Coaching psychology is an established and increasingly popular change methodology in organisations (Grant et al. 2010b). Research on coaching at the individual level has been related to: increased goal striving, hope and wellbeing (Green et al. 2006); increases in goal commitment, attainment and environmental mastery (Spence and Grant 2007); increases in cognitive hardiness, mental health and hope (Green et al. 2007); the reduction of workplace stress and anxiety (Gyllensten and Palmer 2005); improvements in transformational leadership (Grant et al. 2010a; Cerni et al. 2010); and the enhancement of outcome expectancies and self-efficacy (Evers et al. 2006).
While the published research tends to focus heavily on individual level outcomes, the importance of coaching for groups and teams has been asserted previously (Arakawa and Greenberg 2007). Despite these assertions, the impact of coaching at the level of the group, team, organisation or system has largely been ignored. The limited research that does exist in this area has focused mostly on return for investment (Feggetter 2007; McGovern et al. 2001; Palmer 2003). The impact of coaching for leadership development on broader organisational measures such as collaboration, communication flow, relationships and the wellbeing of others in a system, has been left empirically untested. If the wellbeing of organisational members is of any importance, this focus must shift. Focusing on the broader organisational impacts of coaching may provide a deeper understanding of coaching, at both the individual and organisational levels, and give greater clarity on the role of coaching in the process of effective organisational change.
Complex adaptive systems theory
Complex Adaptive Systems theory (CAS) is a promising approach to understanding organisational level dynamics, behaviour and outcomes. CAS theory describes organisations as diverse networks of interacting systems that grow and adapt in response to change in the internal and external environment (Eidelson 1997).
According to CAS theory, systems adapt in novel ways. System components interact, creating feedback and feed forward loops which in turn further affect ongoing behaviour and the trajectory of change (Cavanagh 2006). The recursive and iterative structure of these loops means that change is usually non-linear. This non-linearity renders prediction difficult and limits the utility of linear statistical approaches (Cavanagh and Lane 2012).
From a CAS perspective, the networks of communication and relationships that exist between individuals become the dynamic connections that shape organisational subsystems giving rise to organisational behaviour. By focusing more directly on this interconnectivity (i.e., the pattern, absence and quality of these connections), CAS theory may help us to better understand the potential drivers of emergent organisational change.
The role of the leader
Leaders are active influential agents in organisations. It has been suggested that Leadership is concerned not only with the direct influence of the leader on subordinates, but also with the indirect influence that leadership exerts throughout and around the system at large (Osborn et al. 2002). As change in a system occurs, leaders and other agents adjust to new information. As agents in a system expand (or contract) their behavioural repertoires, the behavioural repertoire of the system as a whole expands or contracts (Kauffman 1993). Coaching is designed to change the individual leader and the way in which that leader interacts with the system. As system members adapt to these changes, the system itself is altered.
The key data of interest, from a CAS perspective, are the nature of the interactions between the system members. These data are essentially relational. They exist between individuals rather than being embedded in a given individual. One way of assessing the influence of coaching on an organisation is to assess changes in the way members of an organisation are connected and interact, i.e. changes in the pattern and quality of their communication. However accessing and analysing relational data presents significant difficulties.
Analysing relational data
Interaction research has been plagued with a heavy commitment to a laboratory-based experimental methodology in order to control for confounding factors. However, in doing so, laboratory studies limit the real life application of their findings (McGrath 1997; Moreland et al. 1994). Consistent with a call for a new and more complex research framework (Frey 1994; Fuhriman and Burlingame 1994; Moreland et al. 1994), McGrath (1997) suggests that group research needs to occur in more realistic settings, thus allowing the group to function as complex, adaptive, dynamic systems.
One research project that sought to take up this challenge was conducted by Losada and Heaphy (2004). They filmed groups of executives during their yearly strategic planning meetings, coding their interactions over time on a range of dimensions (Losada and Heaphy 2004). These dimensions were: (1) the ratio of positivity to negativity, (2) the ratio of advocacy to enquiry and (3) the ratio other-focus to self-focus in the verbal utterances of team members. These dimensions were drawn from the work of previous researchers such as Bales (1950): Bales and Cohen (1979), Gottman (1981); Gottman et al. (1977), Argyris and Schon (1978), Hax and Majluf (1991), and Buber (1970, as cited in Losada and Heaphy 2004).
