A season long investigation into coaching behaviours as a function of practice state: the case of three collegiate coaches

Abstract

Coach behaviour, practice activities and the relationship between the two are seen as essential components in athlete learning and development. This study examined the coaching intervention behaviours employed by three collegiate coaches as a function of ‘practice state’ over a season long period. Systematic observation of these coaches' behaviour was conducted using the Coach Analysis Intervention System (CAIS). Coaches also participated in semi-structured interviews to explore how and why they used the intervention behaviours in the ways they did within the different practice states. Whilst the data from the systematic observation were analysed descriptively and inferentially, the interview data were analysed deductively. Although multivariate analysis of variance tests revealed significant differences for the practice state of two of the three coaches, follow-up analyses revealed that the main differences in coaching behaviour were between ‘other’ states when compared to ‘training’ and ‘playing’ states. The results demonstrated limited changes to coaching behaviour as a function of ‘practice state’ for the three coaches, intimating that the drivers of the coaches' design and implementation of practice sessions and the delivery of instruction were their existing ‘folk pedagogies’ rather than scientifically-based evidence.

Keywords::

• coach behaviour
• season long
• mixed methods
• folk pedagogies

Introduction

Coach behaviour, practice activities and the relationship between the two are seen as essential components in athlete learning and development (Ford, Yates, & Williams, 2010), with coach behaviour connecting player understanding with the concepts and skills of practice activities (Cushion, 2010). A significant body of research has focused specifically on coach behaviour during practice (Gilbert & Trudel, 2004) with the use of systematic observation tools consistently identifying ‘instruction’ as the most frequently used behaviour by coaches (e.g. Cushion & Jones, 2001; Ford et al., 2010; Kahan, 1999; Partington & Cushion, 2013; Pereira, Mesquita, & Graça, 2009). However, the relationship between coach behaviour and practice activity (i.e. type/form) is relatively under-researched. Lacy and Martin (1994) were the first to investigate this relationship, finding that coaches mostly used ‘skill-work’ (77.8%) in preparation for game play, whilst spending a small proportion of practice (7.7%) in games-based activities. More recently, Ford et al. (2010) examined coaches' behaviour in the context of ‘training’ and ‘playing’ form activities.1 Here, coaching behaviours' were also placed into two different categories; instructional (i.e. instruction, questioning, feedback and modelling), and support and encouragement (i.e. hustle, praise, scold, and silence). Time spent in ‘training form’ equated to 65% and ‘playing form’ 35%, with only minor variations between the different age and skill groups. Ford et al. (2010) also found that ‘instructional’ coach behaviour dominated regardless of practice type, with the authors suggesting that an over-reliance on this kind of behaviour limited opportunities for players to problem-solve (Ford et al., 2010). However, a key limitation was the examination of coach behaviour in isolation, with no insight into coaches' underpinning knowledge and understanding of coaching strategies (i.e. why they did what they did).

A deeper understanding of the behaviours and practice strategies used by coaches requires a greater consideration of the observed context and situation (van der Mars, 1989a). Interviews to support systematic observation increases the richness of data in light of the context they are collected (Smith & Cushion, 2006). They also allow for an examination of ongoing practice and the use of certain coaching behaviours and strategies (Cushion & Jones, 2001). For example, Partington and Cushion (2013) examined how in-practice behaviours varied between different ‘practice states.’ Combining data from systematic observation and interpretive interviews revealed ‘how coaches constructed and continue(d) to construct social reality, given their interests and purposes’ (p. 2). Similar to previous studies, high rates of instruction were recorded especially in ‘training’ forms, and high rates of silence in ‘playing’ forms. However, the coaches were unable to recognize or explain the difference utilized in the two activity, providing evidence that coaches have low levels of self-awareness, subsequently limiting opportunities for player learning (e.g. De Marco, Mancini, & West, 1997). Instead, research suggests that coaches' practice activities and behaviours are driven by their own set of beliefs or ‘folk pedagogies’ (Bruner, 1999) about the ways in which people learn or ought to learn. Simply, coaches' personal biographies impact the way practice is structured and the intervention behaviours' used (Partington & Cushion, 2013). These beliefs are powerful and enduring, and are based on the accumulation of both ‘tried and tested’ methods (Harvey, Cushion, & Massa-Gonzalez, 2010), and ‘the processes of intuition, emulation and tradition…rather than on empirical research’ (Williams & Hodges, 2005, p. 637). This has resulted in coaches' practice tending to follow a simple-to-complex structure, i.e. unopposed, isolated technical practice, progressing to a practice within a more complex context, culminating with some form of game play (Rink, 2005). Based on the theoretical foundations of behaviourism underpinned by a linear, process–product approach to learning, this has become the traditional paradigm of coaching (Cushion, Ford, & Williams, 2012; Williams & Hodges, 2005). These patterns of coaching tend to be relatively stable and, irrespective of the needs of athletes, only minor differences have been found as a function of the age or skill level of the players coached (Cushion et al., 2012).

Consequently, not only do coaches generally lack an understanding and awareness of their own behaviour, and, in particular, how this changes (or not) within different types of practice or ‘practice states’ (Jolly, 2010; Partington & Cushion, 2013), it also leaves practice to be guided by a ‘traditional’ approach rather than an understanding and awareness of the needs of learners. Importantly, the link between the instructional behaviours utilized by coaches (the what) and how and why these are employed within the different types of practice has been largely ignored (Janelle & Hillman, 2003). Despite this, evidence suggests that traditional, deep-seated practice can be challenged when coaches connect athlete understanding with the concepts and skills relevant to the objectives of sessions (Cushion et al., 2012; Harvey et al., 2010). Therefore, coaches require the tools to reflect upon the goals of their session and the more general environment to determine the relevance of given behaviours and practices. This, in turn, would allow coaches to critically consider their practice and the extent they align coaching behaviour, practice structure and coaching objectives (Cushion et al., 2012). However, at the same time, it is also difficult to be overly prescriptive, as coaching behaviour is ‘very situation specific and dependent on the interaction of a myriad of influencing contextual variables’ (Jones, 1997, p. 30). Indeed, just because a ‘successful’ coach uses a specific behaviour/practice type frequently, or coaches in a particular ‘style’ or manner, does not mean that it will be applicable or effective for another coach in a different context. Coach behaviours and practice structures need to be responsive to individual differences in the needs of learners with an appropriately diverse range of approaches (Cushion et al., 2012). Thus, research has a role to play through the continuing examination of behaviour and practice in coaching, as well as the underlying cognitions or thought processes informing that practice (Cushion et al., 2012). However, research can be developed by moving to an examination of the interaction of knowledge, thought, and behaviour as manifested in coaches' practice (Cushion, 2010; Cushion et al., 2012).

