This study attempted to empirically assess a measurement instrument for entrepreneurial actions as regards sustainable CE within established organisations. The central research question endeavoured to determine whether the hypothesised measurement instrument for these actions had acceptable psychometric attributes. Specifically, the study was concerned with the reliability, validity and unidimensional capacity of the measurement instrument. In this regard, reliability refers to the instrument’s ability to measure the construct of interest consistently while validity refers to how successfully the instrument accurately measures the construct of interest (Hair et al. 2010). Unidimensionality of a measurement scale refers to the scale’s items being strongly associated with each other and representing a single construct (Hair et al. 2010:125). Furthermore, the study also aimed at establishing whether the hypothesised measurement instrument for entrepreneurial actions for sustainable CE had acceptable intercultural psychometric portability. The hypothesised measurement instrument, with the exception of item EA4, was developed and applied in diverse demographic and cultural contexts (Kuratko et al. 2005a). Many psychometric instruments fail the test of intercultural psychometric portability (Dolnicar & Grün 2007; Van Wyk & Adonisi 2011). Accordingly, the research objectives were (1) to assess measurement reliability of the hypothesised measurement instrument for entrepreneurial actions relating to sustainable CE; (2) to assess construct validity of the said measurement instrument; (3) to determine whether this instrument was unidimensional; (4) to determine the intercultural psychometric portability of the hypothesised instrument and (5) to make a contribution to the CE knowledge.

Entrepreneurship within established organisations: Kuratko (2013:5) posits that entrepreneurship is a ‘dynamic process of vision, change, and creation’; it is the process by which new organisations come into existence. However, a wide range of activities such as ‘creation, founding, adapting, and managing a venture’ have been used to define the phenomenon (Kusumsiri & Jayawardane 2013:26). CE is the term scholars generally use to refer to entrepreneurship practiced within established organisations and the scope of the phenomenon has ultimately expanded significantly, becoming more refined (Corbett et al. 2013:812). However, different opinions still exist among CE researchers regarding the attributes that must exist in order to label an organisation entrepreneurial (Covin & Miles 1999:49), resulting in different definitions and terms used for the phenomenon, while the CE construct itself takes many forms (Menzel, Aaltio & Ulijn 2007:733; Sharma & Chrisman 1999:13).

Covin and Miles (1999:50) define CE as the process of innovation coupled with the presence of the objective to rejuvenate or purposely redefine organisations, markets or industries in order to create or sustain competitive superiority. CE is considered to adopt five distinct forms (Covin & Miles 1999:50–54; Morris, Kuratko & Covin 2011:99–101): (1) sustained regeneration: the CE phenomenon where the organisation regularly and continuously introduces new products and services or enters new markets; (2) organisational rejuvenation: the CE phenomenon in terms of which the organisation seeks to sustain or improve its competitive standing by altering its internal processes, structures and/or capabilities; (3) strategic renewal: the CE phenomenon where the organisation seeks to redefine its relationship with its markets or industry competitors by fundamentally altering how it competes; (4) domain redefinition: the CE phenomenon in which the organisation proactively creates a new product market arena that others have not recognised or actively sought to exploit and (5) business model reconstruction: a form of CE where an organisation applies entrepreneurial thinking to the design or redesigning of its core business model(s) in order to improve its operational efficiencies or otherwise differentiate itself from industry competitors in ways valued by the market. Therefore, entrepreneurship is more than just individual entrepreneurial activities performed by employees or managers. The phenomenon becomes integrated into the entire organisational fabric and ‘captures the essence of what an organisation is about and how it operates’ (Morris et al. 2011:52). The literature indicates that CE exerts a direct and positive influence on organisational performance, as it is interwoven with the organisation’s mission and vision, strategies, objectives, structures and everyday operations as well as the overall organisational culture (Covin & Slevin 1991; Morris et al. 2011:52). CE is essentially the effort of promoting innovation within an existing organisation through products, processes, strategies, domain or business models in order to discover, assess and ultimately exploit attractive economic opportunities based on a long-term perspective.

