See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/40428160

Validity and Reliability of GPS Units to Monitor Cricket-

Specific Movement Patterns

Article in International Journal of Sports Physiology and Performance · September 2009

DOI: 10.1123/ijspp.4.3.381 · Source: PubMed

CITATIONS READS

249 3,972

4 authors, including:

Carl J Petersen David B Pyne

University of Canterbury University of Canberra
64 PUBLICATIONS 1,153 CITATIONS 486 PUBLICATIONS 22,110 CITATIONS

SEE PROFILE SEE PROFILE

Marc Portus
Praxis Performance Group
87 PUBLICATIONS 3,127 CITATIONS

SEE PROFILE

All content following this page was uploaded by David B Pyne on 26 May 2014.

The user has requested enhancement of the downloaded file.

International Journal of Sports Physiology and Performance, 2009, 4, 381-393
© 2009 Human Kinetics, Inc.

Validity and Reliability of GPS Units to

Monitor Cricket-Specific
Movement Patterns
Carl Petersen, David Pyne, Marc Portus, and Brian Dawson

Purpose: The validity and reliability of three commercial global positioning system
(GPS) units (MinimaxX, Catapult, Australia; SPI-10, SPI-Pro, GPSports, Australia)
were quantified. Methods: Twenty trials of cricket-specific locomotion patterns and
distances (walking 8800 m, jogging 2400 m, running 1200 m, striding 600 m, sprint-
ing 20- to 40-m intervals, and run-a-three) were compared against criterion measures
(400-m athletic track, electronic timing). Validity was quantified with the standard
error of the estimate (SEE) and reliability estimated using typical error expressed as a
coefficient of variation. Results: The validity (mean ± 90% confidence limits) for
locomotion patterns walking to striding ranged from 0.4 ± 0.1 to 3.8 ± 1.4%, whereas
for sprinting distances over 20 to 40 m including run-a-three (approx. 50 m) the SEE
ranged from 2.6 ± 1.0 to 23.8 ± 8.8%. The reliability (expressed as mean [90% confi-
dence limits]) of estimating distance traveled by walking to striding ranged from 0.3
(0.2 to 0.4) to 2.9% (2.3 to 4.0). Similarly, mean reliability of estimating different
sprinting distances over 20 to 40 m ranged from 2.0 (1.6 to 2.8) to 30.0% (23.2 to
43.3). Conclusions: The accuracy and bias was dependent on the GPS brand
employed. Commercially available GPS units have acceptable validity and reliability
for estimating longer distances (600–8800 m) in walking to striding, but require fur-
ther development for shorter cricket-specific sprinting distances.

Keywords: accuracy, precision, tracking

Quantifying movement patterns based on game demands is important in

developing sport-specific conditioning programs and tailoring the length of recov-
ery periods from game and training activities. Most commonly, this practice has
been undertaken through time–motion studies conducted in real time with pen,
paper, and stopwatch (notational) or postgame utilizing video recordings with or
without customized computer analysis (digitizing). Time–motion analysis has
been conducted in a number of sports, including basketball,1 field hockey,2 futsal,3
handball,4 ice hockey,5 rugby,6–8 soccer,9,10 taekwon-do,11 and tennis.12 The time–

Petersen and Pyne are with the Physiology Department, Australian Institute of Sport, Belconnen, ACT,
Australia. Portus is with the Sport Science Medicine Unit, Cricket Australia Centre of Excellence,
Albion, QLD, Australia. Dawson is with Human Movement and Exercise Science, University of
Western Australia, Crawley, WA, Australia.