Importantly, the structure and content of coaching practice will differ between different sports and contexts (e.g. differing situations of competition, organization and training) (Côté, Salmela, Trudel, Baria, & Russel, 1995; Cushion et al., 2012). Furthermore, within sessions different instructional strategies can be employed to adjust the cognitive load of the task for different individuals (Pereira et al., 2009). Therefore, despite the large quantity of behavioural research carried out, when divided into context specific studies, the ability to draw meaningful comparisons from the work seems limited (Cushion et al., 2012). In reality, conclusions have been drawn, often regarding coach effectiveness, based on only a few hours of observation (Cushion et al., 2012). The existing picture of behavioural research can be enhanced both by replicating and building on studies in similar and different contexts through more in-depth season long and/or longitudinal work, and by using more sophisticated, reliable and valid, analytical tools. Moreover, to fully understand the holistic nature of coaching, research should focus on the world of individual coaches, and how they operate within given contexts (Harvey et al., 2010; Potrac, Jones, & Armour, 2002). To this end, research should address individual coaches' interpretations of their experiences and the process by which meanings and knowledge are used to guide actions (Potrac et al., 2002). Therefore, the aim of this study was to examine the relationship between practice state and coach behaviour in three collegiate coaches over the course of a season. The significance of the work lies in generating a more in-depth and nuanced understanding of the coaching process through the provision of context specific coaching research that considers not just the what and why of different practices states and coaching behaviours,' but also how these are employed in relation to athlete learning or are shaped by other means.

Materials and methods

Participants and setting

Three, male collegiate coaches from field hockey, volleyball and basketball participated in the season long study. Each was both the head coach and Director of their respective sports within a UK university.2 This sample of coaches was both purposefully and criterion sampled; that is, the coaches were selected to provide the most information rich data possible, and were required to meet the specific criteria (of coaching full-time in a collegiate setting, being from diverse sports, and were willing to be tracked over a the course of a full season). While participant numbers have little to do with the adequacy or quality of data (Morrow, 2005) this study followed the recommendations of Pollio, Henley, and Thompson (1997) who suggested that the situational diversity necessary for identifying thematic patterns can be provided by three to five participants. This sample then offered a way of understanding the various nuances, contrasts and patterns of coaches' behaviour, focussing on providing a ‘holistic description and explanation’ (Berg, 2007, p. 284) of such behaviour.

Field hockey coach

Stuart (all names used are pseudonyms) coached the women's 1st team. He had 11 years of coaching experience in clubs and more recently as a field hockey development officer. He had degree level education and a level two3 field hockey National Governing Body (NGB) coaching qualification. This was his first ‘performance’ related role with pressures and expectations to produce a winning team. It was also his first ‘full-time’ role at the university. Stuart believed his main function to be to develop players' skill levels, and instil confidence and self-belief into them as athletes.

Volleyball coach

Martin, a former international volleyball player, coached the men's 1st team, and had four years coaching experience. He had a level one³ volleyball coaching qualification and was engaged in postgraduate study in sports coaching. He was formerly an assistant coach at a college in the United States, and had spent three years coaching a women's Age Group (i.e. under 18) volleyball team alongside being a National Age Group coach. Martin described many of the players in his current programme as novices as they had limited playing experience. Therefore, his primary objective was to allow players the opportunity to increase their playing experiences and develop a better understanding of volleyball.

Basketball coach

Will (pseudonym) had coached the men's 1st team for two years, with this being the first year that he was able to recruit players. Previously, he had coached children in schools and on summer camps alongside coaching a National League team. In total, he had 15 years of coaching experience. He had a degree level education and held a level three³ NGB basketball award. Having a range of players from different cultures, background and playing experiences, impacted on Will's expectations of his players and on his coaching: ‘…a balance between challenging the players of a higher ability, yet ensuring that the lower ability players understand the messages being conveyed.’

Procedures

Data were collected through quantitative (systematic observation) and qualitative methods (interpretive interviews) to provide breadth and depth of data, in addition to differential layers of collaborative evidence or ‘triangulation’ (Miller & Glassner, 1997). This enabled an in-depth analysis of evolving practice for each coach.

Systematic observation

The instrument used to collect behavioural data was the Coach Analysis and Intervention System (CAIS), which had previously demonstrated construct and face validity in each of the sports used in this study (see Cushion, Harvey, Muir, and Nelson, 2012). This is a computerized systematic observation tool with the capability to analyse across five levels of simultaneous behaviour (primary behaviour plus secondary detail, context, recipient, timing, content), and to further contextualize this behaviour into three different forms: (a) ‘training’; (b) ‘playing’; and (c) ‘other.’4 The ‘full’ version of the CAIS has 23 primary behaviours.' However in line with Ford et al. (2010) study examining the relationship of coaching behaviour and practice type, this study focused on 11 key ‘instructional’ behaviours within the CAIS in addition to the associated secondary behaviours, which are outlined in Tables 123456. These behaviours were used as they have been consistently reported as the most utilized by coaches (Cushion, 2010; Potrac et al., 2002; Smith & Cushion, 2006). Importantly, the resulting systematic observation examined the relationship between these behaviours and practice states, alongside the secondary detail associated with these behaviours (i.e. recipient, timing, content). The study, therefore, used the original behavioural definitions for the CAIS presented by Cushion, Harvey, et al. (2012), and included the examination of both primary and secondary detail associated with these behaviours. Primary behaviours included feedback (general, specific and corrective), instruction, and questioning while the secondary detail added descriptions of these. For example, corrective feedback could have been given concurrently (i.e. while completing the action), directed to an individual (i.e. recipient), and/or was technical in nature (i.e. content). Moreover, questioning could have been directed to the whole team (i.e. recipient), tactical (i.e. content) and convergent or divergent (i.e. closed or open) (see Cushion, Harvey, et al. [2012] for more detail).

Table 1 Table showing the percentage and Rate Per Minute of ‘primary’ coaching behaviours used by Stuart a collegiate field hockey coach, over the course of a season.