Losada’s particular contribution lay in the development of mathematical algorithms aimed at assessing the dynamic nature of these interactions across time (Losada 1999). Losada’s algorithms highlighted the importance of connectivity. Connectivity can be thought of as the influence members have on each other, as measured by recurrent patterns of behaviour over time. It was found that connectivity was strongly associated with team performance (Losada and Heaphy 2004).
Losada (1999) realised that the mathematical formula he developed to match the time series data he had witnessed through team observation, was the same set of differential equations used by Lorenz (1963; Strogatz 2001) to understand change in the complex adaptive system of weather. The Lorenz equations have greatly assisted in the examination of how complex systems evolve over time, helping to predict the behaviour of many different types of complex systems based on change in key variables (Thompson and Stewart 1986). Applying these equations to group interaction research has been a promising step forward considering groups as CAS.
Losada later developed the Meta-Learning model (Losada and Heaphy, 2004). This model mathematically related ratios of Advocacy/Inquiry, Other/Self and Positivity/Negativity to Connectivity and the performance of business teams (Losada and Heaphy, 2004). It was found that the most important ratio to consider was that of positivity to negativity and that, on average, a ratio of approximately 5 to 1 of positivity over negativity was indicative of high performing teams.
The research of Losada and his colleagues is particularly important for two reasons. Firstly, alongside Gottman and Levenson (1992) they are among the first to apply non-linear methods to assessing group dynamics. Secondly, this research formalises the mathematical link between basic positivity to negativity ratios and the previous time series data using the Lorenz equations (Losada 1999). Losada’s model and supporting data suggest that simple and easily obtainable measures of positivity and negativity could be used to predict patterns of performance in senior leadership teams.
The individual experience of positivity and negativity
At the dyad level, Gottman has shown that Positivity to Negativity (P/N) ratios could be used to discriminate distressed from non-distressed couples (Gottman et al. 1977), and that low P/N ratios predicted a significantly greater risk for marital dissolution and lower marital satisfaction (Gottman and Levenson 1992). More recently, Fredrickson and Losada (2005), used the mathematical formula from the meta-learning model to predict levels of positivity to negativity in individuals characterised with flourishing mental health. Waugh and Fredrickson (2006) have also found that a similar threshold of P/N can determine those who are able to reach a complex understanding of others from those who cannot.
P/N ratios of between about 3:1 and 8:1 tend to represent flourishing in different forms, at the individual level (Fredrickson and Losada 2005; Waugh and Fredrickson 2006), at the dyadic level (Gottman and Levenson 1992) and at the team or group level (Losada and Heaphy 2004). It would seem a logical next step to consider how P/N ratios may be important at the organisational level. If P/N ratio of interactions in organisations is related to individual experience, we may find that the distributed pattern of experiences in an organisation can relate to measures such as organisational climate, commitment and collaboration. The pattern of positivity and negativity that characterises the communications within an organisation may also have significant influence over the experience of individuals on factors such as wellbeing, engagement and satisfaction (Harter et al. 2003).
The organisational level: social network analysis
Analysing the impact of coaching on the quality of communication at the organisational level, may help us to understand the mechanism by which leaders are able to influence the way systems are experienced, organised and interact. While it is likely that leaders influence the quality of communication and relationships between individuals, to date, analysis of relational data such as interactions, communication and relationship quality, has been left untested in coaching research.
Social Network Analysis (SNA) is a relatively new technique, primarily concerned with understanding networks and the way in which the network members are related (Scott 2000). It has been applied in a wide variety of fields including management, anthropology, political science and psychology (Hatala 2006). However, a lack of empirical research on leadership and social networks has been noted (Brass et al. 2004). SNA takes into account the interconnectivity observed between members. Relational data consists of things such as contacts, ties, information flow, influence and communication between individual agents (i.e., network members, or “components of a system” to use Scott’s terminology). These relations do not belong to the individual agents but are part of the relational system between system agents, or system components (Scott 2000).
SNA is a technique that allows researchers to focus, at a systems level, on the relational data in networks. In doing so, it allows research questions to focus on emergent properties and interconnectivity of a system (Scott 2000). Hence it has potential to yield a more ecologically valid analysis than more common linear approaches used in leadership research.
Balkundi and Kilduff (2006) outline the potential for SNA in investigating leadership and highlight three networks of interest: (a) the direct ties surrounding leaders, (b) the pattern of direct and indirect ties embedding the leader in the organisation or system and (c) the inter-organisational linkages formed between leaders across organisations.