   381
382   Petersen et al

motion analysis process has traditionally been very time intensive as digitizing
takes approximately 8 h to produce data for one player for a whole 80-min rugby
match.13 Although a high degree of validity (~2% typical error) has been reported
for estimation of total distance,6,13 time–motion analyses can also have poor com-
parability between notational and digitizing methods, with one study reporting a
27.5% difference in estimating time spent in work.13
The extended game duration and large playing field in cricket makes it diffi-
cult to conduct time–motion studies with pen and paper or video recordings.
Accordingly, to date there is only one time–motion study examining batting
requirements14 and one study on the fielding requirements of cricket.15 While
these studies14,15 divided the fielding and batting requirements of cricket into com-
ponents (standing, walking, jogging, striding, sprinting, playing a shot, and lateral
motion), the authors only reported temporal data with no estimates of distances
covered in each movement category. With a lack of field markings, accurate mea-
sures of distance are difficult to obtain with some forms of time–motion analy-
sis—hence the need for GPS as opposed to time–motion analysis.
Recently improved miniaturization and enhanced battery life have made
global positioning system (GPS) athlete-tracking units a more convenient, less
time-consuming, and increasingly popular method to quantify movement patterns
and physical demands in sport. Global positioning system technology has been
used to quantify the physiological demands of athletes training and/or competing
in events such as horse racing,16,17 orienteering,18 triathlon,19 Australian football,20
Gaelic football,21 rugby,22 and soccer.23 These studies have employed several dif-
ferent commercially available GPS brands, including Fidelak Equipilot,16 Polar,17
Garman Forerunner,18 GPSports,20,21 and FRWD.23 However, currently in elite
cricket there are only two brands of GPS unit in regular use (MinimaxX and
GPSports). To date, only one study24 using the MinimaxX GPS unit has been
published. A comparison was undertaken of four different tennis drills performed
on a tennis court with six repetitions of 30- and 60-s work periods. The authors
used distance covered and movement speed as measures of intensity, yet surpris-
ingly no reliability data were reported. Another study20 investigated the validity of
the GPSports SPI-10 unit, finding that it overestimated distances by a mean of
4.8%. However, the distances used only ranged from 128 to 1386 m, whereas it is
likely that some cricket players cover distances at least 10-fold greater (~10,000
to 15,000 m). Furthermore, no account of possible variations in the accuracy of
different movement velocities was reported. Coutts and Duffield25 compared dif-
ferent movement velocities between three different GPS models (including the
GPSports SPI-10) from the same manufacturer and reported a typical error (test–
retest reliability) of 5.3% in estimating the distance covered in low-intensity activ-
ity. However, the walking component of each 130-m lap was only 14 m. In cricket,
walking forms up to 90%26 of total movements, and, given that an innings typi-
cally lasts 3.5 h, walking needs to be assessed over an extended duration.
A recent study27 used a 487-m lap repeated 14 times (total circuit distance =
6818 m) with movement patterns simulating field hockey match play to assess
validity of the GPSports SPI Elite 1-Hz GPS unit. The authors used only one unit
and reported the validity of distance and mean speed for four types of hockey-
specific shuttles, total lap distance, and overall total distance. Although significant
differences (P > .01) were found from the criterion distance during the hockey-
 GPS Unit Validity and Reliability   383

specific shuttles (from a mean underestimation of 1.2% to an overestimation of

0.8%), the mean lap distance and total circuit distance were not significantly dif-
ferent from the criterion distance (mean overestimation of 0.04 and 0.06% respec-
tively). What remains unknown from this investigation is the validity and reliabil-
ity of the individual movement patterns (walking, jogging, cruise, sprint)
comprising the hockey simulation circuit.
Although the interest in GPS monitoring of cricket is growing, the reliability
and validity of this methodology have not yet been established. Cricket is charac-
terized by long periods of low-intensity activity interspersed with short bursts of
high-intensity activity.28 The patterns of the movement are presumably influenced
by both the positional requirements and versions of the game. Fast bowlers appear
to perform a large number of longer high-intensity efforts compared with other
players.26 In comparison, wicketkeepers appear to have a higher proportion of
moderate-intensity activity (jogging) with very few high-intensity bursts.26 In
terms of the different versions of the game: Twenty/20 cricket (T20) is the shortest
form of the game and presumably more intensive than both One Day International
(ODI) and First Class cricket. Therefore, it is prudent to assess the validity and
reliability of GPS measurements over sport-specific distances and locomotion
patterns to determine the measurement error. If the measurement error is less than
the within-player variability in movement patterns from game to game, then
coaches and practitioners can confidently use the GPS measurements to estimate
game distances and player movement velocities. An understanding of the game-
to-game variation in individual distances and velocities to distinguish between
heavy and light game load is necessary for prescription of individualized training.
From our pilot data, fast bowlers have within-player game-to-game variability of
~20% in distances and velocities. The aim of this study was to determine the reli-
ability and validity of GPS monitoring for quantifying the movement patterns of
cricket. A secondary aim was to compare the reliability and validity of the three
different GPS units (ranging from 1 to 5 Hz).