In the field of entrepreneurship, the term sustainability takes on many different conceptualisations and dimensions. Some researchers’ perspectives are those of sustainable entrepreneurship in relation to environmental stewardship (e.g. Miles, Munilla & Darroch 2009; Schaltegger & Wagner 2011; Shepherd & Patzelt 2011). However, other researchers use the term ‘sustainability’ in relation to corporate entrepreneurial behaviour, implying ongoing improvements or enduring entrepreneurial capabilities within established organisations, resulting in a sustainable competitive advantage and sustained performance, and employ terms such as ‘sustainable’ CE (Kelly 2011:74), ‘sustaining CE’ (Kuratko, Hornsby & Goldsby 2004:79), ‘sustainable competitive advantage’ (Urban & Nikolov 2013:384), ‘strategic entrepreneurship’ (Hitt et al. 2011:57) and ‘continued entrepreneurship’ (Davidsson 1991:405) to define the phenomenon. Ultimately, CE should lead to sustainability of a venture resulting from the organisation’s ability to ensure continuous improvement and radical innovation (Morris et al. 2011:375, 403). This study uses the term ‘sustainability’ in this context of ongoing improvements or enduring entrepreneurial capabilities within established organisations, resulting in a sustainable competitive advantage and sustained performance. By adopting the CE typology of Covin and Miles (1999) and Morris et al. (2011), we define sustainable CE as the effort of promoting sustained innovation relating to products, processes, strategies, domain or business models within an existing organisation in order to discover, assess and ultimately exploit attractive entrepreneurial opportunities, to bring about ongoing improvement in organisational performance.

The Theory of Entrepreneurial Actions for Sustainable corporate entrepreneurship: Entrepreneurial actions are activities upon which organisations embark to exploit entrepreneurial opportunities unnoticed or unexploited by rivals; these actions constitute a fundamental organisational behaviour in which organisations move into new markets, seize new customers and/or combine existing resources in new ways (Kuratko et al. 2005a:276; Smith & Di Gregorio 2002). In this regard, entrepreneurial actions are critical in pursuing a CE strategy as they are a conduit for exploiting identified competitive advantages. It is evident from the literature that the construct entrepreneurial actions has been operationalised variously and used in different contexts by several scholars (e.g. Ireland et al. 2001; Kuratko, Ireland & Hornsby 2001; Kuratko et al. 2005a; Simon & Shrader 2012; Wensley et al. 2011), while others have used it without offering any operational definition (e.g. Hornsby et al. 2013). For instance, as noted, Kuratko et al. (2005a:276) define entrepreneurial actions as ‘any newly fashioned set of actions through which companies seek to exploit entrepreneurial opportunities that rivals have not noticed or exploited’. This definition is similar to the one offered by Ireland et al. (2001:50), hinted at above, who define entrepreneurial actions as ‘newly fashioned behaviours through which companies exploit opportunities others have not identified or exploited’ and are ‘oriented to novelty’. According to Kuratko et al. (2001:60), the defining characteristic of entrepreneurial actions is novelty in relation to new resources, markets or customers or a new combination of resources, markets and customers.

Furthermore, Kuratko et al. (2001:61), posit that ‘product, process, and market innovations’ are often a result of ‘newly fashioned entrepreneurial actions’, through which organisations exploit opportunities on a first-mover basis. As mentioned, Smith and Di Gregorio (2002) postulate that these entrepreneurial actions make up a fundamental organisational behaviour by which organisations move into new markets, seize new customers and/or undertake new combinations of existing resources. According to Wensley et al. (2011:133), entrepreneurial actions involve ‘arranging or organising the human and capital assets under someone’s control for the purposes of creating a viable profit-making entity’. However, these authors (Wensley et al. 2011:133) clearly conceptualise entrepreneurial actions as various actions undertaken to achieve specific entrepreneurial objectives, such as successfully achieving the creation of customer capital and operationalise the concept in terms of: (1) knowledge exploration practices that allow organisational members to call upon aspects of practice latent in the periphery as they are needed, instead of deciding ahead of time what an individual needs to know and making this explicitly available to the exclusion of everything else; (2) the presence of open-minded practices within an organisation, which leads to the questioning of established organisational approaches to problem solving and organisational beliefs, even when these have led to successful outcomes in the past and (3) knowledge exploitation practices. Thus, there are different types of entrepreneurial actions.