					Practice State
		ANOVA's (df 2, 30)	Train (N = 11)	Play (N = 11)	Other (N = 11)	Totals
Behaviour		F	p	η^2	M	SD	M	SD	M	SD	M	SD
Instruction	%				16.04	8.56	10.85	7.17	0.33	0.50	27.22	16.23
	***RPM	26.12	.01	.64	1.34^b	0.55	0.96^c	0.49	0.05^b,c	0.09	2.35
Gen Feedback Pos	%				11.27	4.42	3.92	3.62	0.44	0.67	15.62	8.71
	***RPM	23.64	.01	.61	1.01^a,b	0.52	0.32^a,c	0.20	0.08^b,c	0.13	1.42
Silence	%				6.13	3.10	6.95	4.64	0.28	0.38	13.35	8.12
	***RPM	15.70	.01	.51	0.51^b	0.27	0.69^c	0.41	0.03^b,c	0.03	1.23
Spec Feedback Pos	%				7.31	3.71	5.02	3.99	1.08	0.75	13.41	8.46
	***RPM	5.79	.01	.28	0.66^b	0.47	0.45	0.32	0.17^b	0.15	1.27
Questioning	%				4.23	2.83	2.64	1.70	2.78	1.47	9.67	6.00
	RPM	2.81	.08	.16	0.34	0.18	0.24	0.10	0.38	0.15	0.95
Corrective Feedback	%				3.90	2.38	2.13	2.25	1.44	1.56	7.46	6.18
	RPM	2.04	.15	.12	0.33	0.17	0.21	0.24	0.19	0.12	0.73
Spec Feedback Neg	%				3.12	3.01	1.92	1.12	1.13	1.05	6.16	5.18
	RPM	1.33	.28	.08	0.22	0.17	0.18	0.09	0.14	0.08	0.55
Positive Modelling	%				2.98	3.12	0.67	0.83	0.73	0.75	4.38	4.70
	RPM	2.58	.09	.15	0.30	0.47	0.05	0.05	0.09	0.08	0.43
Gen Feedback Neg	%				0.84	0.86	0.46	0.54	0.10	0.25	1.40	1.70
	RPM	2.84	.07	.16	0.07	0.07	0.04	0.04	0.02	0.05	0.14
Negative Modelling	%				0.86	0.89	0.32	0.51	0.13	0.19	1.32	1.60
	*RPM	4.41	.02	.23	0.79	0.07	0.03	0.03	0.02	0.03	0.84
Physical Assistance	%				0.00	0.00	0.00	0.00	0.02	0.08	0.02	0.80
	RPM	1.00	.38	.06	0.00	0.00	0.00	0.00	0.01	0.02	0.01
Totals	%				56.65		34.86		8.46		100.00
	RPM				5.57		3.16		1.17		9.90

Notes: ***significant difference in one-way analysis of variance test (p < .001), **significant difference in one-way analysis of variance test (p < .01). *significant difference in one-way analysis of variance test (p < .05).

a significant difference in mean scores between training and playing form (p < .05).

b significant difference in mean scores between training and other form (p <  .05).

c significant difference in mean scores between playing and other (p < .05)

Table 2 Table showing the percentage and Rate Per Minute of ‘primary’ coaching behaviours used by Martin, a collegiate volleyball coach, over the course of a season.

				Practice State
		ANOVA's (df 2, 18)	Train (N = 7)	Play (N = 7)	Other (N = 7)	Totals
Behaviour		F	p	η^2	M	SD	M	SD	M	SD	M	SD
Silence	%				11.45	6.44	7.47	8.95	0.75	0.65	19.67	16.03
	**RPM	10.64	.00	.54	0.85^b	0.40	0.53^c	0.28	0.14^b, c	0.12	1.52
Corrective Feedback	%				11.07	4.99	3.73	0.82	1.67	1.17	16.48	6.97
	**RPM	5.35	.02	.37	0.88^b	0.49	0.40	0.28	0.32^b	0.17	1.61
Spec Feedback Pos	%				5.76	3.49	6.44	5.03	1.65	1.51	13.84	10.4
	RPM	.91	.42	.09	0.42	0.23	0.53	0.29	0.34	0.30	1.295
Instruction	%				6.67	2.96	5.56	5.17	0.17	0.45	12.40	8.58
	**RPM	10.74	.00	.54	0.51^b	0.18	0.41^c	0.23	0.06^b, c	0.15	0.98
Gen Feedback Pos	%				8.79	3.97	3.03	2.80	0.33	0.34	12.14	6.38
	*RPM	8.85	.00	.50	0.69^b	0.38	0.31	0.28	0.07^b	0.08	1.07
Spec Feedback Neg	%				6.09	3.27	3.49	1.40	1.01	1.10	10.59	5.77
	RPM	1.76	.20	.16	0.48	0.31	0.43	0.41	0.19	0.19	1.09
Positive Modelling	%				5.57	3.76	0.74	0.79	0.93	0.93	7.24	5.48
	RPM	3.01	.08	.25	0.46	0.35	0.12	0.16	0.23	0.25	0.80
Questioning	%				2.01	2.13	1.39	0.77	0.65	0.74	4.06	3.64
	RPM	.06	.95	.01	0.14	0.14	0.13	0.07	0.12	0.14	0.40
Negative Modelling	%				1.92	1.74	0.30	0.34	0.11	0.20	2.33	2.28
	*RPM	3.94	.04	.31	0.16	0.15	0.04	0.04	0.03	0.05	0.22
Gen Feedback Neg	%				0.30	0.31	0.18	0.35	0.29	0.44	0.78	1.10
	RPM	.27	.77	.03	0.03	0.03	0.03	0.07	0.05	0.06	0.10
Physical Assistance	%				0.44	0.75	0.05	0.12	0.00	0.00	0.48	0.87
	RPM	2.05	.16	.19	0.04	0.07	0.00	0.01	0.00	0.00	0.04
Totals	%				60.07		32.37		7.56		100.00
	RPM				4.65		2.93		1.54		9.12

Notes: ***significant difference in one-way analysis of variance test (p < .001), **significant difference in one-way analysis of variance test (p < .01). *significant difference in one-way analysis of variance test (p < .05).

a significant difference in mean scores between training and playing form (p < .05).

b significant difference in mean scores between training and other form (p <  .05).

c significant difference in mean scores between playing and other (p < .05)

Table 3 Table showing the percentage and Rate Per Minute of ‘primary’ coaching behaviours used by Will, a collegiate basketball coach, over the course of a season.