Methods
Experimental Design
The viability of GPS monitoring for cricket applications was evaluated by sepa-
rately investigating the validity, reliability, and practicality of commercially avail-
able GPS units. The same participant (male, aged 32 y) participated in all of the
measurement trials to eliminate between-subject variability in estimates of reli-
ability and validity.

GPS Units
Three different types of GPS units were used in this investigation: SPI-10 and
SPI–Pro (GPSports, Canberra, Australia) and MinimaxX (Catapult, Melbourne,
Australia). The MinimaxX and SPI-Pro units operate with a 5-Hz GPS signal,
whereas the SPI-10 uses only a 1-Hz GPS signal. Two individual MinimaxX units
(MinimaxX-A and MinimaxX-B) and two individual SPI-Pro units (SPI-Pro-A
and SPI-Pro-B) were used to determine the magnitude of between-unit variation.
384   Petersen et al

Both brands of GPS unit are currently used in elite Australian and New Zealand
cricket to assess distances covered in various velocity bands (detailed below). The
GPS unit under evaluation was positioned via an elasticized shoulder harness to
sit between the scapulae of the participant at the base of the cervical spine. Two
units separated horizontally by 10 cm were worn simultaneously in each trial.

Distance Trials
The estimated distance covered in different movement categories (walking, jog-
ging, running, and striding) by Academy fast bowlers26 were used to validate each
GPS unit against the criterion distance covered on an internationally certified syn-
thetic 400-m athletics running track. These velocity zones have been used in Aus-
tralian cricket for the last 2 y. The criterion distance and velocity band for each
movement category were as follows: walking (up to 2 m·s−1) 8800 m, jogging (2
to 3.5 m·s−1) 2400 m, running (3.5 to 4 m·s−1) 1200 m, and striding (4 to 5 m·s−1)
600 m. Twenty trials were conducted for each unit for each distance. Pacing feed-
back was given every 200 m to ensure adherence to the specified locomotion
speed, and timing was conducted with a manual stopwatch. Because 400-m athlet-
ics tracks are measured 30 cm from the inner edge of lane 1, the participant was
instructed to try to maintain this radius (with the center of his shoulders) around
the track.

Sprint Trials
The two MinimaxX units and the two SPI-Pro units were used in 20 trials with
sprints conducted from a standing start over 20, 30, and 40 m. Electronic timing
gates were used to obtain a criterion sprint time accurate to 0.01 s. The estimated
sprint distance was determined with Logan plus v4.0 (Catapult, Melbourne, Aus-
tralia) or Team AMS v2.0 (GPSports, Canberra, Australia) software for each trial.
The start time was determined by the first increase above zero on the velocity
trace and then, using each associated sprint time (from timing gates), the associ-
ated distance was determined. It was not possible to test the SPI-10 in the sprint
trials given the limitations of a 1-Hz (1 cycle per second) sampling rate for sprints
lasting ~3 s.
A cricket-specific run-a-three test was also used to evaluate the ability of the
GPS units to monitor sprinting activities. Batsmen often complete multiple runs at
a time in a game involving changing direction 180° at each turn. While the length
of the pitch between crease lines is 17.68 m, a batsman will use the bat to touch
over the crease line so they generally run a shorter distance. The run-a-three test
involved a set of electronic light gates positioned on both the crease marks and
one camera placed perpendicular to each crease mark to record where the batsman
turned. A 1-m calibration mark was made from the crease mark to calibrate the
distance for software analysis (SiliconCoach, New Zealand). The horizontal dis-
placement of the MinimaxX or SPI-Pro units’ position from the crease lines was
determined. Twenty trials of the run-a-three test were used to validate each Mini-
maxX or SPI-Pro unit. The criterion distance covered in each particular run-a-
three test trial was determined using the video recordings, and these were com-
pared with the distances obtained from the GPS units.
 GPS Unit Validity and Reliability   385

Between-Unit Variation and Time of Day

To assess the variation between different GPS units, a custom-designed platform
was attached to a calibrated trundle wheel (refer to Figure 1). Nine MinimaxX
units were placed in a vertical position on the platform. The trundle wheel was
then wheeled over 1200 m at a walking speed around an athletics track with three
trials performed at 4:00 PM, and a further three trials at 9:30 AM on the next
morning. This method permitted the assessment of the effect of time of day on
reliability and validity.