This study distinguishes the two concepts of entrepreneurial actions and CE as separate but related constructs. On the one hand, entrepreneurial actions pertaining to established organisations are the specific entrepreneurial behaviours of managers, as well as of the other individual employees within an organisation and act as a conduit through which CE is practised (Hitt et al. 2001). CE, on the other hand, is the ‘ostensibly larger topical domain’ (Covin & Lumpkin 2011:855) embracing all aspects of entrepreneurship inside established organisations such as entrepreneurial orientation/posture, strategic entrepreneurship, corporate venturing and intrapreneurship. In other words, CE is the bigger and overarching phenomenon, an overall construct capturing all entrepreneurial activities within established business organisations (De Jong et al. 2011:4). Indeed, in this respect, CE includes entrepreneurial behaviour and orientation in established organisations (Urbano & Turró 2013).

However, entrepreneurial actions or behaviours can be considered to be the ‘conduit’ of CE (Hitt et al. 2001). As Kuratko et al. (2001:61) put it, such actions are in fact ‘novel behaviours’ that the organisation

Entrepreneurial actions are actually specific to identified entrepreneurial activity in a given decision-making context for the organisation and do reflect a unique decision environment, ranging from new venture creation to product introduction in a dynamic environment (Simon & Shrader 2012:292–293). For example, Simon and Shrader (2012), in their empirical study on entrepreneurial actions and optimistic overconfidence, examined five different types of entrepreneurial actions, including venture creation, introducing products that entail more utilisation of the organisation’s resources and introducing products in hostile or dynamic environments.

By adapting the definitions offered by Kuratko et al. (2005a:276) and Ireland et al. (2001:50), this study defines managers’ entrepreneurial actions for sustainable CE as a set of newly fashioned activities within an established organisation aimed at ensuring ongoing exploitation of entrepreneurial opportunities unexploited by rivals. Satisfaction with organisational outcomes in relation to the implemented entrepreneurial activities is an important aspect of these entrepreneurial actions as it creates the basis for the decision whether to sustain, enhance or revise such activities in order to achieve a sustainable competitive advantage and ongoing improvements in performance. Managers’ entrepreneurial actions have been identified as playing an important role in instigating entrepreneurship within established organisations and are considered as ‘critical pathways to competitive advantage and improved performance’ (Kuratko et al. 2001:60). In this respect, their actions are the channels by which sustainable CE and sustained organisational performance can be achieved. The presence of an appropriate CE climate or of organisational antecedents is critical in instigating these entrepreneurial actions. However, managers at different levels have differing views concerning the feasibility and/or desirability of the organisational antecedents (top management support, rewards, time availability, work discretion and organisational boundaries) for promoting entrepreneurial actions (Hornsby et al. 2009:241).

According to Kuratko et al. (2005b:708), managers undertake entrepreneurial actions only once they are convinced that the organisational antecedents to those actions exist and when the managers are in fact aware of their existence. Once individuals recognise as well as interpret these antecedents as indicators of the existence of an internal environment supportive of such actions, they are led to assess their capacities for entrepreneurial undertaking on the basis of what they perceive to be a set of organisational resources, opportunities and obstacles to engaging in entrepreneurial actions (Chen, Greene & Crick 1998; Kuratko et al. 2005a:280). This is critical for sustaining CE, as it instigates continuity or discontinuity of current entrepreneurial actions and was the reason the researchers added the fourth item to measure entrepreneurial actions for sustainable CE.

Therefore, entrepreneurial actions do not occur in a vacuum as they actually take place within the context of the organisation’s complete range of actions (Dess, Lumpkin & Covin 1997). In this regard, an organisation’s entrepreneurial actions could have a mediating effect on the ability of a corporate climate to stimulate sustainable CE. Therefore, an appropriate measurement instrument for entrepreneurial actions is fundamental to the study of corporate entrepreneurial climate and sustainable CE. Because an organisation’s entrepreneurial actions constitute its fundamental behaviour by which it penetrates new markets, seizes new customers and/or undertakes new combinations of existing resources (Smith & Di Gregorio 2002), it is expected that organisations exhibiting more entrepreneurial actions will tend to be on the path to sustainable CE.

