Representative Learning Design and Practice Assessment in Sport

Definition and Core Concepts of Representative Learning Design

• Representative Learning Design is defined as the degree to which information sampled in experimental and practice tasks is representative of the specific performance contexts that the tasks are attempting to simulate (Krause et al., 2018).

• It serves as a framework to support coaches, researchers, and sports scientists in developing training and practice sessions.

• Functionality refers to the degree to which an athlete is able to use the same information sources, such as visual cues, that are present during competition to contextualize their decisions and movements.

• Action Fidelity refers to the degree to which an athlete’s movement behavior, specifically spatiotemporal kinematics, during practice replicates movement performance during competition.

Essential Terminology

• Representative Learning Design

• Representative Practice Task

• Representativeness

• Representative Design

• Representative Task Design

• Representative Experimental Design

• Specificity of Learning Hypothesis

• Ecological Validity

• Perception-Action Coupling

• Especial Skills

• Stimulus-Response Compatibility

• Expert-Performance Approach

Critical Analysis of Training Equipment: "Crazy Catch"

• The "Crazy Catch" is marketed as a tool to improve reactions and catching skills, with testimonials suggesting it is suitable for beginners through professionals. England Women’s Cricketer Heather Knight states it is a "fun, easy way to improve your reactions."

• From a Representative Learning Design perspective, practitioners must evaluate the tool based on the following criteria:

  • Representative speed?

  • Representative trajectory?

  • Representative distance?

  • Representative force?

  • Representative auditory cues?

  • Representative visual background?

  • Representative decision making?

  • Representative concentration?

• Similar critical evaluations are required for other equipment like Spinfire ball machines or Technogym resistance equipment.

Barriers and Arguments Against Representative Design

• Several factors contribute to the resistance against implementing representative designs in training:

  • Logistics: It is often easier to set up non-representative, static drills.

  • Reductionist Beliefs: The belief that skills must be learned in constituent parts before being integrated.

  • Desire for Consistency: A preference for blocked practice or "repetition without repetition" (consistent responses).

  • Tradition: Adherence to the sentiment that "That's just the way it's always been."

  • Injury Risk: Concerns that highly representative, high-intensity environments may increase the likelihood of injury.

The Importance of Design in Research and Practice

• There is a distinction between Representative Learning Design (used in practice) and Representative Experimental Design (used in research).

• Lack of representative design can lead to results that are reliable but not valid. A design that is both reliable and valid must be representative.

• Pinder et al. (2011) emphasizes that small changes in task constraints in sport studies can cause substantial changes in performance outcomes and movement responses.

• Static tests lack functionality and do not successfully represent the constraints of performance environments. Practitioners are encouraged to design dynamic interventions considering interacting constraints on movement behaviors.

Research Evidence for Representation and Coupling

Farrow & Abernethy (2003): Perception-Action Coupling

• This study examined if the degree of perception-action coupling affects the ability to anticipate the direction of a tennis serve.

• The experiment compared "coupled" conditions (where movement is linked to perception) and "uncoupled" conditions.

• Results indicated that expert performance in prediction accuracy was significantly higher in coupled conditions (approximately $80\%$) compared to uncoupled conditions (approximately $65\%$). Novices showed little difference between conditions, performing at approximately $50\%$ accuracy for both.

Carboch et al. (2014): Ball Machines vs. Real Servers

• This study investigated if the movement profile of a tennis return changes when receiving a ball from a machine versus a real server.

• Findings showed that players have a shorter initial movement time and a longer backswing duration when facing a ball machine.

• The researchers concluded that the use of ball machines should be limited because they do not adequately represent the cues from a human server.

Roca et al. (2014): Cognitive Processes in Footballers

• This study compared the cognitive processes of semi-professional footballers in stationary conditions versus movement conditions (moving in conjunction with a video sequence).

• Data for verbal statement categories showed higher values for the movement group across several metrics:

  • Monitoring: Stationary = $2.93$; Movement = $3.60$

  • Evaluation: Stationary = $0.27$; Movement = $0.93$

  • Prediction: Stationary = $0.87$; Movement = $1.43$

  • Planning: Stationary = $0.30$; Movement = $1.40$

• While there was no significant difference in anticipation accuracy, the difference in decision-making accuracy approached significance, with the movement group performing better. Decision-making accuracy for the movement group was $67\%$ compared to $49\%$ for the stationary group.