SNA has been used to investigate group performance and leader reputation (Mehra et al. 2006a), leadership distribution in teams (Mehra et al. 2006b) transformational leadership, group interaction and organisational climate (Zohar and Tenne-Gazit 2008), advice and influence networks of transformational leaders (Bono and Anderson 2005), and social capital in relation to intra-firm networks (Tsai and Ghoshal 1998). However, these studies have been cross-sectional. The authors are unaware of any SNA research that has focused on changes in groups, teams or organisations following an intervention designed to improve patterns of interactions and communication. If we are to understand organisations as CAS with non-linear emergent properties, research that measures change over time is required.
Leadership coaching is typically conducted at the individual level. It is concerned with supporting changes in leaders that enhance the effectiveness of the organisation and the relationships that comprise the organisation. Hence, this study seeks to understand the impact of coaching at several levels - individual, relational and organisational. As such, this study represents the first of its kind in the field. Following Balkundi and Kilduff’s (2005) call to action, analyzing not only the network of communication ties that directly surround and embed leaders in a system, but also the quality (positivity / negativity) of these connections represents an important extension of SNA research on leadership and organisational change. By analyzing these data both pre and post leadership coaching, an understanding of the effectiveness of coaching in creating broader organisational change, may emerge.
Hypotheses on individual, relational and systemic level impact of developmental coaching of leaders.
Individual Dimension - development through coaching, has been shown to be beneficial at the individual level for those receiving coaching (Green et al. 2006). Therefore, direct change should be observed in the level of wellbeing, goal attainment and transformational leadership for those who receive coaching. Specifically:
Hypothesis 1: A Significant positive increase will be observed in measures of wellbeing at the conclusion of the intervention period for those receiving coaching.
Hypothesis 2: A Significant positive increase will be observed in goal attainment measures at the conclusion of the intervention period for those receiving coaching.
Hypothesis 3: A Significant positive increase will be observed in360 feedback measures of Transformational leadership at the conclusion of the intervention period for those receiving coaching.
Relation dimension - Theories of transformational leadership suggest that leaders higher in transformational leadership, are better able to build trust, act with integrity, inspire others, encourage innovative thinking, and help others to develop for themselves (Avolio et al. 1995). These five features of transformational leadership are all relational in some way. If transformational leadership qualities are observed to be changing in an individual, then this change in experience of the leaders must be transmitted somehow to those with whom they are connected. This transmission may occur through the quality of interaction that these leaders have with those around them.
For the purpose of this study, there are two measures of change in the quality of interactions that may access this hypothesised change in dynamics. The first is the coached leader’s perception of the quality of interactions between themselves and others. In social networking analysis this perception is thought of as an outward directed relation (Communication Out). It is derived from the ratings a target individual (the coached leader) gives to their relations with others in the system (Freeman 2004). The second measure of change in the quality of system interactions is comprised of ratings made by system members of the quality of interactions they have with a target individual (e.g. the coached leaders) (Freeman 2004). This metric is thought of as an inward directed relation (Communication In).
Communication Out and Communication In may be, but are not necessarily equal. It is possible that undergoing individual leadership development may change how a person perceives their relations with others independent of any measurable or noticeable change in behaviour. If actual behaviour is not changed, it is unlikely the individual receiving development would be experienced by others as different. If no change in the leaders was experience by others, it is unlikely that these others would then undergo any individual level change themselves as a result of the development of the leaders with whom they are connected. In other words, in the absence of behaviour change in the coached leaders, the wellbeing of others in the system and any system level measures are likely to remain unchanged.
However, if coached individuals are experienced as improved on measures of transformational leadership (e.g. trust, inspiration etc.), transformational leadership theory predicts an increase in psychological wellbeing among others in the system (Nielsen et al. 2008). Trust in the relational context has been shown to play a pivotal role in relation to wellbeing, health and life satisfaction (Helliwell and Wang 2011; Helliwell and Putnam 2004). From a network perspective, it would follow that those most connected to leaders with improved levels of transformational leadership would be most likely to experience change in wellbeing.
In order to assess the potential ripple effect that leadership coaching may have in an organisation, change in the quality of interactions as perceived by those that received coaching (Communication Out) and any change to the experience of the quality of interactions by others, of those that received coaching (Communication In) need to be assessed. If the Coaching Ripple Effect is observed, a positive change will be seen in measures of Communication Out and Communication In, for those that received coaching compared to those that did not. Specifically:
Hypothesis 4: There will be a positive increase in Coached participants’ mean perception of the quality of their communication (Communication Out), compared to non-coached participants following the coaching intervention.
