A video analysis framework for the rugby league tackle

ABSTRACT

Rugby league tackle video analysis research typically uses technical criteria from coaching cues or tackle variables from rugby union. As such, content validity and relevance could be questioned. A video analysis framework that establishes appropriate variables for rugby league is therefore required. The study aimed to adopt a 5-stage process to establish a video analysis framework for the rugby league tackle, which was content valid, relevant and reliable.

The 5-stage process included 1) creation of draft video analysis framework, using available rugby tackle research, 2) expert group recruitment and critique, 3) refinement of framework to establish content validity, 4) response process validity task and agreement within expert group, 5) intra- and inter-reliability testing using Kappa statistics.

The agreed framework comprised six phases including; tackle event, defensive start point, pre-contact, initial contact, post-contact and play-the-ball. Within the identified phases, 63 variables were established. The intra- and inter-reliability testing resulted in strong agreement within all phases.

The video analysis framework can be used in rugby league tackle research, categorising complex tackle events, such as injurious or optimal tackles, improving both player welfare and performance. The application of the framework to future rugby league research will increase coherence and usefulness of research findings.

KEYWORDS:

• Rugby league
• tackle
• variables
• video analysis
• injury
• performance

Introduction

Rugby league is a physically demanding contact sport, which involves frequent exposures to physical collisions and tackles (Gabbett and Domrow 2005). Success is often dependent on the tackling ability and capacity to tolerate physical collisions (Gabbett and Ryan 2010), as rugby league involves approximately 700 tackles per match for all ability levels (King et al. 2010). Within the tackle event, the tackler(s) (i.e. defender) attempts to reduce or stop the momentum of the ball carrier. In rugby league, during the tackle more injuries occur to the ball carrier than the tackler (Gabbett 2000; Gabbett and Domrow 2005; Fitzpatrick et al. 2018).

Video analysis has been adopted by many sports as a tool to quantify performance (Hendricks et al. 2014, 2018; Speranza et al. 2017a) and identify mechanisms of injury (Hendricks et al. 2015; Burger et al. 2016, 2017; Davidow et al. 2018). Detailing the exact moment of injurious events via video analysis is a key step within injury prevention. For example, the concurrent use of video analysis and injury surveillance data can be used to identify high(er) risk activities, e.g. the shoulder charge in rugby league (Cummins and Orr 2015). In doing so, governing bodies can implement strategies to mitigate high-risk activities. Within rugby union, using video analysis, the identification of higher technique proficiency in tackling has been associated with a lower risk of injury (Burger et al. 2016; Hendricks et al. 2016; Davidow et al. 2018). This outlines the benefits of aligning video analysis within injury research.

Despite the continued use of video analysis research in rugby union determine the mechanisms of injury, the association between specific rugby league tackle variables and injury risk has yet to be investigated. This appears to be due to the lack of research validating rugby league tackle variables, and to date, a comprehensive framework linking tackle type and execution with the outcome of the tackle does not exist for rugby league. Previously, performance-related research using video analysis methodologies (Gabbett and Ryan 2010) have adapted a six variable criteria from a standardised one-on-one tackling drill designed by two experienced rugby league coaches. The relationship between tackling ability and player performance was investigated, showing that players who participated at a higher level performed better within the drill (Gabbett and Ryan 2010). Additionally, significant relationships were observed between the criteria and the proportion of missed and dominant tackles during a rugby league match. Strong associations between one-on-one tackles during a training drill and match-play tackle characteristics in rugby league players have also been found (Speranza et al. 2017a). Tackles during rugby league match play were coded using variables previously used in rugby union research (Quarrie and Hopkins 2008; Hendricks et al. 2014) which may be limited due to the inherent differences (Collins 2006; Eaves and Evers 2007). For example, in rugby league, the ball carrier is required to ‘play the ball’ following a tackle (Eaves and Evers 2007). Whereas in rugby union, the ball carrier is required to ‘release the ball’, consequentially each team competes for possession, forming a ruck. Therefore, the application of rugby union variables may not be appropriate to investigate a rugby league specific tackle event. Additionally in rugby union research, phases of the tackles have been identified (Burger et al. 2016; Davidow et al. 2018; Hendricks et al. 2018; Tierney et al. 2018), however, they may not cover all aspects of the rugby league tackle. Therefore, content validity when using rugby union variables for rugby league may be inappropriate (ODonoghue 2014).