This study attempted to empirically assess an instrument for measuring entrepreneurial actions for sustainable CE within established organisations, with a view to constructing a more psychometrically sound measure of the phenomenon, based on the instrument to be tested. Accordingly, the study was aimed at assessing reliability, validity and unidimensionality of the postulated measurement scale, as well as establishing the scale’s intercultural psychometric portability.

An assessment of measurement reliability: The consistency of a particular measurement instrument’s scores is referred to as the measurement reliability (Hair et al. 2010; Said, Badru & Shahid 2011; Streiner 2003). Scholars usually use Cronbach’s alpha to assess this, with a commonly accepted reliability threshold of coefficient alpha (α) ≥ 0.70. In the context of structural equation modelling (SEM), squared multiple correlations (SMCs), factor loadings and error variances are used to measure reliability. According to Hooper, Coughlan and Mullen, (2008), items that have SMCs less than 0.20 should be considered for deletion because such levels of SMCs are an indication that the item is measuring something else than it was intended to do. Standardised loadings are used to measure individual indicator reliability with an ideal threshold of at least 0.70, giving a reliability of at least 0.50; that is, at least 50% of the measure’s explained variance should be a function of its factor (Bagozzi & Yi 2012).

An assessment of construct validity: The extent to which a set of measured items reflects the theoretical latent construct the items are meant to measure is known as the construct validity (Hair et al. 2010:708; Westen & Rosenthal 2003:608–609). In this respect, it essentially involves the accuracy of the measurement itself, and this is one of the main objectives of Confirmatory Factor Analysis (CFA): to assess the construct validity of a postulated measurement theory (Hair et al. 2010:709). To do so, the study examined discriminant, convergent and face validity of the postulated measurement scale (Hair et al. 2010:709–710). In addition to the above, the underlying theory is essential when making a decision about construct validity. Theoretical support, rather than just empirical justification, should be the basis for model re-specification (Hair et al. 2010:647).

Convergent validity is the extent to which items that are indicators of a specific construct converge or share a high proportion of variance in common (Hair et al. 2010:710). When indicators to latent variables correlate with each other to an acceptable extent, then convergent validity is achieved. Based on the measurement model’s goodness of fit (GOF), the conventional rule of thumb suggests that convergent validity is indicated when factor loadings are ≥ 0.70 for all indicators (Garson 2012). According to (Van Dyne & LePine 1998:112), evidence of convergent validity can also be established ‘when each item has a statistically significant loading on its specified factor’. For the postulated measurement model, convergent validity was also assessed using average variance extracted (AVE) which, according to Hair et al. (2010:688), is a ‘summary measure of convergence among a set of items representing a latent construct’ and is the ‘average percentage of variation explained (variance extracted) among the items of a construct’. In this respect, AVE ≥ 50% demonstrates that acceptable convergent validity has been achieved (Hair et al. 2010:709; Shook et al. 2004:400). For a construct with n items, its AVE can be computed as the total of all squared standardised factor loadings divided by the number of items (Hair et al. 2010:709).

Discriminant validity is the extent to which a construct is truly distinct from other constructs – the higher the discriminant validity, the more evident it is that the construct is unique from other constructs and vice versa, while the presence of cross-loadings implies a discriminant validity problem (Hair et al. 2010:710). In other words, when measures of different concepts are distinct, then discriminant validity is demonstrated, which is the case when there are low correlations among the concepts (Bagozzi et al. 1991). Unfortunately, different sources report different criteria for what could be considered as low correlation. However, Brown (2006) indicates that correlations ≥ 0.85 between constructs count for poor discriminant validity. Basically in SEM, if the measurement model is found acceptable, such a model presents its own evidence of discriminant and convergent validity (Anderson & Gerbing 1988; Shook et al. 2004). Acceptability of a measurement model would mean that it has fit indices ≥ 0.90 and significant factor loadings ≥ 0.70, with average variance extracted ≥ 50%, showing acceptable convergent validity (Shook et al. 2004:400).

Face validity is the degree to which the content of the items is consistent with the way the construct is defined and is based solely on the researcher’s judgement (Hair et al. 2010:710). This was vital for the study as the measurement model involved inclusion of a new item for the scale previously used in different research undertakings. Face (or content) validity was considered during the stage of piloting the hypothesised measurement instrument; the researchers were satisfied that all the four items were consistent with the definition of entrepreneurial actions for sustainable CE.