				Practice State
		ANOVA's (df 2, 21)	Train (N = 8)	Play (N = 8)	Other (N = 8)	Totals
Behaviour		F	p	η^2	M	SD	M	SD	M	SD	M	SD
Instruction	%				20.63	11.61	16.13	7.76	0.46	0.37	37.23	19.73
	RPM	n/a	n/a	n/a	1.70	1.10	1.45	0.65	0.06	0.04	3.21
Silence	%				7.36	5.10	8.44	4.44	0.86	0.79	16.66	10.33
	RPM	n/a	n/a	n/a	0.57	0.34	0.77	0.29	0.11	0.09	1.45
Questioning	%				5.31	2.72	4.86	3.16	3.75	2.46	13.92	8.33
	RPM	n/a	n/a	n/a	0.46	0.22	0.49	0.23	0.49	0.21	1.44
Gen Feedback Pos	%				5.41	3.45	2.97	2.17	0.56	0.29	8.93	5.92
	RPM	n/a	n/a	n/a	0.46	0.34	0.31	0.19	0.08	0.04	0.85
Spec Feedback Neg	%				2.11	1.40	2.97	1.55	2.23	1.97	7.31	4.92
	RPM	n/a	n/a	n/a	0.16	0.10	0.25	0.12	0.28	0.16	0.69
Spec Feedback Pos	%				2.12	1.69	3.83	3.86	1.03	0.52	6.98	6.07
	RPM	n/a	n/a	n/a	0.16	0.11	0.28	0.17	0.16	0.12	0.59
Corrective Feedback	%				2.51	1.31	1.60	0.99	0.44	0.44	4.55	2.73
	RPM	n/a	n/a	n/a	0.20	0.11	0.15	0.13	0.08	0.09	0.43
Gen Feedback Neg	%				0.79	0.72	1.25	0.91	0.68	1.21	2.71	2.84
	RPM	n/a	n/a	n/a	0.07	0.07	0.10	0.06	0.06	0.01	0.23
Positive Modelling	%				0.87	0.54	0.20	0.41	0.08	0.15	1.13	1.09
	RPM	n/a	n/a	n/a	0.07	0.04	0.02	0.03	0.01	0.02	0.09
Negative Modelling	%				0.23	0.32	0.02	0.07	0.05	0.13	0.29	0.51
	RPM	n/a	n/a	n/a	0.02	0.02	0.00	0.01	0.00	0.01	0.02
Physical Assistance	%				0.12	0.17	0.12	0.13	0.06	0.12	0.29	0.41
	RPM	n/a	n/a	n/a	0.01	0.01	0.01	0.01	0.01	0.02	0.03
Totals	%				47.45		42.38		10.17		100.0
	RPM				3.87		3.83		1.34		9.04

Table 4 Table showing the percentage of ‘secondary’ behaviours used by Stuart, a collegiate field hockey coach, over the course of a season.

	Practice State
	Train (N = 11)	Play (N = 11)	Other (N = 11)	Totals
	M	SD	M	SD	M	SD	M	SD
Recipient
Team	9.42	7.86	4.70	3.62	4.69	3.69	18.82	15.17
Group	9.66	7.58	6.39	6.72	0.77	1.25	16.82	15.54
Individual	31.41	16.14	16.65	11.11	2.44	0.83	50.50	28.08
Other	0.09	0.21	0.18	0.35	0.27	0.41	0.54	0.96
Totals	50.58		27.92		8.18		86.68
Timing
Pre	1.88	1.78	0.96	1.50	0.50	0.62	3.34	3.88
Concurrent	36.32	16.69	21.11	14.25	0.63	0.91	58.07	31.85
Post	8.10	3.65	3.21	2.83	3.93	2.10	15.24	8.58
Totals	46.23		25.28		5.07		76.65
Content
Technical	40.26	17.73	18.20	12.49	5.70	4.23	64.15	34.45
Tactical	6.88	8.31	7.87	7.17	1.61	1.76	16.36	17.25
Other	3.44	4.27	1.86	2.24	0.87	1.31	6.17	7.81
Totals	50.58		27.92		8.18		86.68
Questioning
Convergent	3.64	2.49	2.31	1.47	2.24	1.15	8.19	2.71
Divergent	0.61	0.66	0.33	0.36	0.54	0.87	1.48	1.08
Totals	4.26		2.64		2.78		9.67

Table 5 Table showing the percentage of ‘secondary’ behaviours used by Martin, a collegiate volleyball coach, over the course of a season.

	Practice State
	Train (N = 11)	Play (N = 11)	Other (N = 11)	Totals
	M	SD	M	SD	M	SD	M	SD
Recipient
Team	8.47	2.54	3.50	2.12	5.36	3.90	17.33	8.55
Group	5.28	4.65	3.95	3.61	0.22	0.28	9.45	8.55
Individual	34.78	17.34	17.30	7.61	1.20	0.89	53.28	5.84
Other	0.10	0.17	0.14	0.28	0.03	0.09	0.27	0.53
Totals	48.63		24.89		6.81		80.33
Timing
Pre	2.34	1.85	0.25	0.48	1.03	0.72	3.62	3.04
Concurrent	35.81	13.96	20.66	11.20	0.61	0.46	57.09	25.62
Post	8.46	5.21	2.59	2.64	4.51	3.47	15.56	11.32
Totals	46.61		23.50		6.16		76.27
Content
Technical	44.01	18.10	21.77	10.93	4.81	3.59	70.60	32.62
Tactical	1.34	1.80	2.02	1.77	0.77	0.91	4.13	4.48
Other	3.27	3.83	1.10	1.12	1.22	0.92	5.60	5.85
Totals	48.63		24.90		6.81		80.33
Questioning
Convergent	1.61	2.10	1.17	0.93	0.44	0.47	3.21	2.01
Divergent	0.40	0.31	0.22	0.59	0.22	0.31	0.84	0.68
Totals	2.01		1.39		0.65		4.06

Table 6 Table showing the percentage of ‘secondary’ behaviours used by Will, a collegiate basketball coach, over the course of a season.

	Practice State
	Train (N = 11)	Play (N = 11)	Other (N = 11)	Totals
	M	SD	M	SD	M	SD	M	SD
Recipient
Team	5.82	3.18	9.02	9.09	6.13	4.06	20.97	16.32
Group	13.73	10.21	11.12	5.13	0.54	0.62	25.39	16.95
Individual	19.72	12.61	13.58	6.62	2.43	1.69	35.71	20.93
Other	0.82	1.21	0.23	0.25	0.24	0.48	1.29	1.93
Totals	40.09		33.94		9.32		84.34
Timing
Pre	1.20	1.29	0.60	0.74	0.41	0.38	2.21	2.41
Concurrent	30.59	15.88	25.12	14.31	0.50	0.60	56.21	30.80
Post	2.99	1.86	3.35	3.12	4.66	2.76	11.01	7.73
Totals	34.78		29.08		5.57		64.43
Content
Technical	19.16	12.03	15.62	9.65	2.56	2.39	37.34	24.07
Tactical	7.24	5.43	10.86	5.32	1.05	1.07	19.15	11.82
Other	13.68	5.91	7.46	7.93	5.71	3.07	26.85	16.82
Totals	40.09		33.94		9.32		83.34
Questioning
Convergent	5.07	2.82	4.42	2.57	3.45	2.20	12.94	3.99
Divergent	0.24	0.18	0.47	0.72	0.30	0.28	1.01	0.96
Totals	5.31		4.88		3.75		13.95

Following ethics committee approval, and coach and performers' informed consent, coaches were filmed in a total of 26 training sessions (n = 11 field hockey; n = 7 volleyball; n = 8 basketball) over the length of one season/year. Session durations averaged M = 90.92 minutes (SD = 16.45), M = 76.14 minutes (SD = 14.96), M =  113.13 minutes (SD = 14.55), for field hockey, volleyball and basketball, respectively. As coaching behaviour has been shown to alter during the course of a season principally in relation to team outcomes (Potrac et al., 2002), observations taken at a single phase of the season only provide a ‘snap shot’ of a practitioner's coaching behaviour at a particular time. By comparison, observations spread over the length of a playing season provide a better means for obtaining a more accurate account of a coach's pedagogical strategies (Kahan, 1999; Potrac et al., 2002). Consequently, to develop a comprehensive description, the coaches were observed for the duration of a season.