The first video analysis framework in rugby union has recently been established (Hendricks et al. 2020). Consequently, future video analysis-based research in the population will be able to compare findings between studies (Hendricks et al. 2020), which is not possible in rugby league at present. Therefore, a content valid, relevant and reliable video analysis framework is required in rugby league, to determine the appropriateness of phases and variables suitable for use in research (ODonoghue 2014). This can then be used to establish mechanisms of injury or successful performance during a rugby league tackle. Therefore, the purpose of the study was to develop a video analysis framework with specific phases, variables, descriptors along with operational definitions which can be used within future injury or performance-related research when investigating the rugby league tackle.

Materials and methods

Design

A 5-stage process was undertaken (Figure 1) based upon previous recommendations to create a video analysis framework which has evidence of content, response process and internal structure validity (Cook and Beckman 2006). During this process, the phases, variables, descriptors and definitions within the rugby league tackle event were established.

Figure 1. The process of phase, variable, descriptor and definition creation for the video analysis framework.

Stage 1 – Rapid review of literature and creation of draft variable list

Stage 1 included a rapid review of relevant online peer-reviewed literature using the electronic databases PubMed and Web of Science. This was to capture the most common tackle phases, variables, descriptions and definitions used in literature (Grant and Booth 2009). The time frame for inclusion was any study up until January 2018. Key search terms were used, including ‘rugby’, OR ‘rugby league’, in combination with ‘tackling’ OR ‘tackle’ OR ‘variables’. Phases, variables and descriptors were only extracted from the literature if an appropriate definition was provided. The variables and their descriptions and defintions were extracted from a total of 13 articles (Deutsch et al. 2007; Quarrie and Hopkins 2008; Wheeler et al. 2010; Fuller et al. 2010; King et al. 2010, 2012; Hendricks et al. 2014, 2015; Sewry et al. 2015; Cummins and Orr 2015; Burger et al. 2016, 2017; Speranza et al. 2017a) and a draft list of variables and descriptors with definitons was created, to be critiqued in stage 2. Based on the literature, phases of the tackle were identified (Fuller et al. 2010; Hendricks et al. 2012) and the variables from the literature were aligned to the specific phase.

Stage 2 – Expert group recruitment and critique

Six experienced practitioners from relevant disciplines within rugby league were invited to participate, based on their specific expertise. The recruitment and formation of the expert focus group followed previous recommendations, regarding size and areas of expertise (Adams and Wieman 2011). One practitioner declined, leaving five experienced practitioners. The participants were all familiar with each other prior to the formation of the expert group, which facilitated high levels of interaction (Gill et al. 2008). The participants were informed of the overall purpose and had the opportunity to ask any questions regarding the scope of the study. The expert group consisted of an international coach (12 years experience as a full-time professional coach), an academy coach (4 years experience as a full-time professional coach), a performance analyst from a professional Super League club (9 years experience as a full-time professional analyst), and two elite level referees (3 and 17 years experience as full-time professional referees). The purpose of the expert group was to ensure a high level of content validity by sense checking and discussing which phases, variables, descriptors and definitions accurately represent important information relating to the rugby league tackle. During the meeting, to introduce different topics of discussion, the researcher asked open-ended questions to facilitate open conversations which were not influenced through researcher bias (Adams and Wieman 2011). Throughout the process, the expert group refined the list of phases, variables, descriptors and definitions and provided further recommendations. They also advised on any recent rule changes to ensure the outcome was relevant to the current tackle scenario.

