People’s ability to collaborate with each other in organisations is becoming a business imperative. Effective collaboration can enhance product innovation and service delivery, and increase efficacy (Barber & Goold, 2014; Miller & Katz, 2014; Patel, Pettitt & Wilson, 2012; Sanker, 2012; Simatupang & Sridharan, 2005). Leveraging these benefits could be of considerable value to South African businesses. A valid measure of collaboration could help South African business leaders optimise opportunities and address obstacles to effective collaboration.

Globalisation, technology advancements and the fast rate of change are all catalysts for increased complexity in the business environment (Barber & Goold, 2014; Kaats & Opheij, 2014; Miller & Katz, 2014; Sanchez, 2012). South African organisations face the same complexities as the rest of the world. The country also faces unique challenges, such as electricity shortages, labour unrest, poor quality of education and increasing debt (International Monetary Fund, 2013; World Economic Forum, 2015). Higher education enrolment rates are not foreseen to produce the level of skills required for future competitiveness (World Economic Forum, 2015). These issues place additional pressure on businesses to operate efficiently and innovatively.

Coping with this complexity requires staying abreast of new developments, solving difficult problems and at the same time seizing opportunities. This is not easy for individual entities to achieve alone (Barber & Goold, 2014; Kaats & Opheij, 2014; Woodland & Hutton, 2012). To deal with this complexity, many organisations are moving from hierarchies to network systems. This change is necessary but can involve complicated interactions, meaning that communication, learning and innovation must be fast and effective (Kaats & Opheij, 2014; Miller & Katz, 2014). A powerful strategy to attain improved results under these conditions is through collaborative partnerships (Barber & Goold, 2014; Kaats & Opheij, 2014; Miller & Katz, 2014; Sanchez, 2012; Thomson, Perry & Miller, 2007; Woodland & Hutton, 2012). This necessitates establishing an environment that fosters a collaborative culture. Further to improving business results, a collaborative culture can also improve employee morale and retention (Murphy, Arenas & Batista, 2014). Exposure to collaboration allows individuals to perform roles other than their own, enhancing self-development (Sanchez, 2012). People also experience expanded thinking, broadened and diverse networks and inspiration from co-collaborators (Bushe, 2006). Collaboration can also provide strong social connections, which have been found to promote positive mental health (Bond, Lusher, Williams & Butler, 2014).

Conversely, efforts to collaborate can result in failure. There are several potential obstacles to achieving a collaborative environment. Poorly designed partnership agreements usually result in ambiguous roles and the likelihood of poor communication (Patel et al., 2012). Structures, processes and technologies that do not support collaboration can create difficulties (Hudson, Hardy, Henwood & Wistow, 1999; Kaats & Opheij, 2014) as can weak knowledge management (Patel et al., 2012). Collaboration requires significant energy in managing varying priorities, conflicts and territory issues (Sanchez, 2012). Given these challenges, before embarking on collaboration, leaders should assess whether the reward outweighs the effort and establish that all stakeholders share this perception (Marek, Brock & Savla, 2015; Patel et al., 2012).

Evaluation and feedback on how the organisation is progressing towards a collaborative culture is considered to be a valuable endeavour (Majchrzak, Jarvenpaa & Bagherzadeh, 2015; Marek et al., 2015; Murphy et al., 2014; Wong, 2006). Several measurement instruments have been developed over the years, many of which apply to specific industries or disciplines. This study reviewed various collaboration instruments for their suitability for validation in a South African context and application to a general business environment. The collaboration instrument selected for this study was the Thomson, Perry and Miller (2007) Collaboration Instrument. This instrument was based upon a five-factor model of collaboration as illustrated in Figure 1.