Statistical Analyses
Validity was estimated by subtracting the estimated distance traveled for each trial
from the associated criterion distance. For each locomotion pattern, the standard
deviation of the percentage errors for each unit gave the standard error of the esti-
mate (SEE). Precision of estimation was indicated with 90% confidence limits.
Reliability was estimated using the typical error (TE) expressed as the coefficient
of variation (with 90% confidence limits). Bias was determined by subtracting the
criterion distance from the GPS estimated distance, and then dividing the differ-
ence score by the criterion distance. The data were tested for heteroscedasticity by

Figure 1 — Trundle wheel with milk crate platform. Nine GPS units are positioned in
cradles on this platform for simultaneous recording of distance traveled.
386   Petersen et al

plotting a figure of absolute difference against the mean and computing the
correlation.29

Results
Validity
Validity was assessed over a range of distances and locomotion patterns specific
to cricket. Using all distances, the Pearson correlation (r) for the distance around
the track and the estimated distance from the GPS units was 0.99. The data dis-
played a trivial amount of heteroscedasticity R2 = .03, indicating that as the size
of the measurement increases there are only trivial measurement differences.
The degree of validity varied as a function of distance and intensity of the loco-
motion pattern (Table 1). The validity of the SPI-10 unit during walking to strid-
ing locomotion patterns ranged from 0.5% to 2.1%. Estimation of running had
more than twice the error of walking, jogging, or striding (Table 1). The mean
SEE of the SPI-Pro units for the same locomotion patterns varied between 0.4%
to 3.7%, with estimation of jogging having the greatest error. Larger errors were
evident in sprinting short distances (20–40 m) where the SEE for the SPI-Pro
units ranged between 3 and 11%. Similarly, the SEE for the MinimaxX units
varied between 2 and 4% for walking to striding and 5 and 24% for sprinting.
The validity of the MinimaxX units improved the greater the sprint distance in
the range of 20 to 40 m.

Reliability
Similar to validity, the reliability of GPS estimation of locomotion patterns was
better for longer distances (Table 2). The SPI-10 had good reliability with the TE
<2% over the different distances walking to striding, while the SPI-Pro was <4%.
The MinimaxX units displayed similar reliability over the same locomotion pat-
terns with TE <4%. The reliability of the SPI-Pro units during sprinting ranged
from 2% to 13%, with the shorter sprints (20 m) tending to be less reliable com-
pared with the longer (40 m and run-a-three) sprints. The reliability of the Mini-
maxX units during sprinting ranged from 4% to 43%, again with estimation of
the shorter sprints (20 m) less reliable than the longer (40 m and run-a-three)
efforts.

Bias
The SPI-10 unit underestimated the criterion distance of walking through to strid-
ing by 1% to 3% (Table 3). Similarly, the SPI-Pro units underestimated the crite-
rion distance of walking through to striding by 1% to 4%. In sprinting, the SPI-Pro
units substantially underestimated the sprinting distances by between ~6% to
20%. In contrast, the MinimaxX units overestimated the criterion distance of
walking to striding by up to 3%. However, the MinimaxX units substantially
underestimated the sprinting distances by between ~20 and 40%.
 Table 1 Standard error of the estimate of commercially available GPS units in cricket-specific movement
patterns and distances (percent standard error of the mean ± 90% confidence limits)
MinimaxX-A MinimaxX-B SPI-10 SPI-Pro-A SPI-Pro-B
Walking 8800 m (up to 2m/s) 3.8 ± 1.4 2.0 ± 0.8 0.6 ± 0.2 1.0 ± 0.4 0.5 ± 0.2
Jogging 2400 m (2–3.5 m/s) 2.6 ± 1.0 1.8 ± 0.7 0.5 ± 0.2 3.7 ± 1.4 1.5 ± 0.5
Running 1200 m (3.5–4 m/s) 2.8 ± 1.0 3.0 ± 1.1 2.1 ± 0.8 2.4 ± 0.9 0.7 ± 0.2
Striding 600 m (4–5 m/s) 1.7 ± 0.6 1.8 ± 0.7 0.8 ± 0.3 3.0 ± 1.1 0.4 ± 0.1
Sprinting 20 m (+ 5 m/s) 15.2 ± 5.6 23.8 ± 8.8 — 5.5 ± 2.0 10.5 ± 3.9
Sprinting 30 m (+ 5 m/s) 14.4 ± 5.3 19.7 ± 7.2 — 4.2 ± 1.5 7.6 ± 2.8
Sprinting 40 m (+ 5 m/s) 14.9 ± 5.5 16.1 ± 5.9 — 2.9 ± 1.1 7.7 ± 2.8
Sprinting Run-a-Three (+ 5 m/s) 12.7 ± 4.7 5.3 ± 2.0 — 6.7 ± 2.5 2.6 ± 1.0