The key objective of the first expert group meeting was to critique the phases (e.g. pre-contact phase), variables (e.g. upper body movement), descriptors (e.g. no upper body lean) and definitions (e.g. ball carrier displayed no body lean and was in an upright position) developed from the literature and identify any important missing elements. A video loop of different rugby league tackles was displayed during the meeting to ensure that the expert group were prompted of all possible components of the tackle. The video clips were selected by the research team to ensure a variety of tackle types and scenarios were considered. During the meeting, the draft video analysis framework was edited, annotated, and specific key themes were extracted. In addition, the meeting was recorded and referred to within Stage 3 in order to establish and extract further information (Adams and Wieman 2011).

Stage 3 – Refinement of phases, variables, descriptors and definitions

The updated phases, variables and descriptors and definitions developed in Stage 2 were refined by the research team to remove duplication or unclear information. The refined video analysis framework was then sent to the expert group to review independently prior to a secondary meeting.

Stage 4 – Second expert group meeting, response process task and agreement

A second meeting took place to raise any potential issues found when reviewing the video analysis framework. As a group, during the meeting they agreed that all relevant points within a rugby league tackle were clearly described. By the end of the meeting, all phases, variables, descriptors and definitions were established.

To ensure response process validity, the international coach, academy coach, performance analyst and one researcher then coded six tackles independently from pre-existing match footage using a paper copy of the finalised variable list. The aim of this was to allow the experts to go through the process of using the video analysis framework, to determine if anything was missing and check the definitions were clear. Once completed, results were compared and discussed between each coder. The experts were given the opportunity to discuss anything pertinent from the coding task. From this, the group agreed that no further changes to the video analysis framework were required.

Stage 5 – Establish reliability and finalise video analysis framework

The final stage was to establish internal structure validity through an acceptable level of reliability (Cook and Beckman 2006). To test the overall reliability of the variable framework, an intra and inter-coder reliability analysis was completed. For intra-coder reliability, 30 randomly selected tackles were coded by the lead author (3 years’ experience coding RL matches) twice using SportsCode elite (version 6.5.1). Coding of the same 30 tackles was separated by seven days (Wheeler et al. 2010). For inter-coder reliability, a second coder (5 years’ experience coding RL matches) then coded the same tackles. Kappa statistics (κ) were used to evaluate intra- and inter-coder reliability for each randomly selected tackle (James et al. 2007). Kappa values between 0.90 and 0.99 show almost perfect agreement between repeated measures, values between 0.8 and 0.89 represent strong agreement, and 0.6 to 0.79 represent moderate agreement (ODonoghue 2014).

Results

An overview and the descriptions of each phase are presented in (Table 1). The tackle variables, descriptors, and definitions for each phase are presented in (Table 2–7). In total, there were six phases, 63 variables, and 241 categorical descriptors with 6 continuous (time) variables. The phases were categorised as tackle event (Table 2), defensive start point (Table 3), pre-contact phase (Table 4), initial contact phase (Table 5), post-contact phase (Table 6) and play the ball phase (Table 7). Where appropriate, example images are provided for the descriptors in the supplementary material.

Table 1. Overview of tackle phase variables, descriptors and definitions

Table 2. Tackle event variables, descriptors and definitions

Table 3. Defensive start point variables, descriptors and definitions

Table 4. Pre-contact phase variables, descriptors and definitions

Table

Table 5. Initial contact phase variables, descriptors and definitions

Table 5. (Continued)

Table 6. Post-contact phase variables, descriptors and definitions

Table 6. (Continued)

Table 7. Play the ball phase variables, descriptors and definitions

Intra- and inter-coder reliability

Intra-coder reliability for the 30 tackles were: Tackle event variables κ = 0.95, defensive start point variables κ = 1, pre-contact variables κ = 0.94, initial contact variables κ = 0.89, post-contact variables κ = 0.9, play the ball variables κ = 0.96. The inter-coder reliability was assessed using the same methods and were: Tackle event variables κ = 0.92, defensive start point variables κ = 0.85, pre-contact variables κ = 0.81, initial contact variables κ = 0.82, post-contact variables κ = 0.81, play the ball variables κ = 0.88. The specific kappa statistics for each variable areincluded in the supplementary material.

