measurements between trials (reliability). Statistical signicance was set at P 5 0.05 for all variables. The strength of the CV (510%) was quantied in accordance with Atkinson and Nevill (1998). Data are reported as means and standard deviations (s) throughout and analysed using SPSS (v.16; SPSS Inc., Chicago, IL).

1615

case (10 m sprint distance; 95% LOA 1.10 6/7 1.29). However, using the CV method, these same results were reported as reliable, with results consistently below 10% (CV 4.81% to 8.06%). For all speed variables, there were signicant differences (P 5 0.05) between timing gate and GPS values (Table I). Mean biases ranged from 2.01 km h71 to 2.19 km h71, with random errors of 2.18 km h71 to 3.62 km h71, which, in one instance, equated to a total error of 5.67 km h71 (10 m sprint; 2.05 /7 3.62 km h71). Values ranging from 5.68% to 9.81% (CV) were evident for all speed measures. The closest level of agreement for both speed (95% LOA 2.03 + 2.85 km h71; 95% ratio LOA 1.08 6/7 1.12) and distance (95% LOA 0.47 + 1.34 m; 95% ratio LOA 1.05 6/7 1.15) were found during the moving 10 m sprint. Reliability GPS measurements of speed, distance, and proper acceleration demonstrated no signicant differences (P 4 0.05) for any of the measured variables between the rst and second trial. As shown in Table II, using the 95% LOA method, mean differences between trials ranged from 0.04 m to 0.23 m for distance and from 0 km h71 to 0.09 km h71 for speed. Measurement of peak speed demonstrated the greatest reliability (95% LOA 70.00 + 0.8 km h71; 95% ratio LOA 1.00 6/7 1.03; CV 0.78%). Table II shows the results from the speed assessment using timing gates. Small mean biases (70.01 km h71 to 70.14 km h71) for all sprint intervals and random errors between test 1 and 2 ranging from 0.56 to 1.64 km h71 were demonstrated. Using the CV method, all sprint intervals showed high levels of reliability (1% to 1.54%). The largest errors were found in the moving 10 m sprint.

Downloaded by [ ] at 13:09 27 October 2012

Results Concurrent validity The 95% LOA showed a systematic underestimation (P 5 0.05) of the GPS against that of the timing gates for all variables of distance (Table I). Heteroscedastic errors were present for all measures of distance (r 71; P 5 0.05) and 10 m moving speed (r 0.79; P 5 0.05). For these variables only, the 95% ratio LOA showed systematic biases ranging from 1.05 to 1.12 and random errors ranging from 1.11 to 1.29, translating to a 39% total error in the worst

Table I. Validity of measured distance and timing gate speed against GPS measurements (n 19). Measured distance/timing gate (mean + s) Distance (m) 10 m sprint 20 m sprint 30 m sprint Moving 10 m Speed (km h71) 10 m sprint 20 m sprint 30 m sprint Moving 10 m GPS measure (m) (mean + s) CV (%) 95% LOA 95% ratio LOA

10 + 0 20 + 0 30 + 0 10 + 0 16.52 + 1.19 20.48 + 1.15 22.73 + 1.23 27.02 + 1.20

9.07 + 1.11a 17.91 + 1.66a 27.98 + 1.45a 9.52 + 0.68a 14.46 + 1.94a 18.28 + 1.66a 20.72 + 1.43a 24.98 + 1.97a

8.06 8.09 5.00 4.81 9.81 8.54 6.61 5.68

2.05 + 3.62 2.19 + 3.34 2.01 + 2.18

1.10 6/7 1.12 6/7 1.07 6/7 1.05 6/7 1.15 6/7 1.12 6/7 1.09 6/7 1.08 6/7

1.29 1.21 1.11 1.15 1.30 1.20 1.11 1.12

Note: CV coefcient of variation; 95% LOA 95% limits of agreement; 95% ratio LOA 95% ratio limits of agreement. aSignicantly different (P 5 0.05) from measured distance or timing gate speed. Moving 10 m speed between timing gate 2 and 3 during the 30 m sprint.