Video recordings were digitally captured and three coders trained to use CAIS (Cushion, Harvey, et al. (2012) populated and saved coded behaviours to a timeline. Intra- and inter-observer reliability checks were completed on approximately 30% of the data. This was in line with Baumgartner, Jackson, Mahar, and Rowe (2007) recommendations with checks completed on the primary behaviours, the secondary detail, as well as the training ‘state’ durations (i.e. ‘training’, ‘playing’ and ‘other’). Results revealed intra- and inter-observer agreement levels of above 80% (van der Mars, 1989b). In addition, a two-way (coder x session) analysis of variance test on these data was non-significant, suggesting that differences in data coding were not attributable to individual coder variability (Williams, Ward, Smeeton, & Allen, 2004).

Interpretive interviews

Interpretive interviews enabled an exploration of how and why the coaches structured practice and used intervention behaviours in a particular way within such practices. This gave an in-depth explanation of the reasons why they coached the way they did in their respective context (van der Mars, 1989a).

Each coach was interviewed twice, once during the competition season and once at its conclusion. The first interview had two parts. The first explored coaches' biographies. Here, the coaches were asked how long they had been coaching, about their previous playing experience and to identify significant others from their coaching experiences. Part two of the interview focused on the coaches, understanding of their current practices, their level of self-awareness of these practices, and the potential factors that influenced them.

The second interview, conducted at the conclusion of the study, was also separated into two parts. In part one, the coaches were given the opportunity to talk about their current coaching behaviour in the context of those coded by the CAIS. For example, the coaches were asked to talk about whether their behaviour was different depending on the ‘practice state.’ They were also asked about which primary and secondary behaviours they felt they used, and what types of secondary interventions they employed. After this was completed, the second part of the interview was based on the findings of the quantitative research. Behavioural data from their own coaching sessions were presented to the coaches, with the subsequent interview focusing on their reflections about the data and what the latter demonstrated about the coaching that took place.

Data analysis

Time spent in each practice ‘state’

The length of time of each coaching session was recorded, with the time spent each of the three ‘practice states’ being subsequently calculated. The percentage time spent in the three different practice ‘states’ was analysed using a one-way analysis of variance (ANOVA).

Coaching behaviours

Each coach's data were analysed independently, within each of the three practice states, resulting in three sets of descriptive statistics (i.e. means and standard deviations) for each of the 11 coach behaviours. These included: (a) total frequency of behaviours; (b) percentage of total behaviours in a session; and (c) Rates per Minute (RPM). This provided a comparison of the RPM's between each ‘practice state.’ For the secondary behaviours, descriptive statistics were calculated (i.e. means and standard deviations) for: (a) total frequency of behaviours; and (b) percentage of total behaviours in a session for each of the four categories of secondary behaviours (i.e. recipient, timing, content and questioning).

Before the one-way multivariate analysis of variance (MANOVA) tests were conducted, various assumption checks were performed on the data. To control for homogeneity of variances across groups, the test statistic chosen for interpretation was Pillai's Trace, which is the most robust against violations of the heterogeneity of variance assumption in MANOVA (Tabachnick & Fidell, 2007). Three one-way MANOVA tests were conducted to assess if there would be one or more mean differences in the coaching behaviours between practice states (‘training,’ ‘playing’ and ‘other’). If the first MANOVA yielded a significant result, then a series of one-way ANOVA's were conducted. Finally, a series of post hoc analyses examined individual mean difference comparisons across the coaching behaviour subscales in each of the three practice states. Inferential statistical analyses were completed using the RPM figures as the dependent variable.

Interview data

Interviews were tape-recorded and transcribed verbatim. In terms of analysis, the interview transcripts were firstly subjected to a line-by-line examination where meaningful and significant excerpts of text were highlighted. Secondly, these excerpts were grouped within similar sub-categories, a largely deductive process assisted by the predefined levels of behaviours (i.e. ‘practice states,’ primary behaviours and secondary behaviours) identified in the CAIS.

Results and analysis

This section presents individual coach data, and while it was not the aim of the study to ‘compare’ the three coaches, presenting results in this way shows the behavioural pattern of each coach (also reported in Tables 123456) as well as their rationale for acting as they did. To aid the presentation of the results, Tables 123 have been ordered according to the behaviours that occurred most often (by total percentage).

Stuart

Time use analysis

Stuart's practice sessions totalled 1000 minutes. On average, Stuart spent 41% of this time in the ‘training’ practice state (M = 41.45; SD = 18.11), 35% in the ‘playing’ practice state (M = 35.09, SD = 16.12), and nearly a quarter of session time in the ‘other’ practice state (M = 23.36, SD = 8.30). The difference between the time spent in the three practice states was significant, F(2,30) = 4.23, p = 0.02. However, post hoc analyses revealed the only significant difference was the time spent in the ‘other’ state, which was lower when compared to ‘training’ (M_diff = 18.09, p < 0.01).

Coaching behaviours

Five behaviours comprised almost 80% of Stuart's coaching (‘instruction,’ ‘general feedback positive,’ ‘silence,’ ‘specific feedback positive,’ and ‘questioning’). Of these, ‘instruction’ was the most employed in both in ‘training’ and ‘playing’ practice states (see Table 1). ‘General feedback positive’ was the second most employed behaviour overall, although used more in the ‘training’ practice state. ‘Specific feedback positive’ and ‘silence’ were the next highest used behaviours overall. ‘Silence’ had a slightly higher RPM and percentage in ‘playing’ practice states, while ‘specific feedback positive’ had a higher RPM and percentage in ‘training’ states. Contextualizing these with secondary behaviours, Stuart, irrespective of state, was predominantly ‘concurrent,’ ‘technical’ in nature and directed to individuals (see Table 4).