Discussion

The aim of the current study was to develop a valid video analysis framework that accurately reflects the complexity of the rugby league tackle through different phases, variables and descriptions. The variables were created for use in rugby league video analysis research to provide opportunities to better understand both injury and performance mechanisms during the tackle event.

The importance of video analysis for enhancing player and team performance is well documented (Hendricks et al. 2020). In rugby league tackle research there are currently no video analysis frameworks of tackle-related variables which can be used in future research. This has recently been undertaken in rugby union (Hendricks et al. 2020). Without a well-established rugby league video analysis framework, content validity may be questioned (ODonoghue 2014). The current study provides further detail to that of Gabbett (2008) which described six rugby league tackle variables. Whilst the variables used within Gabbett (2008) described are relevant to rugby league, given they were developed by rugby league coaches, no defensive start point, pre-contact or play-the-ball variables were included. The current study created a video analysis framework which covers all aspects of a rugby league tackle. Therefore, future research can utilise all specific phases or individual variables described in the current study, and be confident of their content validity (ODonoghue 2014).

The relevance of variables within the video analysis framework is important to ensure variable validity (O’Donoghue 2009). Rugby league tackle-related research has previously used rugby union specific tackle variables, which may not represent the specific aspects of rugby league. As a result, findings may not directly translate and potential issues of variable validity may arise. Stage 1 of the current study was completed to provide a foundation for the video analysis framework and although only a rapid review was complete instead of a more comprehensive systematic review, the variables captured at this stage allowed the expert group to successfully critique variables for relevancy, as well as content validity. Stages 2–4 ensure the variables are relevant through consultation with expert practioners within the field. This is an important process to undertake as rugby league and rugby union tackle events are disparate. For example, in rugby league the attacking team can only concede six tackles before a turnover (Gibbs 1993), whereas rugby union has an unlimited number of tackles (Wheeler et al. 2013). For that reason in rugby league, it is important that the defending team reduces the chance of an offload during the tackle, in order to minimise the territory gain of the attacker during the set of six tackles. To do this, the defending side frequently engage two or more tacklers in the tackle event (King et al. 2010). Consequentially, the behaviours of tacklers in rugby league and rugby union are likely different. Twenty of the 63 variables established in this study were also included within the rugby union video analysis framework (Hendricks et al. 2020). This allows some comparison between rugby league and rugby union, but also ensures the framework is specific to rugby league. Therefore, rugby league research should adopt the rugby league specific variables within the video analysis framework to capture the most appropriate information.

An important methodological aspect for video analysis is to clearly describe and define actions and events to reduce bias and improve reliability (James et al. 2007). Therefore, the agreed descriptors of the variables were developed to be used as part of the video analysis framework and the reliability of each variable was established. The intra- and inter-reliability scores were ‘strong’ to ‘almost perfect’. The variable list therefore shows comparable reliability to previous video analysis research using kappa statistics in rugby union (Hendricks et al. 2016, 2018; Burger et al. 2017; Tierney et al. 2018) and rugby league (King et al. 2010). As a result, the rugby league video analysis framework demonstrates both strong reliability and strong relevancy to a rugby league tackle. Researchers can therefore be confident that the variables within the list are valid for research (ODonoghue 2014). This is vital for the advancement of rugby league video analysis research, not only to better understand mechanisms of the tackle, but to have consistent measurements between studies.

Recommendations and future direction

The 5-stage process has guided the creation of the video analysis framework which is relevant, reliable and content valid. The provision of in-depth definitions, images (supplementary material) and a familiarisation process allows researchers to accurately and precisely code the tackle within video analysis investigations. Future studies can utilise appropriate variables which are relevant to the research question to analyse tackle events in rugby league. For example, research aiming to better understand the ‘play-the-ball’ can use variables within the post-contact and play-the-ball phase. The consistent use of the video analysis framework will allow appropriate comparisons within and between research to increase the coherence and usefulness of future research findings. In turn, this will allow rugby league practitioners and governing bodies to make informed decisions for performance and injury prevention.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary data

Supplemental data for this article can be accessed here.

Additional information

Funding

No funding is associated with this study.