1616

M. Waldron et al. of the 95% ratio term revealed a total error reaching 39% (1.10 6/7 1.29; CV 8.09%) for 10 m distance and a 45% total error for 10 m speed (1.15 6/ 7 1.30; CV 9.81%). However, similar method comparisons using 1 Hz GPS devices have previously reported coefcients of variation of 1% (Portas et al., 2007) or signicant overestimations of mean speed, particularly during test protocols involving directional changes (Macleod et al., 2009). Such ndings are inconsistent with the results of the present study where a 5 Hz model was used (CV range 5.68% to 9.81%). The differences between previous investigations and the current analysis may be due to differences in test procedures or disparities in sampling rate. It is also possible that the custom algorithms integrated within the newer 5 Hz devices may account for the clear differences in the level of agreement between methods. The notable trend for the current improvements in agreement as the distance and speed of the sprint increased is consistent with previous ndings (Gray et al., 2010; Petersen et al., 2009). Petersen and colleagues reported agreements of 5.4 + 2%,

As shown in Table III, proper accelerations ranging from 5.1 g to 12.52 g were found over sprint intervals of both 10 m and 30 m. Table III shows the average values found for 10 m peak proper acceleration (7.32 + 2.25 g to 7.03 + 1.82 g) and 30 m peak proper acceleration (8.28 + 1.81 g to 8.33 + 1.56 g), both of which displayed a CV of over 5%. While no signicant differences were found between test 1 and 2 for any of the measured values, random errors ranging from 2.3 g to 1.73 g were recorded for 10 m and 30 m sprints, respectively. The frequency of proper accelerations 45 g for the 30 m sprint demonstrated the greatest parity between trials (CV 14.12%), equating to a potential 93% total error. Discussion In accordance with previous research (Dufeld et al., 2010; Petersen et al., 2009), the results of the current analysis demonstrate the systematic underestimation of GPS measurements compared with measured distances and timing gate calculations of speed at all measured intervals. In the present study, application

Downloaded by [ ] at 13:09 27 October 2012

Table II. Reliability of sprinting distance and speed measured by the GPS and timing gates (n 19). Test 1 (mean + s) Distance (m) 10 m sprint 20 m sprint 30 m sprint Moving 10 m GPS speed (km h71) 10 m sprint 20 m sprint 30 m sprint Moving 10 m Overall peak speed Timing gate speed (km h71) 10 m sprint 20 m sprint 30 m sprint Moving 10 m Test 2 (mean + s) CV (%) 95% LOA 95% ratio LOA

9.0 + 1.16 17.90 + 1.66 27.98 + 1.45 9.52 + 0.68 14.46 + 1.94 18.28 + 1.66 20.72 + 1.43 24.98 + 1.97 29.48 + 1.56 16.52 + 1.19 20.48 + 1.15 22.73 + 1.23 27.02 + 1.20

9.04 + 0.93 18.14 + 1.68 28.06 + 1.44 9.58 + 0.61 14.41 + 1.74 18.34 + 1.74 20.81 + 1.43 25.00 + 1.87 29.47 + 1.48 16.53 + 1.18 20.6 + 0.9 22.87 + 1.03 27.15 + 1.47

1.99 2.06 1.84 2.30 2.06 1.92 2.02 1.62 0.78 1.13 1.00 1.35 1.54

0.04 + 0.72 70.23 + 1.42 70.08 + 1.75 70.06 + 1.13 0.05 + 1.05 70.05 + 1.17 70.09 + 0.84 70.02 + 1.35 0.00 + 0.8 70.01 + 0.54 70.12 + 1.09 70.14 + 0.94 70.12 + 1.64

1.00 6/7 0.98 6/7 0.99 6/7 0.99 6/7 1.00 6/7 0.99 6/7 0.99 6/7 0.00 6/7 1.00 6/7 0.99 6/7 0.99 6/7 0.99 6/7 0.99 6/7

1.08 1.08 1.06 1.13 1.07 1.06 1.07 1.05 1.03 1.04 1.05 1.04 1.06

Note: CV coefcient of variation; 95% LOA 95% limits of agreement; 95% ratio LOA 95% ratio limits of agreement. Moving 10 m speed between timing gate 2 and 3 during the 30 m sprint.

Table III. Peak proper accelerations and frequency of proper accelerations using integrated accelerometry (n 19). Test 1 Accelerometer (mean + s) Test 2 Accelerometer (mean + s) CV (%) Peak acceleration (g) 10 m sprint 30 m sprint Frequency (45 g) 10 m sprint 30 m sprint 95% LOA 95% ratio LOA

7.32 + 2.25 8.28 + 1.81 6.84 + 2.98 14.26 + 5.24

7.03 + 1.82 8.33 + 1.56 6.63 + 3.11 13.26 + 4.62

5.01 5.16 4.69 14.12

0.28 + 2.30 1.02 6/7 1.30 70.05 + 1.73 0.99 6/71.22 0.21 + 1.68 1.04 6/7 1.24 1.00 + 5.43 1.02 6/7 1.91

Note: CV coefcient of variation; 95% LOA 95% limits of agreement; 95% ratio LOA 95% ratio limits of agreement.

