GPS Validity & Reliability – A Collection of Independent Research Findings
GPS technology provides coaches, sport scientists and athletes the opportunity to collect individual physical data that has not previously been available and on a time efficient, relatively low cost, non-invasive and potentially real time basis. As well as the practical benefits, there has been a surge in uptake through the elite sport setting probably due to those in the elite setting wanting to gain a competitive advantage and/or not fall behind their opposition. Although it first became available to the sports industry in 2003, little validation research was published until 6-7 years later. Given the increase in popularity, GPS technology has been exposed to the High Performance team sport environment before thorough research has been conducted into data accuracy and global definitions of metrics. This pattern endures as manufacturers release new models and firmware updates ahead of academic research.
I was surprised to struggle to find a collection of the independent findings from the literature readily available on the internet (published literature reviews aside), so here is my selection of some of the research:
|Course Design||Parameter(s)||Criterion measure(s)||Conclusion Excerpt|
|Edgecomb and Norton, 2006||GPSports
|Predetermined marked circuit||Distance||Trundle wheel||GPS overestimated true distance travelled by players (average <7%). There is relatively large error when small distances are involved (<200m) but these become less over longer movement patterns (>2km).|
|Macleod et al, 2009||GPSports
|Simulated team sport (hockey) circuit||Distance and speed||Trundle wheel and timing gates||Valid tool for measuring speed and distance during hockey|
|Coutts and Duffield, 2010||GPSports
SPI-10, SPI Elite and WiSPI
|Simulated team sport circuit||Total distance, high intensity distance (>14.4km/hr), very high intensity running distance (>20km/hr) and peak speed during 20m sprint||Measuring tape and timing gates||Accurate and reliable information on total distance travelled during team sport running patterns; however data from different devices should not be used interchangeably. 1 Hz may not provide accurate information regarding high intensity activities, especially if they are completed over a non-linear path|
|Gray et al, 2010||GPSports
WI SPI Elite
|Linear and non-linear (including curved) 200m courses||Distance across different movement velocities and mean and peak velocities||Actual distance using total station EDM/theodolite||GPS distance demonstrates reduced validity in non-linear movement patterns, including curved or circular paths, as movement intensity increases… Although 1-Hz GPS receivers should be considered a reliable tool for measuring distance travelled by athletes in field-based team sports, multiple changes in direction at high speed may reduce both reliability and validity.|
|Jennings et al, 2010||Catapult MinimaxX
1 & 5 Hz
|Straight line running and change of direction courses at various self-selected speeds and simulated team sport circuit||Distance across different movement velocities||Measuring tape and goniometer for change of direction sections, and timing gates||The reliability and validity of GPS to estimate longer distances appears to be acceptable (<10%). However, currently available GPS systems maybe limited for the assessment of brief, high speed straight line running, accelerations or efforts involving a change of direction. An increased sample rate improves the reliability and validity.|
|Portas et al, 2010||Catapult MinimaxX
1 & 5 Hz
|Linear, multidirectional and soccer specific courses||Distance||Trundle wheel and tape measured distance||Both methods produced valid and reliable measures of linear motion and could be used to precisely quantify total distance motion in linear sport activity such as running. For multidirectional motion, both 1-Hz and 5-Hz were valid and reliable in less challenging scenarios but not in the more complex where reliability decreased… presently NdGPS is incapable of detecting small but practically important changes (SWC)|
|Waldron et al, 2011||GPSports
|Straight line sprinting||Distance, peak speed and proper acceleration (integrated accelerometry)||Measuring tape and timing gates||GPS devices (5 Hz, SPI-Pro) can be used to quantify small, yet practically significant 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.|
|Varley et al, 2012||Catapult MinimaxX V2.0 5 Hz
|Straight line running||Acceleration, deceleration and constant velocity||Instantaneous velocity using a tripod mounted laser||Superior validity and inter-unit reliability of V4.0 Minimax compared with the older V2.0 units… The latest V4.0 units sampling at 10 Hz provide sufficient accuracy to quantify the acceleration, deceleration and constant velocity running phases in team sports.|
|Johnston et al, 2012||Catapult MinimaxX V2.5
|Simulated team sport circuit (Coutts and Duffield, 2010) and flying 50m sprint||Total distance, peak speed, player load, distance covered/time spent/number of efforts across different movement velocities||Measuring tape and timing gates||This study demonstrated that 5-Hz GPS units are capable of measuring the fundamental movement demands of TD and peak speed. However, because of the levels of reliability revealed in this study of 5-Hz GPS units it is recommended that they are only used to measure the distance covered, time spent, and number of efforts performed at different speed zones at low speeds|
|Buchheit et al, 2013*||GPSports
SPI-ProX and SPI-ProX2
(although looks like the 5 Hz model in the pic?)