Results from the inferential analysis revealed a statistically significant MANOVA, Pillai's Trace = 1.39, F(22, 42) = 4.36, p < 0.001, effect size 0.70, implying that 70% of the variance was accounted for by practice state. As a result, one-way ANOVA's on the RPM data were completed which found significant differences between practice states for five behaviours: ‘negative modelling,’ ‘silence,’ ‘general feedback positive,’ ‘specific feedback positive’ and ‘instruction’ (see Table 1). However, while these initial analyses suggested differences in Stuart's behaviour between states, post hoc tests showed that only one behaviour, ‘general feedback positive,’ was significant when comparing ‘playing’ practice states to ‘training,’ with this being lower in the ‘playing’ state (see Table 1). Significant differences between ‘other’ states when compared to ‘training’ and ‘playing’ states were noted for ‘silence,’ ‘general feedback positive,’ ‘specific feedback positive’ and ‘instruction’ (‘training’) and ‘silence’ and ‘instruction’ (‘playing’), with all these behaviours significantly lower in ‘other’ states when compared to training (see Table 1).

The lack of significant differences between Stuart's behaviour in ‘training’ and ‘practice’ states reflected the fact that the top five behaviours in each of these two states were the same (i.e. ‘instruction,’ ‘general feedback positive,’ ‘specific feedback positive,’ ‘silence,’ and ‘questioning’). In both interviews, Stuart commented on ‘how poor the [players'] basic techniques were’ resulting in him offering players a ‘basic plan in skills development’ giving a technical focus to his coaching. This focus (see Table 4) was consistent with his past experiences of coaching field hockey in schools as well as his desire to reinforce the technical skills of field hockey via recent ‘information from England Hockey’ (the English field hockey governing body) where he also worked as a performance coach and coach educator. Instilling these technical skills were paramount to the on-going development of the players, explaining why Stuart spent more time in technical practice. Here, he would ‘work for 20 minutes on this [a technique] and then go on to how its related to the game.’ The caveat to this ‘jigsaw’ was, of course, if the players did not get the first part right, ‘it was pointless moving on to the next bit [the game].’

In the second interview, Stuart reflected on his approach and noted that he ‘led the programme,’ was ‘talking a lot,’ giving ‘loads of instruction’ by ‘bark[ing] orders from the sideline or tell[ing] people how it is’ most of which he suggested ‘fell on deaf ears.’ Indeed, Stuart noted that ‘I suppose I was pretty damning in some of the way that I spoke with the girls, because there was a time pressure on top of all of this’ further highlighting that he had ‘spoon-fed and…[then]…got annoyed that people aren't meeting my high expectations.’ Stuart recognized that the high levels of instruction aligned with his previous experiences as a player, where he perceived himself to have been treated poorly.

Moderately high levels of positive feedback were another key aspect of the learning environment observed. Despite recognizing that high levels of ‘general feedback positive’ ‘meant nothing’ to the players, this was not a deliberate coaching strategy by Stuart. Rather, he viewed as something ‘…that has come from where I started as a coach, which was primary school level and it was all really fun…and about having bums on seats and keeping people engaged.’ In the second interview, Stuart noted that the players in his current environment should ‘still enjoy training and have a laugh…but with direction and specific knowledge and understanding of how they can improve.’ Moving forward, he admitted a need to ‘empower the players more…[and]…stop instructing’, allowing players to ‘work more as groups’ to ‘problem solve in situations which are more game realistic.’ In addition, he recognized that ‘if I am going to develop players to…play at the higher level, I cannot stand there and control and instruct them.’

The lack of significant differences in Stuart's behaviour between states, particularly between the ‘training’ and ‘playing’ were linked to a ‘reactive’ planning approach which ‘changed weekly’ depending on the ‘availability of the players for the team’ and ‘weather conditions.’ This reactive approach also resulted in an admission that he, at times, needed to ‘just go into game play to allow them to just play hockey.’ And while ‘playing’ practice states revealed lower rates per minute of ‘instruction’ and ‘general feedback’ and slightly higher rates of ‘silence,’ the specific planning of, and rationale for, this change in behaviour remained unarticulated. Hopper (2002) noted that players need ‘progressive and skilful’ instruction during games to prevent these becoming times when the coach simply ‘rolls out the ball’ (Metzler, 2011) in the implicit belief that players will automatically transfer learning from technical skill drills to games without guidance from the coach.

A seemingly problematic issue with Stuart's coaching was the amount of time in ‘other’ states, which was sometimes as high as 40% of total session time. Stuart admitted there were ‘timing issues’ in his sessions ‘because I do talk.’ Having said that, Stuart also used this time deliberately: (a) to give his players time to ‘go and discuss and talk about things, apart from hockey’; and (b) to engage with players to ‘get them on board…to call someone over’ and ask questions such as ‘…when you're getting the ball, what are your options? Where do you go?,’ as well as giving them supplementary feedback in a variety of ways. Far from being time off task then, it could be argued that such a state incorporated some crucial facets of coaching.

Martin

Time use analysis

Martin's practice session totalled 533 minutes. Of all the three coaches featured, he spent the greatest amount of time in the ‘training’ practice state (M = 45.29, SD = 12.69) and the most amount of time in the ‘playing’ practice state (M = 39.14, SD = 12.02). This meant he spent the least amount of time of all three coaches in the ‘other’ practice state (M = 15.71, SD = 12.02). Results revealed significant differences between the time spent in the three practice states, F(2,18) = 16.10, p < 0.001. However, post-hoc analyses revealed the only significantly differences between states were the percentage time spent in the ‘other’ state which was lower when compared to both ‘training’ (M_diff = 29.57, p < 0.001) and ‘playing’ (M_diff = 23.43, p < 0.001).

Coaching behaviours

Five behaviours (‘silence,’ ‘corrective feedback,’ ‘specific feedback positive,’ ‘instruction,’ and ‘general feedback positive’) comprised nearly 75% of Martin's behaviour. ‘Silence’ was the largest behaviour for Martin overall, in both ‘training’ and ‘playing’ states (see Table 2). ‘Corrective feedback’ was the second largest behaviour overall, and was the highest and second highest behaviour for Martin in both ‘other’ and ‘training’ practice states, respectively. ‘Specific feedback positive’ was the third largest behaviour, and the second most employed behaviour in ‘playing’ and ‘other’ practice states. ‘General feedback positive,’ ‘instruction’ and ‘specific feedback negative’ were the next most employed behaviours in the ‘training,’ ‘playing’ and ‘other’ states, respectively. In relation to secondary behaviours, Martin's actions were ‘technical’ in nature, ‘concurrent’ and ‘directed to individuals’ irrespective of state (see Table 5).

