Agility training: how to improve skill and technique

Overloading on repetitious training plans will not improve an athlete’s skill or technique

I recently spent a week on holiday in the Algarve where my hotel room was overlooking a golf course and driving range. I awoke every morning to the constant thud of metal drivers projecting golf balls over 200 yards down the driving range. It got me thinking about the way in which these recreational and amateur golfers were organising their practise time. It was clear that the majority were hitting drive after drive after drive (with varying degrees of success) in what appeared to be an attempt to ‘groove’ the swing.

The point here is that this type of practise schedule is not consistent with playing a round of golf, where you almost never hit two or three consecutive drives (unless the first ended up off bounds or in a lake). While it’s true that many coaches and players from a variety of different sports organise practise schedules in a similar way, the question is whether there’s sufficient research evidence to support the idea of ‘grooving’, or whether there are more beneficial ways of organising practise?

Blocked and random practise

If a golfer has 100 practise balls to hit on a driving range, there are numerous way of organising this practise. He or she could choose to hit 25 balls with the driver, followed by three more blocks of 25 balls with the 3-iron, the 7-iron and the pitching wedge in sequence. As the name suggests, this is termed ‘blocked practise’ as the participant practises one skill or action repeatedly before moving on to the next.

An alternative approach would be to randomise the order in which the clubs are used (ie driver, pitching wedge, driver, 3-iron, 7-iron, pitching wedge, etc). In this way, the same amount of practise is achieved but in a ‘random practise’ fashion. There various types of schedule that are more or less random than the example given above, but the key factor, is that the same shot is not repeated over and over again.

There appears to be a high degree of consistency in terms of what the research tells us about how these practise schedules influence performance and learning of motor skills. Much of the research is from the 1980s and 1990s but this highlights the fact that this information does not always filter down to players and coaches.

Most studies that evaluated the differences between blocked and random practise assessed performance during the acquisition of skills, and then assessed the amount of learning using retention tests that were often completed a few days after practise ended. Studies using both laboratory tasks and sport-specific skills have reported remarkably consistent results that may surprise some players and coaches.

Most findings suggest that blocked practise leads to better performance during the acquisition of new skills, but that random practise is more effective in terms of learning and transfer of skills to new contexts (1, 2). In essence, the poorer initial performance using random practise actually benefits the learner in the long-term.

One early controlled study compared random and blocked practise schedules for beginners learning to serve at badminton (3). Female college students practised short, long and drive serves from the right service court. These women practised 3 times a week for three weeks, with 36 serves (12 serves of each type) completed in each session. Results revealed significantly better performances in retention trials, and in a transfer task (serving from the left service court) for those women who used random practise schedules.

Similar results have been obtained using expert performers, showing that this effect termed ‘contextual interference’ (see box 1) is not confined to novices. One study used expert baseball players who engaged in extra batting practise (45 extra hits) for 3 days a week, over five weeks (4). For each session, batters faced 15 fastballs, 15 curve balls and 15 changeups in either a random or blocked schedule. Performance was measured in relation to number of solid hits. The results perfectly illustrate the benefits of random practise for future performance benefits; while more solid hits were achieved during practise sessions with the blocked schedule, when a transfer task (simulated game conditions) was introduced, the number of solid hits increased far more by following random practise (57%) compared to blocked practise (25%).

Constant and variable practise

Two other important types of practise termed ‘constant’ and ‘variable’ practise are concerned with practising several different versions of the same skill. Constant practise appears very similar to blocked practise as it involves repetition. An example of constant practise in throwing would be participants throwing the same object, the same distance, to the same target (eg throwing and catching a cricket ball over a 20-meter distance).

In contrast, variable practise concerns repeatedly changing the parameters of a particular skill (eg throwing and catching a cricket ball over different distances, to moving and static targets). It is important to note that variable practise involves changing the parameters of one particular skill, whereas random practise is based around the order in which several different skills are practised. Variable practise is given central importance in a number of different theories of motor learning(1,5), which suggest it leads to better transfer to new tasks.

One particular study concerning the basketball free-throw performance of unskilled university students provides a good example of the importance of variable practise(6). Participants in this research practised free-throw shooting (40 attempts each day) over a period of three weeks. While constant practise conditions (free-throws from the free-throw line only) resulted in a 1% increase in performance following practise, various combinations of variable practise (shooting from in front, behind and to the left and right of the free-throw line) resulted in up to 11% improvements in free-throw shooting performance.

It appears that organising practises around random and variable schedules can offer significant learning benefits in comparison to blocked and constant practises. However, further benefits might also be possible by combining random and variable practise (1). For example, a cricket training session could include practising two or more different skills (throwing and catching) in random order that included constantly changing parameters (different distances, different trajectories, etc).
Recently, researchers presented a good example of how these types of practise can be used for progressive learning in football (7). An example of blocked/constant practise for early learning is that of grid practises utilising a single skill. Blocked/variable practise could consist of grid practises involving multiple skills. Random/constant practise might consist of a conditioned game that focuses on single skills. Finally, theoretically the most beneficial, providing the learners are sufficiently competent, involves match play in small-sided games and is a combination of random and variable practise.

