The repeatability of motion analysis and the reproducibility of athletes in sprint hurdles

  • A. Salo
  • P.N. Grimshaw
  • J.T. Viitasalo


Grainger et al. (1983, Biomechanics VIII-B, pp. 1239- 1247) concluded that biomechanical variables from a single trial based on film data were unstable even in a well-learned pattern such as walking. This leads to the question: how reliable is the analysis of more complex tasks, which are common in most applied sports research? The aim of this study was to investigate the repeatability of motion analysis system and the reproducibility of athletes in the event of sprint hurdles. Eight athletes (4 females and 4 males) performed 2 sets of 4 trials over 4 hurdles in the training situation. The clearances over the third hurdle were filmed with the two genlocked video camera recorders (50Hz). The cameras were located 29 m away from the hurdle in front and symmetrically on both sides of the lanes creating a 90 ' angle from the midpoint of hurdle. The filmed view was 6.5 m in the direction of running ensuring as large an image size as possible. All the eight trials per subject were digitised using the "Kine analysis" software. Additionally two trials (one female and one male) were randomly selected and re-digitised eight times. Twentyeight kinematic variables were studied and coefficient's of variation (CV) were calculated for each subject and for both re-digitid data sets independently. Individual CV's varied from 0.4% to 138.0% for the female athletes and from 0.8% to 181.2% for the male athletes. The ranges for the female and male redigitised trial were 0.1-151.2% and 0.2-198.7%, respectively. There were 8 variables in both female and male groups, in which the CV of the re-digitised trial were greater than the lowest CV of individual athletes in the same gender. This indicates that the digitising and analysis process is sensitive enough to potentially separate technical aspects of performance. However, it also indicates that in these eight variables (which were not the same variables for both genders) the variation of results may be due to the digitising and analyses process. Although, some of the CV's were high, there were only 8 variables were one or more of the females produced over 10% CV (9 variables for males). For the re-digitised trial there were 16 and 13 variables for the female and male, respectively, in which the CV did not exceed the 2% level. Thus, it can be concluded that one operator can perform the digitising process repeatably and that in this context the analyses is reliable for most of the variables. This study, however, did not clarify the accuracy and reliability of motion analysis in absolute terms. Furthermore, the athletes were able to reproduce their performance within certain limits despite such a complex movement. However, there were differences in reproducibility between females and males due to different clearance technique. In conclusion, it is clear that for some variables more digitised trials are needed to clarify individual technical aspects of performance.
Coaching and Sports Activities