Table 2 Reliability of commercially available GPS monitoring (coefficient of variability with 90% confidence
limits) of cricket-specific movement patterns and distances
MinimaxX-A MinimaxX-B SPI-10 SPI-Pro-A SPI-Pro-B
Walking 8800 m (up to 2m/s) 2.6 (2.1–3.6) 1.4 (1.1–2.0) 0.4 (0.3–0.6) 0.7 (0.6–1.0) 0.3 (0.3–0.5)
Jogging 2400 m (2–3.5 m/s) 1.8 (1.5–2.5) 1.3 (1.0–1.7) 0.4 (0.3–0.5) 2.9 (2.3–4.0) 1.1 (0.9–1.5)
Running 1200 m (3.5–4 m/s) 2.0 (1.6–2.7) 2.0 (1.6–2.8) 1.5 (1.2–2.1) 1.8 (1.4–2.5) 0.5 (0.4–0.7)
Striding 600 m (4–5 m/s) 1.2 (1.0–1.7) 1.3 (1.0–1.8) 0.5 (0.4–0.7) 2.3 (1.8–3.1) 0.3 (0.2–0.4)
Sprinting 20 m (+ 5 m/s) 19.7 (15.3–27.9) 30.0 (23.2–43.3) — 4.8 (3.8–6.6) 9.3 (7.3–13.0)
Sprinting 30 m (+ 5 m/s) 15.8 (12.4–22.3) 21.3 (16.5–30.3) — 3.4 (2.7–4.7) 6.3 (5.0–8.7)
Sprinting 40 m (+ 5 m/s) 16.1 (12.5–22.6) 17.1 (13.4–24.2) — 2.3 (1.8–3.1) 5.8 (4.6–8.1)
Sprinting Run-a-Three (+ 5 m/s) 13.6 (10.7–19.1) 5.3 (4.2–7.3) — 6.3 (5.0–8.7) 2.0 (1.6–2.8)

387
388
Table 3 Percent bias (± 90% confidence limits) of GPS measured distances from criterion
distances
MinimaxX-A MinimaxX-B SPI-10 SPI-Pro-A SPI-Pro-B
Walking 8800 m (up to 2m/s) 2.1 ± 1.4 3.2 ± 0.8 −0.8 ± 0.2 −1.1 ± 0.4 −0.8 ± 0.2
Jogging 2400 m (2–3.5 m/s) 0.3 ± 1.0 0.8 ± 0.7 −0.9 ± 0.2 −2.8 ± 1.4 −2.7 ± 0.5
Running 1200 m (3.5–4 m/s) 0.4 ± 1.0 2.6 ± 1.1 −1.0 ± 0.8 −3.8 ± 0.9 −1.2 ± 0.2
Striding 600 m (4–5 m/s) 0.3 ± 0.6 0.0 ± 0.7 −1.7 ± 0.3 −2.7 ± 1.1 −1.0 ± 0.1
Sprinting 20 m (+ 5 m/s) −37.3 ± 5.6 −24.3 ± 8.8 −15.3 ± 2.0 −13.6 ± 3.9
Sprinting 30 m (+ 5 m/s) −27.4 ± 5.3 −20.4 ± 7.2 −10.3 ± 1.5 −11.5 ± 2.8
Sprinting 40 m (+ 5 m/s) −24.1 ± 5.5 −19.5 ± 5.9 −7.7 ± 1.1 −7.4 ± 2.8
Sprinting Run-a-Three (+ 5 m/s) −28.9 ± 4.7 −28.1 ± 2.0 −20.0 ± 2.5 −6.1 ± 1.0
 GPS Unit Validity and Reliability   389

Interunit Variability and Time-of-Day Effects

There were minimal differences in validity and reliability measures calculated
from MinimaxX GPS data collected in the morning compared with the afternoon.
The SEE was 1.7 ± 0.5 and 1.8 ± 0.6 (mean ± 90% confidence limits) and the TE
was 1.3 (1.0 to 1.8) and 1.5 (1.2 to 2.0), respectively, for morning and afternoon
data. Movement data collected in the morning displayed an underestimation bias
of 1%, whereas in the afternoon there was an overestimation bias of 1% (Figure 2).