Theoretical explanations

These findings, and others, pose the intriguing question of why random and variable practise appears to be so effective in terms of learning? Despite some similarities, it appears that random and variable practise is beneficial for different reasons. There are two theories that propose explanations for the effects of random practise:

Learners are able to compare and contrast the variations (such as short, long or drive serve in badminton) that makes each movement distinct from the other(1,8).
The second explanation (not incompatible with the first) suggests that learners partially or completely forget the action plan for a particular skill, because it is interspersed with other skills(1,2).

In short, in a blocked schedule, you can keep drawing on the same solution to the problem, with slight modification, each and every time. Returning to the golf example of the introduction, it is clear that constantly hitting balls on a driving range does not accurately simulate the target context (a round of golf on the golf course), which requires constant problem solving (choice of club, adjusting stance for sloping ground, reduced backswing, etc). Random practise forces the learner to rethink each problem. Both explanations suggest that more cognitive effort is required when using random practise.
Schema theory and the work of Richard Schmidt has been central to understanding the importance of variable practise in skill learning and human motor control(1,2). A schema is defined as ‘a set of rules, formed from abstracting information from related movement experiences that includes the underlying movement principle for this class of responses’(9).

It is proposed that variable practise aids the development of schemas, and allows the participants to learn much more than specific actions. When people practise throwing a cricket ball over a distance of 10, 20 and 30 metres, they are learning a movement pattern that can be generalised and recalled to produce throws of distances not yet practised (eg 15 or 25 metres). This is likely achieved once the learner can recognise and recall the connections between actions and outcomes. As such, there are some features called parameters that can be varied from one performance to another.
For example, throwing a cricket ball back to a wicketkeeper requires changes in parameters of the over-arm throw, depending on distance from target and environmental conditions such as strong winds or sunshine, which might require a low trajectory throw. Consequently the cricketer might need to throw with less force and adjust direction to achieve the desired result. For such skills, once the movement pattern has been learned, it is beneficial to move towards variable practise where the skill is transferred to a variety of different situations requiring altered parameters.

When blocked practise is appropriate

So does this research suggest we should abandon blocked and constant practise schedules? Well, there is evidence to suggest that children and beginners may be particularly suited to blocked practise (2,5). Furthermore, random and variable practise might not be the ideal choice for learners with low levels of self-efficacy, as poor initial performance can damage fragile confidence (5).
Learners with low self-efficacy might therefore benefit more from practise that is likely to bring some relatively quick signs of success (ie blocked practise) and thus build confidence. In the early stages of learning the beginner is trying to establish a basic movement pattern and develop what some theorists term a ‘motor program’. As such, blocked and constant practise schedules are likely to be of benefit as the learner first practises a new skill. Indeed, some theorists suggest until an appropriate coordinated pattern of movement has been properly established, variable practises should be avoided(5).

Moving from blocked to random practise

Box 2 shows an example of different practise schedules for a group of club level tennis players. In a 90-minute session, given warm-up, warm-down, and some conditioned games, there are 30 minutes available during the session for practise. Over three practise sessions, the coach wants to focus on developing three specific strokes identified as areas for improvement. For blocked practise, each session is dedicated entirely to the practise of one stroke. In contrast, the random and serial schedules involve practising all three strokes in one session. Serial practise falls between blocked and random practise and can act as a route of progression after participants have developed basic movement patterns.

Practise could be organised on three courts, with forehand drives played on court A, backhand slices on court B, and forehand volleys on court C.  Participants would rally using the stated strokes on one of these courts until a mistake was made. After this they either move courts in a set sequence (ABCABC – serial practise) or randomly alternate between courts (random practise).

Conclusion

The results of numerous experiments appear to contradict deeply rooted views of sports skill practise, which emphasise repetition and doing things over and over again to groove technique(3,4,6). It is true that lots of practise is required to develop expertise, but repetitiveness in practise is not the most effective approach (1,2). While evidence suggests that repetitive practise schedules often lead to rapid performance improvements during training, random and varied practises tend to produce greater adaptability, and transfer to target contexts more successfully. Many athletes appear to have all the skills when practising, but cannot always transfer these skills into the competitive environment. However, for children, beginners, and those individuals with lower levels of self-efficacy, it appears that blocked and constant practise schedules are valuable, at least in the early stages of learning where participants are developing schemas and building confidence.

Practical implications

• Practitioners teaching several skills or variations of a skill should generally look to progress from blocked and constant practice to random and variable practice once an approximation of the movement pattern has been learned;
• Practitioners should not be perturbed if random and variable practice results in lower levels of performance during practice;
• Blocked and constant practices are important for use with children, beginners and participants with low self-efficacy;
• Random and variable practices can be combined to facilitate even more learning.