Concurrent validity and reliability of GPS 4.2 + 1.5%, and 2.9 + 1.1% (standard error estimate) for sprints exceeding 18 km h71 at increasing distances of 20 m, 30 m, and 40 m, respectively. The reason for such a trend in results may be linked to the deciencies of a 5 Hz sampling rate for assessing intense periods of sprint performance, where rapid changes in linear movement limit the opportunity for positional measures to be recorded. In the present study, the 10 m moving speed demonstrated greater validity for both distance (95% ratio LOA 1.05 6/ 7 1.15; CV 4.81%) and speed (95% ratio LOA 1.08 6/7 1.12; CV 5.68%) compared with any other measure, even though this was equivalent to the shortest distance assessed. As in the study of Jennings et al. (2010), removal of the accelerative period from the sampled performance appeared to reduce the limitations of the 5 Hz sampling frequency. This might be due to the established stability in gait cycle, and a correspondingly lower alteration in the rate of locomotion compared with measurements taken from a standing start. Previous investigations using dual beam timing gates to assess sprint speed have found typical errors between 1% and 2% over 10 m or 20 m distances (Cronin & Templeton, 2008; Duthie et al., 2006; Gabbett, Kelly, & Sheppard, 2009; Moir, Button, Glaister, & Stone, 2004), which are comparable to the values observed in the present study (CV 1% to 1.54%). Reliability was strongest for 10 m and 20 m sprint speed (CV 1% to 1.13%), which may be a result of the participants familiarity with sprint intervals of this magnitude during seasonal conditioning and competition (Waldron et al., 2010). Timing gate reliability was poorer for mean speed at the 30 m interval compared with 10 m and 20 m distances, whereas mean speed measured by the GPS demonstrated an improvement in reliability over increasing distances. Such results may be a consequence of the increased opportunity for participants to deviate away from the assumed linear pathway over greater distances, thus inuencing the timing gate reliability over repeated trials. Previous evidence may support this supposition since, even in conjunction with repeatable horizontal velocity during sprinting movements (CV 1%), greater variability (CV 9% to 34.6%) in centre of mass displacement can be expected between trials (Hunter, Marshall, & McNair, 2004). Therefore, the subsequent accumulation of lateral centre of mass displacement over prolonged movement pathways is to be expected and may be recognized using GPS devices, but not when using the timing gate method. The reliability of the GPS over all variables ranged from 0.78% to 2.3% (CV), with peak speed over 30 m providing the most reliable measure (CV 0.78%; 95% LOA 0.00 + 0.8; 95% ratio LOA 1.00 6/7 1.03). It should be noted that such results

1617

are similar to those previously reported (Petersen et al., 2009) using an identical GPS model, yet using GPS models from different manufacturers appears to show much larger variation in recorded distances (Jennings et al., 2010). There are currently no published data on the absolute testretest reliability of peak speed using GPS over repeated trials, yet this value is often strongly related to speed from timing lvarez, Coutts, gates in validation trials (Barbero-A lvarez, & Castagna, 2010; Coutts Granda, Barbero-A & Dufeld, 2010). It was apparent in the GPS measurements that the true peak speed of each participant was never attained during the present analysis, potentially owing to the short 30 m distances used in the current experimental protocol. Indeed, intervals of between 50 m and 60 m have been suggested as optimal distances for the attainment of peak speed (Young, Benton, & Duthie, 2001). Consequently, the ability to yield such reliable results over repeated trials may be aided since participants were able to reach a sub-maximal peak speed with greater ease than true peak speed. It is possible that sampling peak speed attained over a longer distance may have produced less reliable results. Nevertheless, a total error of 0.8 km h71 provides a useful variable to consider for detection of minor changes in performance. The present results showed that timing gate reliability was superior to GPS reliability. Since the timing gate measures are comparable to previous reports, and that these data were collected simultaneously with GPS, it appears that the GPS measurements are likely to be subject to small technical errors inherent in the measurement tool rather than biological variations (Hopkins, 2000). It is suggested that consideration of analytical goals be made by researchers before selecting timing gates or GPS to measure speed (Atkinson & Nevill, 1998). Based upon the present results from the 95% ratio LOA, to detect a 5% change in 10 m sprint performance using the current GPS model, a sample size of 10 may be used (Atkinson, Nevill, & Edwards, 1999). The current study demonstrated the potential for large variations in the frequency of proper accelerations, particularly over 30 m intervals (CV 14.12%; 95% LOA 1.00 + 5.43 g). The large differences found in the current study between repeated trials provide a poor forecast for researchers attempting, for example, to differentiate between positional groups during competitive team sports using such variables. Based upon the degree of difference required to distinguish between positional groups (*38.4%; Cunniffe et al., 2009), it can be expected that the 91% total error currently found over 30 m sprints would not tolerate such uses and are likely to be beyond the degree of acceptable error for most measurements.