|Simulated match running activities. Data collected during training sessions.||Total distance, distance >14.4 km/hr, distance > 25.1 km/hr, peak speed, accelerations and decelerations||Reliability was assessed between model, unit and software update||Very large variations in common GPS measures (particularly accelerations and decelerations) between models and units from the same manufacturer. Analysis of same data files with different software versions showed substantial differences in the occurrence in accelerations and decelerations.|
* For more on the Buchheit et al (2013) paper look up Jonny’s blog on it here: http://sportsdiscovery.net/journal/2014/03/09/monitoring-with-gps-time-to-slow-down/
The complexity of validity and reliability of the technology to quantify movement demands is significant and stating an across-the-board conclusion is impossible due to the variation of manufacturers, models, sampling frequency and algorithms, plus the different parameters, task designs, criterion measures and statistical analysis employed in the research. Consequently, practitioners working with GPS must take steps to understand the validity and reliability of their own devices within their own setting and usage. Caution is strongly advised if using models and systems interchangeably.
The demands of the movement significantly influences the accuracy of the data collected with trends in reduced accuracy with increases in velocity, decreases in task duration and increases in the complexity of the movement path (i.e. non-linear/curved/multidirectional). Of course these are the kinds of movements that are arguably of most interest in team sports – short, high intensity sprints/accelerations/decelerations/change of directions. Although research has shown this error can be improved with increased sampling frequency, limitations remain. We must question if we are accurately capturing the most crucial movements relating to loading through the body and possibly impact on the game?
As well as sampling frequency we have to consider the variation between manufacturers, models, hardware, software etc. I find it really interesting to see that the manufacturers covered in this sample of literature are limited to only two of the leading manufacturers; I would certainly be interested to read the independent findings into the others (if these even exist?!). Martin Buchheit’s study showed substantial differences in outputs from the same data files using different software versions so care must be given in the applied world to if, when and how we run the updates. Finally there are also environmental factors that can affect the quality of data (see my previous post: Inside the GPS Unit for more detail http://sportsdiscovery.net/journal/2014/03/14/inside-the-gps-unit/)
Whilst the research moves on to explore new ventures; small sided games, metabolic power, accelerometer data for collision sports to name but a few, it is essential we do not forget the foundation of accuracy of the technology.
As well as the research cited above there are a number of review papers that further summarise the findings of GPS validity plus the potential applications of the technology:
Aughey RJ. (2011) Applications of GPS technologies to field sports. Int J Sports Physiol Perform 6: 295-310.
Cummins C, Orr R, O’Connor H, et al. (2013) Global positioning systems (GPS) and microtechnology sensors in team sports: a systematic review. Sports Med 43: 1025-1042.
Dellaserra CL, Gao Y and Ransdell, L. (2014) Use of integrated technology in team sports: a review of opportunities, challenges, and future directions for athletes. J Strength Cond Res 28: 556-573.
[…] GPS companies allow you to access data on how many satellites were locked onto a device during a session and perhaps the development of using GPS in competitive matches emphasises the need for us to use this function. We will also need to carry out validity and reliability research for GPS specifically in the stadium environment. [To revisit GPS validity and reliability research available see this summary in a previous post: GPS Validity & Reliability – A Collection of Independent Research Findings.] […]