Search

Understanding reliability of performance analysis tools is important to ensure match to match comparisons can be undertaken with the knowledge of consistency between coding situations. There are few published studies examining the reliability of commonly used performance analysis tools. The aim of this project was to assess the inter- and intra-rater reliability of the NetballStats application and to make comparisons between live and video-based coding situations. Two ‘coders’ coded eight netball matches using the NetballStats application, coding each match live, then twice from video. Level of agreement was assessed for frequency counts across the variables coded. Results showed that intra-rater agreement was higher than inter-rater agreement and that reliability from video coding is better than from live coding. High frequency events automatically coded by the application and events that are well defined had greater levels of agreement than lower frequency events and subjectively judged events. Live coding situations underrepresent occurrence of events, particularly for high frequency events such as ‘possession’. To ensure reliability between coders, clubs should provide an extensive training program to coders with clear instructions on coding subjective events. Coaches should be aware that live coding underestimates some event types and factor this into their decision making processes.

Studies of training and competition load in sport are usually based on data that represents a sample of a league and or annual training program. These studies sometimes explore important factors that are affected by load, such as training adaptations and injury risk. The generalisability of the conclusions of these studies, can depend on how much load varies between seasons, training phases and teams. The interpretation of previous load studies and the design of future load studies should be influenced by an understanding of how load can vary across seasons, training phases and between teams. The current study compared training loads (session rating of perceived exertion x session duration) between all (8) teams in an elite Netball competition for multiple (2) season phases and (2) seasons. A total of 29,545 records of athlete session training loads were included in the analysis. Linear mixed models identified differences between seasons and training phases (p < .05). There were also differences between teams and a complex set of interactions between these three factors (season, phase, and team) (p < .05). While the absolute value of the training loads reported here are only relevant to elite netball, these results illustrate that when data is sampled from a broader context, the range and variation in load may increase. This highlights the importance of cautiously interpreting and generalisation of findings from load studies that use limited data sets.