Discussion
With rapid commercial advancements (and increased commercial competition) in
GPS microtechnology, it is important to ensure that the validity, reliability, and
practicality to athletic training also improves from the current state. Effective
evaluation of the movement patterns and physical demands of cricket can benefit
the importance of the prescription of fitness and condition programs. The SEE
(validity) of distance for walking to striding ranged from 0.4% to 3.8%, and the
estimate of reliability ranged from 0.3% to 2.9%. Even with the same type of GPS
unit simultaneously receiving the same satellite signals, there were discernable
differences in the estimated distances. A major finding of this investigation was
the shortcomings in confidently measuring short sprint efforts. The typical errors

Figure 2 — Interunit variation in estimated distance (mean ± 90% confidence limits) from
1200-m walks with nine GPS units. The three morning trials were conducted at 9:30 AM
and three afternoon trials at 4 PM.
390   Petersen et al

of validity and reliability of estimating sprint distances over 20 to 40 m ranged

from 3% to 40%. The MinimaxX and SPI-Pro units consistently underestimated
short sprint distances. The magnitude of errors associated with particular GPS
outputs should be clearly understood by conditioning coaches in monitoring
cricket workloads and prescribing training. In particular, conditioning coaches
should be aware of the likely underreporting of high-intensity activity and the
overreporting of low-intensity activity.
There are obvious differences in outputs between different brands of GPS
units, even though the GPS receivers are receiving the same signals from the same
satellites. These errors are most likely attributable to variations in the calculation
of the receiver position from triangulation algorithms, or the Kalman (exclusion
criteria) formula used to determine the most logical position. Errors can also
occur in the smoothing techniques used by the software to exclude erroneous data.
The degree of smoothing inherent in the analysis will affect the accuracy of the
estimated distance or velocity measurements. Therefore, where possible, the same
type of unit should be used when making comparisons between athletes or with
repeated monitoring on the same athlete (or player) over time.
The validity of the GPS units measured here appears acceptable for longer
distances at the slower velocities of locomotion (up to striding) with values of less
than 5% fulfilling the criteria used by Duffield and Drinkwater.14 The MinimaxX
units overestimated (by 0% to 3%), whereas the SPI-Pro and SPI-10 units under-
estimated (up to 4%) these distances. However, caution should be exercised when
interpreting short high-intensity efforts by contemporary commercially available
GPS units, as validity of sprinting distances ranged up to 24% with an underesti-
mation bias up to 37%. The type of GPS unit used must be accounted for as this
will determine the bias it expresses over different locomotion patterns. Further-
more, results from different GPS units should not be used interchangeably when
analyzing high-intensity running.25
Typical errors of ~2% with lower intensity activity (<4 m·s−1) found in the
current study are less than half the error reported in a preliminary study25 that
evaluated three different models (SPI-10, SPI Elite, and WiSPI) of GPS units
(GPSports, Canberra, Australia). The typical error of estimating very high inten-
sity running (>5.5 m·s−1) in Coutts and Duffield’s study25 ranged from 12% to
30%, which is in agreement with our findings. Our data supports their finding of
reduced reliability with increased movement velocities. It is possible that because
the number of GPS signals is lower during shorter sprints, the distance of these
sprints is underestimated and had reduced reliability. A report by the manufactur-
ers of the MinimaxX shows that the MinimaxX GPS unit has a typical error of
0.8% for the criterion distance around a 400-m track.30 Interestingly, this report
indicated improved accuracy during jogging and running compared with walking
(3  3 laps around a small 10-  10-m square), and concluded that the MinimaxX
GPS accuracy typically improves at higher speeds.30 Our data does not support
this conclusion and further refinement of this technology is required before
coaches and scientists can be confident about the accuracy of GPS output to quan-
tify short sprints in team sports.
One study reported validity results for measured distance at lower intensity
activity far tighter than our current results (typical errors of less than ~2%). This
study27 investigated variable movement speeds around a 6.8-km painted hockey-
 GPS Unit Validity and Reliability   391