Real-World Inconsistencies: Career Swing% vs. BP Swing%

• Data comparing career swing percentages to Batting Practice (BP) swing percentages for various athletes (A through H) indicates a significant discrepancy. For example, some career swing percentages are as high as $82.3\%$ while others are drastically lower in practice, suggesting that practice swing behaviors do not always replicate competition behaviors.

The Representative Practice Assessment Tool (RPAT)

• Developed by Krause et al. (2017), the RPAT aims to help practitioners assess current practices and guide the design of new ones.

• The tool was developed using the Delphi method, which involves gathering insights from a panel of experts through multiple rounds of questioning.

• The expert panel consisted of $10$ motor learning academics and $13$ elite tennis coaches.

• Inclusion criteria required >80\% agreement on a Likert scale; items with lower agreement were removed or reworded for the next round.

Evolution of the RPAT through Delphi Rounds

• Round 1 included $10$ initial items.

• In Round 2, modifications included splitting questions into "task goal" and "representativeness," combining items, and removing items (Q9 and Q10) based on panel feedback.

• In Round 3, questions were re-split (such as variability Q4 and Q5) and re-ordered to improve flow.

• Final Scoring System: Uses a 5-point scale (ranging from 1 = "Not at all" to 5 = "Certainly").

Core Components of the RPAT

• Q1: Captures the importance of designing tasks for specific learning outcomes relative to competition.

• Q2: Assesses the impact of constraints on the nature and difficulty of the task for the learner.

• Q3 & Q4: Assesses the importance of variable movement solutions versus constant ones (between-skill and within-skill variability).

• Q5: Assesses the relationship between ball delivery and flight information (early anticipation cues).

• Q6: Assesses movement coordination (action fidelity).

• Q7: Assesses if sufficient affordances exist for context-specific decision making.

Application of the RPAT in Tennis (Krause et al., 2018)

• The tool was used to assess four common tennis drills for ground ball striking.

• All tasks scored high for task goals, but representativeness was variable.

• Task 4 was the most representative (score of $65/70$), followed by Task 3 ($57/70$), then Tasks 1 and 2 ($48/70$).

• Key findings from the application:

  • Players hit the ball significantly faster and from deeper in the court during practice tasks compared to matchplay.

  • Practice tasks 1 and 2 saw more balls hit "in," more forehands, and more topspin compared to matchplay.

  • Practice tasks 1, 2, and 3 resulted in fewer winners than in matches.

  • The RPAT highlighted specific areas for improvement, such as ball feed in Task 4 and skill variability in Task 1.

• Psychological constraints are often under-represented in practice designs.

• Gender differences suggest that task designs need to be sensitive to gender-specific representativeness.

The Expert Performance Approach and Performance Assessment Model (PAM)

The Expert Performance Approach

• A theoretical framework devised by Ericsson & Smith (1991) for understanding elite performers.

• Key Tenets:

  • Observe expert-novice differences in-situ to identify underlying skills.

  • Examine practice histories and use learning studies to understand how advantages are developed.

  • Use study designs (e.g., occlusion paradigms) to identify mechanisms behind performance differences in representative situations.

Performance Assessment Model (PAM) for Australian Football (AF)

• Developed by Bonney et al. (2019) to create a structured framework for assessing technical skills, specifically the drop-punt kick.

• The model features five levels of increasing representativeness and performance demand:

  • Foundation Stone: Notational Analysis. Identifying key components (e.g., kicking and handballing frequency) from match play that require assessment.

  • Level 1: Laboratory Test. Controlled environments providing high reliability (e.g., $VO_2$ max tests or biomechanical movement patterns).

  • Level 2: Field-Based Test (Static). Isolating technical elements in a closed environment (e.g., AFL Draft Combine). It separates the skill from competition context.

  • Level 3: Field-Based Test (Dynamic). Incorporates a minimum of three match-specific components and higher intensity. Perception-action coupling is essential, though it remains structured without opponents.

  • Level 4: Field-Based Test (Small-Sided Game). Integrates technical, tactical, physical, and psychological components in an open environment replicating competition.

  • Level 5: Match Play. The most representative level, encompassing all components, but the least controllable due to weather, tactics, and opposition.

• Successful talent prediction is more likely when tests are representative and use an integrated approach rather than isolated assessments.