The inferential analysis revealed a statistically significant MANOVA, Pillai's Trace = 1.60, F(22, 18) = 3.29, p = 0.006, effect size 0.80, which implies that 80% of the variance was accounted for by practice state. As a result, a series of one-way ANOVA's on the RPM data were completed (see Table 2) and significant differences were found between practice states for five behaviours: ‘negative modelling,’ ‘silence,’ ‘corrective feedback,’ ‘general feedback positive,’ and ‘instruction’ (see Table 2). However, while initial analyses suggested differences in Martin's behaviour between states, post hoc tests showed there were no significant differences in the coach's behaviour between ‘training’ and ‘playing.’ Indeed, the only significant differences were between ‘other’ states when compared to ‘training’ where the RPMs of ‘silence,’ ‘corrective feedback,’ ‘general feedback positive’ and ‘instruction’ were significantly lower, and between ‘other’ states when compared to ‘playing’ where instruction was significantly lower (see Table 2).

There were few differences in behaviour between ‘training’ and ‘playing’ states, because all but one of the behaviours used in each state was the same. Martin admitted his coaching was ‘very didactic, very prescriptive. I wasn't at the stage where the players were given autonomy and the opportunity to express themselves very much…. [I suppose] I wasn't meeting individual players' needs as I was predominantly doing all the work and imparting information.’ In Martin's words the players were ‘beginners’ who needed to work on ‘ball control and technique.’ This resulted in him ‘probably trying to fix too much’ and, at times, ‘not being overly pleased with the players’ as they attempted to meet the ‘vision of how I want the game to be played.’

Martin linked his behaviours to his ‘playing vision’ and the needs of the players in the context at that particular time. However, these behaviours also resulted from prior experiences and socialization into a ‘progressive type of coaching’ session or ‘teaching model’ from a previous PE teacher and coach. This teacher had a ‘massive influence’ to the extent where Martin ‘probably copied quite a lot of what he did, even in terms of personality.’ This included a practice structure model incorporating a ‘warm-up, a sort of introduction of a theme, a bit of teaching on it, but it will still be a game context and then put it into a competition phase.’ Martin ‘liked’ this as it allowed him to ‘introduce it simple, and progressively make it harder.’ Martin also commented on other approaches such as ‘whole-part-whole’ or games based practice as a ‘waste of time,’ arguing that ‘without the technique you're never going to be able to execute the tactical component.’ So, only once his players got ‘the toolbox…complete enough’ could he ‘set the context then stand back and let them do it.’ Martin also commented how this approach was reinforced through the influence of other coaches and, in particular, those whose beliefs aligned with his own developing ones. Such beliefs were manifest in the provision of ‘technical’ and ‘corrective feedback’.

Prior to the study Martin admitted to not really considering the ways in which he structured practice, and behaved as he thought this was correct ‘because I'm copying other coaches…in my head I know what I am looking for in sessions and perhaps I don't plan, as in writing it down.’ However, in the second interview, he revealed that studying for a part-time master's degree, in addition to undertaking a number of professional development initiatives alongside his coaching has resulted in more time reflecting on the impact of his behaviour. He specifically detailed how he had been influenced by notions from self-determination theory. This, and the ‘different [National Level] coaches I have worked with’ had, therefore, begun to change his perception that coaches always have to ‘justify their role,’ making him, for example, ‘more comfortable with silence’ as a deliberate coaching strategy.

Will

Time use analysis

Will's practice sessions totalled 905 minutes. On average, he spent the least amount of time in the ‘training’ practice state of all three coaches (M = 40.50, SD = 13.66) and, like Stuart, nearly a quarter of session time in the ‘other’ practice state (M = 23.88, SD = 6.69). Will also spent similar amounts of time in the ‘playing’ practice state to that of Stuart (M = 35.75, SD = 15.35). Results revealed significant differences between the time spent in the three practice states, F(2,21) = 3.77, p < 0.05. Post hoc analyses, however, showed that the only significant difference was the percentage time spent in the ‘other’ state, which was lower when compared to ‘training’ (M_diff = 16.63, p < 0.05).

Coaching behaviours

Five behaviours (‘instruction,’ ‘silence,’ ‘questioning,’ ‘general feedback positive,’ and ‘specific feedback positive’) comprised nearly 85% of Will's behaviour. ‘Instruction’ was the most employed coaching behaviour overall both in ‘training’ and ‘playing’ practice states (see Table 3). ‘Silence’ was the second most employed behaviour overall, being the second highest behaviour for Will in the ‘other’ and ‘training’ practice states respectively. ‘Questioning’ was the third most employed total behaviour, with it the most used behaviour in ‘other’ practice states, and the third and fourth highest behaviour in ‘playing’ and ‘training’ practice states. ‘General feedback positive’ was the fourth most employed behaviour overall, with it being the third most employed behaviour in ‘training.’ ‘Specific feedback’ was the fifth highest behaviour, with its use being more prominent in ‘other’ practice states. A slightly different pattern of secondary intervention behaviours was noted for Will who, irrespective of state, used similar levels of ‘technical’ and other (i.e. psychological) content and directed behaviours ‘concurrently’ to the team and small groups, as well as individuals (see Table 6). The initial MANOVA was not significant, Pillai's Trace = 1.28, F(22, 24) = 1.95, p = 0.06, effect size 0.64. As a result, no post-hoc analyses on the RPM data were completed.

Will's perceptions of his coaching behaviours closely mirrored the systematic observation data. His practice, however, had been shaped by a number of previous experiences coaching in terms of ‘assimilating with practices that you feel successful in.’ For example, in the first interview, Will spoke about coaching at many basketball summer camps in the United States where he learned to teach ‘stations and games, little clinic sessions’ taking ‘ownership of different parts of the day…footwork, post development, it might be shooting.’

During both interviews, Will commented that he did not plan his coaching behaviours to align with different practice states. Rather, he stated that he ‘liked to improvise,’ a trait learned during his formative experiences in learning how to coach youth basketball: ‘in a camp environment, improvisation is always key.’ This ‘improvisation’ manifested in the top two behaviours in both ‘training’ and ‘playing’ states being the same, where moments of ‘silence’ were punctuated with ‘instruction’ (‘concurrent’, ‘technical’, directed at individuals, teams as well as small groups of players) (see Tables 3 and 6). Will then supported this with ‘general feedback positive’ when the players worked well, or with ‘specific negative’ or ‘corrective feedback’ when they did not. Furthermore, while he asked the players questions, these were often convergent; as he outlined:

I imagine it was very closed…probably…[because I]…wanted the answer immediately, knew what the answer was and knew that they could give me it. I think it was more so questioning for reinforcement, sometimes merely used to reinforce previous instructions.

Thus, while the notion of stepping back in his coaching was desired, actually doing this seemed to be something that Will had difficulty with. Such a tension provides evidence of an ‘epistemological gap’ (Davis & Sumara, 2003; Partington & Cushion, 2013) between these desires and the realities of Will's coaching practice, where he openly admitted that he was still ‘quite directive with the group.’