An assessment of unidimensionality: The literature reveals that measurement models that are psychometric must satisfy the attribute of unidimensionality, that is, ‘valid and legitimate summing of rating scale items into an interpretable total score rest on the requirement that the items represent one common underlying (latent) variable’ (Hagell 2014:457). In this respect, an underlying assumption and necessary requirement for creating a summated scale is that the testing of unidimensionality requires such a scale to consist of measurement ‘items loading highly on a single factor’ in that scale (Hair et al. 2010:125). According to Nazim & Ahmad (2013:4), a measurement model is unidimensional if there are no redundant items, that is, items with factor loading < 0.5 or items highly correlated to each other. Accordingly, if a measurement scale lacks unidimensionality, this ‘cannot be compensated for by study design or analytical statistics, and hampers the understanding and usefulness of the outcomes’ (Hagell 2014:457).

Therefore, it was deemed essential in this study to assess the hypothesised measurement instrument on the basis of the following hypotheses:

Hypothesis1: The proposed measurement instrument for entrepreneurial actions for sustainable CE has acceptable internal consistency.

Hypothesis 2: The proposed measurement instrument for entrepreneurial actions for sustainable CE has acceptable construct validity.

Hypothesis 3: The proposed measurement instrument for entrepreneurial actions for sustainable CE is a unidimensional scale.

Hypothesis 4: The proposed measurement instrument for entrepreneurial actions for sustainable CE has acceptable intercultural psychometric portability.

Three of the items or indicators for the hypothesised measurement instrument were adapted from the instrument developed by Kuratko et al. (2005a) and used as three single-item latent variables to measure entrepreneurial actions in a study of managers’ corporate entrepreneurial actions and job satisfaction. The three items are (Kuratko et al. (2005a): (1) the number of new ideas suggested, (2) the number of new ideas implemented and (3) the number of improvements implemented without official organisational approval. Unfortunately, the study by Kuratko et al. (2005a) did not report the reliability and validity of the measurement scale for entrepreneurial actions. Since then, there has been little scholarly effort exerted to refine the measurement instrument. Due to this void in literature, the researchers could not find appropriate previous research dealing with assessment of a measurement scale for entrepreneurial actions for sustainable CE. Therefore, this demonstrates a need for developing and validating a measurement instrument for entrepreneurial actions in an organisation so that it could be depended upon as an appropriate reflection of CE levels for the purposes of stimulating enhanced corporate competitiveness and performance.

Therefore, developing a psychometrically sound measurement instrument for managers’ entrepreneurial actions pertaining to sustainable CE would be a significant contribution to the CE domain. In adapting the items developed by Kuratko et al. (2005a), the researchers also changed the wording by way of framing the items into statements requiring responses as indicated in Table 1.

Furthermore, this study aimed at assessing a measurement instrument for entrepreneurial actions for sustainable CE. However, the measurement instrument developed by Kuratko et al. (2005a) did not incorporate the aspect of continuity of entrepreneurial actions for the purposes of sustaining entrepreneurship within established organisations. In their model of sustained CE, Kuratko et al. (2004:79) demonstrated that sustainability is dependent on managers continuing to undertake innovative activities as well as their positive perceptions of these activities, which in turn results in further allocation of the needed organisational support and resources. In this respect, managers’ satisfaction with performance based on their entrepreneurial actions serves as a feedback mechanism for either continuing or discontinuing current strategy (Kuratko et al. 2004; Morris et al. 2011) For instance, empirical research findings by Simon and Shrader (2012:291) indicated that both high and low levels of satisfaction with organisational performance were associated with the type of entrepreneurial actions relating to product introductions. Furthermore, Kuratko et al. (2004) point out that managers’ satisfaction with the outcomes pertaining to their actions should be both intrinsic (i.e. psychological) and extrinsic (i.e. tangible) in order for them to have undiminished participation.

Therefore, when adapting the Kuratko et al. (2005a) measurement, there was still a need to include an item that relates to managers’ perceived satisfaction regarding the entrepreneurial activity being undertaken, that is, whether the activity meets expectations in order for it to be sustained or discontinued (Morris et al. 2011:74). In this regard, the scale used to measure the construct adapted three items developed by Kuratko et al. (2005a) and one item added by the researchers, as listed in Table 1.