specific circuit (1-Hz GPSports Spi Elite unit). The 7 km·h−1 average movement
speed indicates an overall intensity of walking and with only nine trials of the
complete circuit (only one unit used) the authors reported a TE of 0.1%. It would
be interesting to see if this TE increased with a greater number of trials and with
more regular high-intensity efforts.
Cricket conditioning coaches currently use GPS microtechnology to track
patterns of movements both in a game and between-game variation to modify the
type, duration, and intensity of conditioning sessions.26 The reliability of estimat-
ing longer distances is acceptable for game and training applications. However,
the underestimation bias displayed with the GPS units studied here during sprint-
ing may mean that total game sprint distance is underreported, whereas the total
game distance may be overreported when using MinimaxX units. The noise of up
to 30% in the measurements of cricket specific-sprinting distances is too large for
identifying the smaller within-player game-to-game variability of 20% established
with our pilot game data. Conditioning coaches should focus on game-to-game
variations in total distance covered rather than the sprinting workload when pre-
scribing and evaluating interval training drills.
In a comparison study of GPS technology (SPI-10) to computer-based track-
ing software (Trakperformance), Edgecomb and Norton20 concluded that comput-
er-based tracking was as accurate as GPS technology. Their reliability results
(TEM of 5.5%) for movements up to a running intensity were more than twice the
value found in our present study for the equivalent SPI-10 GPS unit. Improve-
ments made in the intervening period to the GPS technology (updates in firmware
or software) may be the reason for these improved results. With the increased reli-
ability and the practical convenience of GPS monitoring, it appears that notational
analysis and manually controlled computer tracking software will soon be super-
seded in sports allowing GPS technology. However, notational analysis and man-
ually controlled computer tracking software is often the only option for indoor
sports, certain contact sports (safety of landing on the unit), and sports that impose
restrictions on using GPS technology in competition.

Practical Applications
The estimations of distance with GPS were remarkably consistent between the
morning and afternoon. This finding confirms those of MacLeod et al,27 who also
reported that different configurations of satellites at different times of the day did
not substantially affect the GPS output. Our results indicate that temporal longitu-
dinal comparison of GPS data can be made with confidence. However, it is notice-
able to any frequent user of GPS technology that satellite reception can vary with
location. Therefore, we advise practitioners to routinely check indicators of signal
quality and number of satellites being used when interpreting GPS data. It is also
advisable for coaches to check their own units to establish individual unit bias
and, if possible, use the same unit for a particular player (if possible, do not use
units interchangeably25).
In conclusion, in comparison with other workload-monitoring techniques,
GPS athlete tracking technology is practically superior. This technology can pro-
vide acceptable validity and reliability for estimating longer distances of walking
to striding intensity. However, there are shortcomings in the ability of the technol-
392   Petersen et al

ogy to reliably quantify shorter cricket-specific sprinting distances. Further

improvements are required for detailed analysis of high-intensity efforts in the
disciplines of batting, bowling, and fielding.