Collaboration has been the focus of much scholarly research. The volume of literature suggests a keen interest in exploring this construct. Despite this interest, it is apparent that a common and accepted definition does not exist. In some cases, definitions are specific to certain environments such as education or nursing. This lack of clarity around collaboration is considered an obstacle to supporting and improving collaboration (Hudson et al., 1999; Patel et al., 2012; Thomson et al., 2007). However, there are generic definitions that support similar themes. The following operational definition, highlighting these themes, is suggested for the purpose of this study: collaboration is a combined undertaking between mutually dependent parties that have shared responsibilities on an agreed process with an envisaged outcome to deliver a result considered superior to what the parties could achieve individually (Kaats & Opheij, 2014; Mattessich, Murray-Close & Monsey, 2001; Sanker, 2012; Schwarz, 2006; Thomson et al., 2007). The literature discussion that follows is framed within:

stages of collaboration

Because there is no evidence of an existing collaboration measure in South Africa, use of an international instrument was considered appropriate. With a limited number of tests developed specifically for South Africa, it is not unusual to make use of international instruments (Foxcroft & Roodt, 2005; Gajda, 2004; Orchard et al., 2012). It is, however, necessary to be cautious when using a measurement instrument originally validated in a different setting (Gregory, 2005). Certain instruments developed in other countries have been empirically tested and proved to have good portability to a South African context (Campbell & Young, 2011; Collings, Valjee & Penning, 2013; De Bruin, Swartz, Tomlinson, Cooper & Molteno, 2004; Vermeulen, Schaap, Mitchell & Kristovics, 2009). However, specific studies support the assertion that not all assessment instruments are suitable for testing in a South African environment without some adjustment (De Klerk, Boshoff & Van Wyk, 2009; Meiring, Van de Vijver & Rothmann, 2006; Van Eeden & Wissing, 2008). For these reasons, it is required that instruments are appropriately tested for each setting (Donald, Thatcher & Milner, 2014; Foxcroft & Roodt, 2005; Gregory, 2005). This study therefore aimed to validate the Thomson, Perry and Miller Collaboration Instrument (Thomson et al., 2007) in the South African context, with the objective of using it in South African organisations. A secondary objective was to test convergent validity with the norms factor of the Thomson, Perry and Miller Collaboration Instrument (Thomson et al., 2007) with the Trust in Teams Scale (Adams, Waldherr & Sartori, 2008).

This study could further the body of knowledge on collaboration through testing a conceptual model’s reliability and construct validity in the South African context. Validation of the existing instrument could contribute to more reliable measures in the South African context. This may enable organisations to understand the extent to which the factors of collaboration are a strength or weakness in their organisation. Equipped with this knowledge, leaders could prioritise initiatives to create the conditions that enable improved collaboration. Human resources functions are being called upon to enhance their contribution and become strategic partners to business (Beatty, Huselid & Schneier, 2003). An area where human resources practitioners can increase their value-add to companies is through the provision of human capital measures that support implementation of the business strategy (Beatty et al., 2003; Chrysler-Fox & Roodt, 2014). Collaboration is an example of one such measure. This is particularly true for South Africa where human resource management has predominantly focused on industrial relations as opposed to increasing organisational effectiveness (Horwitz, 2008).

Descriptive statistics for the items in the Revised Thomson, Perry and Miller (2007) Collaboration Instrument show that all items received the minimum and maximum responses on the Likert scale. The mean ranged from 2.63 to 5.92 with standard deviations ranging from 1.220 to 1.648. All items in the Trust in Teams Scale (Adams et al., 2008) also received the minimum and maximum responses on the Likert scale. One item was removed as a duplication. The mean ranged from 4.90 to 5.89 with standard deviations ranging from 1.111 to 1.637. The mode and median indicated a high predominance of the 6 rating.

Exploratory Factor Analysis of the Revised Thomson, Perry and Miller (2007) Collaboration Instrument indicated Kaiser-Meyer-Olkin (KMO) measure at an acceptable level of 0.95, which is above the value (0.6) suggested by Tabachnick and Fidell (2007). The Bartlett’s test of sphericity was significant at 7584.2 (df = 465; p = 0.00). Principal Components Analysis with Varimax rotation indicated five factors with Eigenvalues above one. The five factors explained 65.3% of the total variance.