Downloaded by [ ] at 13:09 27 October 2012

1618

M. Waldron et al. References

Atkinson, G., & Nevill, A. M. (1998). Statistical methods for addressing measurement error (reliability) in variables relevant to sports medicine. Sports Medicine, 26, 217238. Atkinson, G., Nevill, A. M., & Edwards, B. (1999). What is an acceptable amount of measurement error? The application of meaningful analytical goals to the reliability of sports science measurements made on a ratio scale. Journal of Sports Sciences, 17, 18. lvarez, J. C., Coutts, A. J., Granda, J., Barbero-A lvarez, Barbero-A V., & Castagna, C. (2010). The validity and reliability of a global positioning satellite system device to assess speed and repeated sprint ability (RSA) in athletes. Journal of Science and Medicine in Sport, 13, 232235. Bhattacharya, A., McCutcheon, E. P., Shvartz, E., & Greenleaf, J. E. (1980). Body acceleration distribution and O2 uptake in humans during running and jumping. Journal of Applied Physiology, 49, 881887. Bland, J. M., & Altman, D. G. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, i, 307310. Bouten, C. V. C., Koekkoek, K. T. M., Verduin, M., Kodde, R., & Janssen, J. D. (1997). A tri-axial accelerometer and portable data processing unit for the assessment of daily physical activity. IEEE Transactions on Biomedical Engineering, 44, 136147. Coutts, A., & Dufeld, R. (2010). Validity and reliability of GPS units for measuring movement requirements in team sports. Journal of Science and Medicine in Sport, 13, 133135. Cronin, J. B., & Templeton, R. L. (2008). Timing light height affects sprint times. Journal of Strength and Conditioning Research, 22, 318320. Cunniffe, B., Proctor, W., Baker, J. S., & Davies, B. (2009). An evaluation of the physiological demands of elite rugby union using GPS tracking software. Journal of Strength and Conditioning Research, 23, 11951203. Dufeld, R., Reid, M., Baker, J., & Spratford, W. (2010). Accuracy and reliability of GPS devices for measurement of movement patterns in conned spaces for court-based sports. Journal of Science and Medicine in Sport, 15, 523525. Duthie, G. M., Pyne, D. B., Ross, A. A., Livingstone, S. G., & Hooper, S. L. (2006). The reliability of ten meter sprint time using different starting techniques. Journal of Strength and Conditioning Research, 20, 246251. Frost, D. M., Cronin, J. B., & Levin, G. (2008). Stepping backwards can improve sprint performance over short distances. Journal of Strength and Conditioning Research, 22, 918922. Gabbett, T. J., Kelly, N. J., & Sheppard, J. M. (2009). Speed, change of direction speed and reactive agility of rugby league players. Journal of Strength and Conditioning Research, 22, 174 181. Gray, A. J., Jenkins, D., Andrews, M. H., Taaffe, D. R., & Glover, M. L. (2010). Validity and reliability of GPS for measuring distance travelled in eld-based team sports. Journal of Sports Sciences, 28, 13191325. Hopkins, W. G. (2000). Measures of reliability in sports medicine and science. Sports Medicine, 30, 115. Hunter, J. P., Marshall, R. N., & McNair, P. (2004). Reliability of biomechanical variables of sprint running. Medicine and Science in Sports and Exercise, 36, 850861. Jennings, D., Cormack, S., Coutts, A. J., Boyd, L., & Aughey, R. J. (2010). The validity and reliability of GPS units for measuring distance in team sport specic running patterns. International Journal of Sports Physiology and Performance, 5, 328341. Macleod, H., Morris, J., Nevill, A., & Sunderland, C. (2009). The validity of a non-differential global positioning system for assessing player movement patterns in eld hockey. Journal of Sports Sciences, 27, 121128.