As presented in Table 1, the measurement scale has four items as the observable variables: (1) improvements implemented without organisational approval, (2) new ideas suggested, (3) new ideas implemented without official organisational approval and (4) managers’ satisfaction with the outcomes of implemented entrepreneurial activities of the organisation. Participants were asked to indicate their responses on a 5-point Likert scale of strongly disagree to strongly agree. Their responses indicated the extent to which they agreed or disagreed with each of the statements,

Figure 1 presents the hypothesised measurement model for entrepreneurial actions for sustainable CE.

The postulated CFA model presented graphically in Figure 1 shows the four items of interest (EA1 to EA4) and their respective error terms eEA1 to eEA4. The model provides the basis for the assessment of the measurement scale.

The classical test theory postulates that multi-item scales enhance minimisation of random measurement error as well as maximisation of measurement reliability and validity (Nunnally & Bernstein 1994). Therefore, in line with the classical theory, the latent variable entrepreneurial actions for sustainable CE were represented by four measures of the same underlying construct. As mentioned, these four measures of entrepreneurial actions comprised three items developed by Kuratko et al. (2005a) and one added by the researchers.

The hypothesised model was tested on the basis of the formed four manifest variables (indicators) as described by Hair et al. (2010) which were subjected to CFA procedures. A measurement model’s validity depends basically on two important factors (Hair et al. 2010:664): (1) establishing the measurement model’s acceptable levels of GOF and (2) finding specific evidence of construct validity. Therefore, the study proceeded accordingly and used CFA to assess the hypothesised measurement instrument. To establish GOF, the study reported on the following indices, which should provide adequate model fit evidence (Hair et al. 2010:672; Hoe 2008:77): the chi-square (χ²) value, degrees of freedom (df), one absolute index – the root mean square error of approximation (RMSEA), one increment index – comparative fit index (CFI) and the standardised root mean square residual (SRMR).

Given the high sensitivity of χ² to sample size, especially if the sample size is more than 200, the ratio between χ² and df (χ²/df), which in AMOS appears as the Chi-Square minimum discrepancy divided by its degrees of freedom (CMIN/DF), was used to solve this limitation of χ², and a ratio of 3 or less was considered to be a reasonably good indicator of model fit (Byrne 2010; Hoe 2008). Values for RMSEA range from 0.00 to 1.00, with values < 0.08 indicating acceptable model fit, although a more recent publication places a stringent upper limit of 0.07 (Steiger 2007). Values for CFI range from 0.00 to 1.00, with values ≥ 0.95 suggesting acceptable model fit, although CFI ≥ 0.90 may still be meaningful (Bagozzi 2010). SRMR values also range from 0.00 to 1.00, with values ≤ 0.08 suggesting acceptable fit while values falling below 0.05 would be indicative of well-fitting models (Byrne 2010; Hooper et al. 2008). The fit indices reveal whether the postulated model fits the data or not. An inadequate fit of the model could further be investigated to detect the source(s) of misfit. The focus of the assessment was placed on parameter estimates (assessing the fit of individual parameters) using the following three criteria (Byrne 2010): these estimates’ feasibility, standard errors’ appropriateness and these estimates’ statistical significance.

On feasibility of parameter estimates, the assessment was carried out through a consideration of viability of estimated values for parameters (i.e. whether estimates indicate correct sign and size, and are consistent with theory; Byrne 2010)). According to Byrne (2010:67), parameters demonstrating ‘unreasonable estimates are correlations > 1.00, negative variances, and covariances or correlation matrices that are not positive definite’. In this respect, estimates not meeting the criteria are candidates for deletion as they are considered non-viable. The second step considered the appropriateness of standard errors, which indicate ‘the precision with which a parameter was estimated, with small values suggesting accurate estimation’ (Byrne 2010:67). Excessively large or small standard errors are an indication of poor model fit, although currently there are no definitive criteria to determine ‘small’ and ‘large’ (Jöreskog & Sörbom 1989). The third step involved the testing of the statistical significance of parameter estimates. This was performed through the assessment of critical ratio (CR) values of these estimates. Those parameter estimates with CR < 1.96 in absolute terms are considered statistically non-significant.