The rotated factor matrix for the Revised Thomson, Perry and Miller (2007) Collaboration Instrument on a five-factor model resulted in only two items loading on factor five. These two items were not theoretically relevant and it was decided to delete the two items for further analyses. Governance and administration loaded on one factor, as well as two autonomy items, which was considered theoretically relevant. Rotation converged in 10 iterations. This model was further analysed by means of Principal Components Analysis with Varimax Rotation on four components. Eigenvalues above one were indicated as: Component 1 = 14.529, Component 2 = 2.446, Component 3 = 1.795 and Component 4 = 1.516. The total variance explained by the four components is 69.95%, with Component 1 explaining 25.25%, Component 2, 21.38%, Component 3, 15.13% and Component 4, 8.20% of the variance in the data space. The rotated factor matrix for the Revised Thomson, Perry and Miller (2007) Collaboration Instrument on a four-factor model confirmed governance and administration as one factor. Rotation converged in eight iterations. Principal Components Analysis with Varimax Rotation of the Revised Thomson, Perry and Miller (2007) Collaboration Instrument on a four-factor model is reported in Table 2.

Exploratory Factor Analysis of the Trust in Teams Scale (Adams et al., 2008) indicated the KMO measure (KMO) above 0.6, at an acceptable level of 0.96. The Bartlett’s test of sphericity was at a significant level Chi-square = 9064.85 (df = 171; p = 0.000). Two factors with Eigenvalues were above one, explaining 78.3% of the total variance. The two components of the Trust in Teams Scale (Adams et al., 2008) were named Component 1: integrity and Component 2: competence.

Reliability of both the Revised Thomson, Perry and Miller (2007) Collaboration Instrument and the Trust in Teams Scale (Adams et al., 2008) was calculated using Cronbach’s alpha coefficient. Cronbach’s alpha values for the Revised Thomson, Perry and Miller (2007) Collaboration Instrument are factor 1: governance (α = 0.95), factor 2: mutuality (α = 0.92), factor 3: norms (α = 0.9) and factor 4: autonomy (α = 0.85) and for the Trust in Teams Scale (Adams et al., 2008) are factor 1: integrity (α = 0.97) and factor 2: competence (α = 0.95).

The CFA was done by means of EQS. The CFA of both the Revised Thomson, Perry and Miller (2007) Collaboration Instrument and the Trust in Teams Scale (Adams et al., 2008) are reported in Table 3. Items that cross loaded and were initially omitted though did not increase fit indices. This could be due to similarity of questions. The final factors thus included the cross loading items. None of the Mardia’s coefficients were less than three, which is an indication of a significant non-normality. Chi-square indices of the different factors showed a weak to good fit with the data. Weak fits are indicated in governance, integrity and competence. Collaboration total has a good fit of 4.3. Trust in Teams total has a weak fit of 10.01. Comparative fit indices are between 0.86 and 0.99, which is between a reasonable and good fit (Hair, Black, Babin, Anderson & Tatham, 2006). No CFA was performed on autonomy as there were only three items.

The results of this study provide acceptable indices of the Revised Thomson, Perry and Miller (2007) Collaboration Instrument when applied to a South African sample. It affirms the premise by Foxcroft and Roodt (2005) that instruments used in alternative countries must be validated for a South African context. It addresses the recommendations of Thomson et al. (2007) to refine their scale with the addition of items and subsequent testing of construct validity in other contexts.

Thirteen of the fourteen new items were retained. Only two items in total were eliminated, both which loaded on factor five in the initial Principal Components Analysis. These items were eliminated as they could not be theoretically explained. The moderate convergent validity with the prediction of norms on the trust factors of integrity and competence could be due to the different reference to team collaboration of the two instruments. The Revised Thomson, Perry and Miller (2007) Collaboration Instrument refers to participants’ experience of individuals in other teams, for example, ‘The people who represent other teams in collaborative endeavours are competent’. The Trust in in Teams Scale (Adams et al., 2008) makes reference to participants’ experience of own teammates, for example, ‘My teammates are capable at their jobs’.