The errors recorded for peak proper acceleration over 10 m and 30 m intervals (CV 5.01% to 5.16%) are larger than previously reported (CV 3.6%) for testretest trials using a hip-tted tri-axial accelerometer to quantify the magnitude of accelerations during over-ground walking (MoeNilssen, 1998). However, it is unclear what level of reliability would be required for performance differences during match-play. Nevertheless, based on arbitrary values of CV 5 10%, the current analysis demonstrates the in-built accelerometer as a reliable tool for measuring the magnitude of proper accelerations. This was particularly noteworthy over 10 m, where less error was apparent, possibly due to the limited opportunity for noise accumulation. The errors found in accelerometer measurements may be confounded by a variety of factors. First, accelerometer measurements may be inuenced by changes in body orientation. Although the use of a tri-axial accelerometer may be thought to negate such an issue, accelerometers vary in sensitivity based upon the axis of measurement, often demonstrating reduced accuracy while measuring in a horizontal axis (Bouten, Koekkoek, Verduin, Kodde, & Janssen, 1997; Wong & Wong, 2008). Given the likelihood of trunk inclinations during maximal sprinting, a disproportionate recording of proper acceleration could be expected when attempting to quantify whole body proper acceleration. Second, the magnitude of the signal obtained by the accelerometer is affected by the stability of the mounting surface and can be sensitive to changes in the integrity of underlying support (Sinclair, Bottoms, Taylor, & Greenalgh, 2010). The GPS vest supplied by the manufacturer does not support the accelerometer housing in the manner often specied for accelerometry purposes (Welk, 2002). Subsequently, the resultant signal received by the accelerometer may be greatly inuenced by external noise. Conclusion Based on the current evidence, it is suggested that the GPS devices (5 Hz, SPI-Pro) can be used to quantify small, yet practically signicant changes in sprint performance, particularly with reference to measures of peak speed in young rugby players. However, it appears that calculations made using either a GPS device or timing gates can differ markedly. Measures of proper acceleration derived from the integrated accelerometer produced questionable results, particularly for the frequency of proper accelerations. As such, it is recommended that researchers refrain from using the current accelerometers for sensitive experimental purposes.

Downloaded by [ ] at 13:09 27 October 2012

Concurrent validity and reliability of GPS

McLellan, C. P, Lovell, D. I., & Gass, G. C. (2010). Creatine kinase and endocrine responses of elite players pre, during, and post rugby league match play. Journal of Strength and Conditioning Research, 24, 29082919. McLellan, C. P., Lovell, D. I., & Gass, G. C. (2011). Performance analysis of elite rugby league match play using global positioning systems. Journal of Strength and Conditioning Research, 25, 17031710. Moe-Nilssen, R. (1998). Testretest reliability of trunk accelerometry during standing and walking. Archives of Physical Medicine and Rehabilitation, 79, 13771385. Moir, G., Button, C., Glaister, M., & Stone, M. H. (2004). Inuence of familiarisation on the reliability of a vertical jump and acceleration sprinting performance in physically active men. Journal of Strength and Conditioning Research, 18, 276280. Petersen, C., Pyne, D., Portus, M., & Dawson, B. (2009). Validity and reliability of GPS units to monitor cricket-specic movement patterns. International Journal of Sports Physiology and Performance, 4, 381393. Portas, M., Rush, C., Barnes, C., & Batterham, A. (2007). Method comparison of linear distance and velocity measurements with global positioning satellite (GPS) and the timing gate techniques. Journal of Sports Science and Medicine, 6, (suppl. 10), 7.

1619

Sinclair, J., Bottoms, L., Taylor, K., & Greenhalgh, A. (2010). Tibial shock measured during the fencing lunge: The inuence of footwear. Sports Biomechanics, 9, 6571. Waldron, M., Twist, C., Daniels, M., & Worsfold, P. (2010). Match demands, heart rate and perceptual responses of elite rugby league players during competitive matches: A case study. In Proceedings of the British Association of Sports and Exercise Sciences Annual Conference, S67. Welk, G. (2002). Use of accelerometry based activity monitors to assess physical activity. In G. Welk (Ed.), Physical activity assessments for health related research (pp. 125130). Champaign, IL: Human Kinetics. Wong, W. Y., & Wong, M. S. (2008). Detecting spinal posture change in sitting positions with tri-axial accelerometers. Gait and Posture, 28, 168171. Young, W., Benton, D., & Duthie, G. (2001). Resistance training for short sprints and maxim-speed sprints. Strength and Conditioning Journal, 23, 713.

Downloaded by [ ] at 13:09 27 October 2012