Hypothesis 5: There will be a positive increase in participants’ perception of the mean quality of their communication with coached individuals (Communication In), compared to non-coached individuals following the coaching intervention.
If hypothesis 4 and hypothesis 5 are supported, an improvement in the wellbeing of those most closely connected to the coached individuals may be observed, as these closely connected individuals would be most exposed to positive relation change with the leaders across the system.
Individuals also have multiple connections across complex systems, only some of which are to those receiving coaching. If one’s experience of the system is related to their place within the structure and architecture of the network as a whole, then it is important to take into account all their connections and not just those with targeted individuals such as the coached leaders.
For example, consider the case of Peter. Peter is directly connected to Meg, a recipient of coaching, who following coaching, has positively changed how she interacts with others. These others, in turn, have positively changed the quality of their interactions with others. If we assume that Peter would observe at least some of these changes, Peter might begin to experience the work place differently, even though he himself had not been coached. However, we also need to take into consideration the number of people to whom Peter is also connected who have shown no change, or even negative change in their interactions. The experience of these connections may serve to inhibit any positive shift in Peter’s experience of the workplace. Hence, it is important to consider the full range of Peter’s experience of interaction in the organisation. By measuring individual change in relation to organisational level interconnectivity, this approach seeks to account for the interconnected context in which an individual is embedded.
In networked organisations, an individual may be connected to a number of coaching recipients. If the coaching is effective, such an individual is likely to experience a proportionally larger degree of change in their experience of the system than someone connected to only one coaching recipient, or only indirectly connected to a coaching recipient. It is plausible that the greater weight of positivity may lead to a comparatively greater shift in the individual’s experience of the organisation, and comparably greater increases in wellbeing and positive organisational indices such as collaboration, engagement and work place satisfaction.
The multiplicity of connections between people, and the large degree of variation in network positioning, present challenges to assessing interconnectivity in large, complex, interconnected systems. To overcome this challenge SNA provides a number of metrics of interconnectivity. One group of such metrics involves the notion of Centrality. Centrality measures different ways in which a person may be embedded in a network. For our purposes two types of centrality are important.
-
1.
Degree Centrality: This measure of centrality provides a basic count of the number and strength of connections an individual has in proportion to the number they could have across an entire network,
-
2.
Closeness Centrality: this is a measure of the degree to which an individual lies a short distance from most other individuals (Scott 2000).
Unfortunately, no analytical techniques currently exist, that enable analysis of changes in centrality in specific sub groups comparative to an entire network. Therefore, the ability to assess network level changes that may be occurring locally to only those that have received coaching is limited. One way to overcome this limitation is to create a sub network or network neighbourhood (Hanneman and Riddle 2005) which only includes those connected to at least one coached individual and the connections that these individuals have with each other. Comparisons can then be made between the centrality measures observed in both the whole network and the coached neighbourhood network, and their relation to change in any of the individual level variables. This means that a stronger relationship between measures of individual level wellbeing and measures of centrality should be observed in the coached neighbourhood network, compared to the whole network for the intervention period. Such an observation would provide supporting evidence for a coaching ripple effect.
Hypothesis 6: The relationship between measures of change in wellbeing and centrality in the quality of interactions within the coached neighbourhood network will be stronger than those observed in the primary or whole network.
Organisational level impacts - The coaching ripple effect would also suggest that if positive change in the quality of communication occurs in key sub networks across the system to a great enough degree, other system level measures may also be observed to change. Specifically, one would expect a positive change in the density of the positive interactions in the system as a whole. For valued and directed networks, network density is defined as the sum of the value of all ties present, divided by the number of possible ties. Consequently, in this study, density is the ratio of the number and strength of all present ties that each individual has in a network, to the theoretical maximum number and strength of all possible ties (Hanneman and Riddle 2005). Density is a measure of the degree of total interconnectivity of the system as a whole. If the positivity of communication across the network is improved to a great enough degree, the density of the quality of the interaction network should improve over the intervention period.
Hypothesis 7: A positive increase will be observed in the density of the quality of interaction network post intervention period.
Exploring the above questions contributes to the field of coaching and leadership research in a number of ways:
-
1.
By adding to the empirical coaching literature on the effectiveness of coaching at the individual level,
-
2.
By being one of the first studies to consider the impact of coaching on wellbeing through a complex adaptive systems theory of organisations,
-
3.
By directly applying social network analysis as a methodology for assessing coaching intervention and organisational change, and
-
4.
By considering the potential impact of leadership coaching and change in the leader, beyond the leader themselves.