Based on the large sample theory, the conduct of CFA or SEM analyses is generally premised on the assumption that the data are multivariate normal (Byrne 2010), an assumption which is specifically required in the use of AMOS (Arbuckle 2007). Of particular concern in the use of CFA or SEM is the situation when the data are multivariate kurtotic; in other words, both the tails and the picks of the multivariate distribution of the observed variables differ from the ones that are characteristic of a multivariate normal distribution (Byrne 2010). Therefore, it is procedurally expected that one inspects the normality of data prior to conducting any analyses. The study estimated multivariate normality of the data and also considered the extent of kurtosis and skewness (Sharma & Sharma 2013). Outliers, that is, cases whose scores are substantially different from all the others (Byrne 2010) are a likely cause of data skewness (McDonald & Ho 2002). Serious multivariate outliers make it inappropriate to analyse the data using the maximum likelihood (ML) estimation method and CFA processes. To detect any outliers, Mahalanobis squared distance values (D²) were computed for each case, after which the resultant D² values were reviewed. The study considered outliers in terms of their inclusion or deletion; these were detected from the univariate distribution of the variables. Factor scores can also be used to detect outliers (Bollen & Arminger 1991). According to the conventional rule of thumb, ‘data may be assumed to be normal if skew and kurtosis is within the range of ± 1.0 (some say ± 1.5 or even 2.0)’ (Schumacker & Lomax 2010:69). However, some scholars have suggested that the recommended level of skewness be a maximum of 2 and of kurtosis a maximum of 7 for a CFA with ML estimation (Curran, West & Finch 1996; Sharma & Sharma 2013; West, Finch & Curran 1995).

Bayesian estimation of hypothesised measurement model: Because the study used categorical data, the parameter estimates generated by the ML method were compared with those generated using the Bayesian estimation method in order to assess the biasness of the ML-derived estimates (Byrne 2010). If the Bayesian estimates are no different from the ML estimates, then this would speak well for the validity of the hypothesised CFA model (Byrne 2010). The given estimation method was used for the final CFA model of the proposed measurement scale.

Undoubtedly, there are a number of limitations to this study. However, these could offer opportunities for new research directions.

One of the limitations was that once the industry strata had been identified, selection of participating individuals was carried out using a non-probability sampling method in order to satisfactorily meet sampling objectives (Cooper & Schindler 2008:396). However, the use of a non-probability sampling procedure does not reflect a true cross-section of the population and is acknowledged as a limitation of this study.

Secondly, the piloting of the measurement instrument should have been much more comprehensive, using a relatively large sample size, say of at least 30 participants (Schriesheim et al. 1993). However, this was not done because of the cost implications; it is consequently acknowledged as another important limitation of the study. In addition, it is very likely that the contextual meaning of the items for the hypothesised measurement instrument may have been interpreted differently by some respondents, thereby detracting from the intended outcome in terms of responses. The piloting phase could have dealt with this aspect as well.

Thirdly, the study’s objective was to assess reliability and validity of the hypothesised measurement of entrepreneurial actions for sustainable CE, which incorporates the three-item scale developed by Kuratko et al. (2005a) to measure managers’ entrepreneurial actions as well as the added fourth item developed by the researchers. The measures of entrepreneurial actions in the original scale developed by Kuratko et al. (2005a) tend to be simplistic. For instance, one can argue that merely suggesting new ideas does not necessarily make such ideas into actions and it may not really matter whether such ideas are approved or not. Similarly, the item (‘I am satisfied with the outcomes of my organisation’s entrepreneurial activities as they meet expectation’) developed by the researchers may benefit from further research. In this respect, although the measurement instrument showed excellent reliability and construct validity, more research still needs to be undertaken in order to refine the instrument theoretically and empirically, based on the premise mentioned earlier that entrepreneurial actions are the conduit through which CE is practiced in existing organisations (Hitt et al. 2001). Moreover, satisfaction with organisational outcomes in relation to the implemented entrepreneurial activities is a critical aspect of these entrepreneurial actions for sustainable CE (Simon & Shrader 2012:291).