References
1. McInnes S, Carlson J, Jones C, McKenna M. The physiological load imposed on
basketball players during competition. J Sports Sci. 1995;13:387–397.
2. Spencer M, Lawrence S, Rechichi C, Bishop D, Dawson B, Goodman C. Time-mo-
tion analysis of elite field hockey, with special reference to repeated sprint activity. J
Sports Sci. 2004;22:843–850.
3. Barbero-Alvarez J, Soto V, Barbero-Alvarez V, Granda-Vera J. Match analysis and
heart rate of futsal players during competition. J Sports Sci. 2008;26:63–73.
4. Alexander M, Boreskie S. An analysis of fitness and time-motion characteristics of
handball. Am J Sports Med. 1989;17:76–82.
5. Green H, Bishop P, Houston M, McKillop R. Norman. R., Stothart P. Time-motion and
physiological assessments of ice hockey performance. J Appl Physiol. 1976;40:159–
163.
6. Deutsch M, Maw G, Jenkins D, Reaburn P. Heart rate, blood lactate and kinematic
data of elite colts (under-19) rugby union players during competition. J Sports Sci.
1998;16:561–570.
7. Duthie G, Pyne D, Hooper S. Time motion analysis of 2001 and 2002 super 12 rugby.
J Sports Sci. 2005;23:523–530.
8. Deutsch M, Kearney G, Rehrer N. Time-motion analysis of professional rugby union
players during match-play. J Sports Sci. 2007;25:461–472.
9. 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.
10. Pereira Da Silva N, Kirkendall D, Leite De Barros Neto T. Movement patterns in elite
Brazilian youth soccer. J Sports Med Phys Fitness. 2007;47:270–275.
11. Heller J, Peric T, Dlouha R, Kohlikova E, Melichna J, Novakova H. Physiological
profiles of male and female taekwon-do (ITF) black belts. J Sports Sci. 1998;16:243–
249.
12. Christmass M, Richmond S, Cable N, Arthur P, Hartmann P. Exercise intensity and
metabolic response in singles tennis. J Sports Sci. 1998;16:739–747.
13. Roberts S, Trewartha G, Stokes K. A comparison of time-motion analysis methods for
field-based sports. Int J Sports Physiol Perform. 2006;1:388–399.
14. Duffield R, Drinkwater E. Time-motion analysis of Test and One-Day international
cricket centuries. J Sports Sci. 2008;26:457–464.
15. Rudkin S, O’Donoghue P. Time-motion analysis of first-class cricket fielding. J Sci
Med Sport. 2008;11:604–607.
16. Hebenbrock M, Due M, Holzhausen H, Sass A, Stadler P, Ellendorf F. A new tool
to monitor training and performance of sport horses using global positioning system
(GPS) with integrated GSM capabilities. Dtsch Tierarztl Wochenschr. 2005;112:262–
265.
17. Vermeulen A, Evans D. Measurements of fitness in thoroughbred racehorses using
field studies of heart rate and velocity with a global positioning system. Equine Vet J.
2006;S36:113–117.
18. Cych P. Possibilities of and constraints on the application of GPS devices in con-
trolling orienteering training. Studies in Physical Culture and Tourism. 2006;13:109–
115.
19. Bentley D, Vleck V. Pacing strategy and performance in elite world cup triathlon: A
preliminary study. Med Sci Sports Exerc. 2004;36:S122.
 GPS Unit Validity and Reliability   393

20. Edgecomb SJ, Norton KI. Comparison of global positioning and computer-based
tracking systems for measuring player movement distance during Australian Football.
J Sci Med Sport. 2006;9:25–32.
21. Gamble D, Young E, O’Donoghue P. Activity profile and heart rate of responses of
referees in Gaelic football. J Sports Sci Med. 2007;S10:93.
22. Hartwig T, Naughton G, Searl J. Motion analysis of adolescent rugby union players:
Linking training and game demands among under 16 players. Journal of Science and
Medicine in Sport. 2006;9:S1;16.
23. Pino J, Martinez-Santos R, Moreno M, Padilla C. Automatic analysis of football
games using GPS on real time. J Sports Sci Med. 2007;S10:9.
24. Reid M, Duffield R, Dawson B, Baker J, Crespo M. Quantification of the physio-
logical and performance characteristics of on-court tennis drills. Br J Sports Med.
2008;42:146–151.
25. Coutts AJ, Duffield R. Validity and reliability of GPS units for measuring movement
demands of team sports. J Sci Med Sport. 2008; .
26. Petersen C. GPS monitoring during the COE 2006 emerging players tournament. In
Cricket Australia Sports Science Sports Medicine Conference proceedings. Brisbane,
Australia; 2007:69–70.
27. MacLeod H, Morris J, Nevill A, Sunderland C. The validity of a non-differential
global positioning system for assessing player movement patterns in field hockey.
Journal of Sport Sciences. 2008:1–8, ifirst article
28. Noakes T, Durandt J. Physiological requirements of cricket. J Sports Sci. 2000;18:919–
929.
29. Atkinson G, Nevill A. Statistical methods for assessing measurement error (reliabil-
ity) in variables relevant to sports medicine. Sports Med. 1998;26:217–238.
30. Catapult innovations. Athletics track application: MinimaxX GPS performance, In-
house report, October. 2007.

View publication stats