Will was also aware that he ‘talked a lot’ when coaching. Consequently, even when a ‘playing’ practice state was engaged with where players were in small teams, Will still felt compelled to concurrently ‘instruct’ his players (see Tables 3 and 6). The rationale given related to the considerable mixed ability of the group. Thus, Will noted that he faced the constant challenge of developing some ‘consistency’ in his coaching, ‘just trying to find the right level at which to deliver’ to ‘balance between losing the interest of the highest basketball IQ…but end up up-skilling those with least experience.’ However, the data illustrated a uniform approach to coaching, regardless of their skill level.

Discussion and conclusion

Previous research (e.g. Cushion, Armour, & Jones, 2003) has demonstrated that coaches base their coaching on previous experience. Simultaneously, the season-long nature of this study provided evidence linking the specific detail of coaches' behaviour and session structures to prior socialization and coaching experiences. The resulting practice was directive and prescriptive in nature, and characterized by ‘folk pedagogies’ (Bruner, 1999) guided by coaching tradition and an uncritical inertia (Cushion et al., 2003; Fernandez-Balboa, 1997). The tacit and implicit nature of coaches' beliefs (Cushion et al., 2003) contributed to low self-awareness and an inability to articulate and explain why the coaches structured practice or used the behaviours in the way that they did in different practice states (Partington & Cushion, 2013). Therefore, it should not be surprising that the coaches suffered from an ‘epistemological gap’ (Partington & Cushion, 2013) between what they did, their own understandings of what they did, and their explanations of why they did it in that particular way (Davis & Sumara, 2003). This lack of self-awareness is significant given that a central plank of NGB coach education courses is reflection (Nelson & Cushion, 2006). The participants in this study, despite having progressed through this system, were not reflective about their practice in any meaningful way (Stuart: ‘I have never really looked at myself and tried to look at where I am going wrong…. I've never really reflected’). Nor did the coaches refer to coach education as a driver for their coaching. This study then, further contributes to the evidence base that coach education is ‘low impact’ on coaches' practice and understanding.

Reflection is a key strategy to enable coaches' beliefs and dispositions to be made explicit (Christensen, 2011), whilst also allowing coaches the opportunity to become more aware of their practice (Gilbert & Trudel, 2006). Arguably, reflection is more effective when coaches have a ‘critical friend’ to help promote deeper levels of engagement (Knowles, Gilbourne, Borrie, & Nevill, 2001). However, as coaches often work in isolation (Gilbert & Trudel, 2006), they are not afforded many opportunities to be involved in peer or group reflection. The use of technology within this study provided opportunity for the coaches to specifically explore the relationship between coaching objectives, practice state, coaching behaviours and the implications on athlete learning. All the participants subsequently perceived benefit from the process, as it gave them an opportunity for meaningful reflection and challenged their current beliefs and practices, beginning a process of ‘conceptual change.’

Although Cushion et al. (2012) have questioned the uncritical use of a blanket ‘positive and instructional’ approach to coaching as was evidenced in this study, at the heart of such generalized action was the coaches' socialized belief that the more positive feedback given, the better. For example, it was clear that Stuart's experiences had led him to overuse positive verbalizations to the point where it was ‘meaning nothing’ to his players. As an alternative to this directive, behavioural approach to coaching, Cushion et al. (2012) suggest the use of coaching behaviour more aligned with constructivist learning theory. Such approaches (e.g. guided discovery and Teaching Games for Understanding) are grounded in the use of questioning. The data suggest that the coaches in the current study would need a considerable shift in approach to achieve this, with questioning accounting for no more than 10% of their recorded behaviour. In addition, the questioning used by the coaches here was convergent (i.e. closed), which only cognitively engages learners to a limited degree.

In addition, the practice activities used by the three coaches in this study differed from Ford et al. (2010), who noted a ratio of 65% to 35% of time in ‘training’ and ‘playing’ form practice respectively. The current study more closely aligns with Partington and Cushion (2013), who found a smaller difference between ‘training’ and ‘playing’ forms of practice. This still means, however, that around half of the practice activities recorded were ‘training’ form based, which has been suggested to have an adverse influence on players' skill development, especially where there is little contextual interference (Cushion et al., 2012).

Using CAIS and its more sophisticated analysis of practice types, identified the time spent by coaches in ‘other’ states which included management (i.e. explaining practices), transitions (i.e. movements between practices) and drink breaks (i.e. allowing players to rehydrate). Time in this category ranged from 16–24% of the practice sessions, comprising almost a quarter of total practice. The amount of time spent in ‘other’ states appears illustrative of a disconnection between the coaches' knowledge and planning, and their actual behaviour in relation to athlete learning (Cushion et al., 2012). This does not mean that learning has to be exclusively related to activity as, inevitably, time is required for managing and organizing practice (as well as the need for drink stops). However, in this study, coach cognitions and behaviours did not intersect with the process of planning and instruction. Consequently, for two of the three coaches featured in this study, the ‘other’ practice state appeared only as a largely unconsidered, ad hoc part of their coaching. In this respect, only Stuart used drinks breaks to speak to the players individually and provide individual and/or small group feedback and questioning, thus keeping athletes cognitively engaged.

The current study then demonstrated coaches' low behavioural self-awareness and a limited understanding of the relationship between coaching objectives, practice state and behaviour. As stated, this lack of self-awareness provides further evidence of the limited impact of coach education, given that reflection is a key element of the system through which the participants had progressed. It is subsequently suggested that increasing use of video technology could better initiate and develop the quality and quantity of coaches' (Carson, 2008) reflections required to genuinely impact on, and progressively improve, practice.

Acknowledgements

This research was supported by a seeding grant from the Carnegie Faculty at Leeds Metropolitan University.

Notes

^1. Training form' activities were defined as those with no game related focus (e.g. fitness, technical and skills related activities), ‘playing form’ activities were defined as those with a game related focus (e.g. small-sided games, conditioned games and phases of play).

^2. Each participant coached one team/squad of players during their weekly team training session ahead of their respective university competitions for which they played between 12 and 15 games in the season. In addition, the volleyball team participated in regional league competition (18 matches) while the basketball team competed in the National League (25 games including playoffs). These two teams, therefore, had weekly collegiate competition as well as additional league matches, some weeks playing two matches in addition to training.

^3. Sports Coach UK (the body responsible for overseeing coaching in the UK) outline the duties/roles of different levels of coaches to be: Level 1 coaches assist more qualified coaches, delivering aspects of coaching sessions, normally under direct supervision; Level 2 coaches prepare for, deliver and review coaching sessions; Level 3 coaches plan, implement, analyse and revise annual coaching programmes (Sports Coach UK, 2012).

^4. Activities in the ‘other’ state category were defined as the time when coaches were managing/addressing the team of players, transitioning or moving players from one practice to another, and the time when players took breaks from practice (i.e. water breaks).