Organizational Social Entrepreneurship: Scale Development and Validation

Social entrepreneurship has become a growing field of research interest. Yet, past research has been held back by the lack of a rigorous measurement instrument. Rather than defining social entrepreneurship as an organizational form that a venture does or does not have, this paper agrees with Dees and Anderson (2006) that the construct is better thought of as a set of practices, processes, and behaviors that organizations can engage in to a higher or a lesser degree. In other words, the construct is a set of behaviors that any organization can engage in. We refer to this construct as organizational social entrepreneurship (OSE). Drawing on previous literature, this paper first develops and then validates scales for measuring OSE as a third-order formative construct. As its second order the OSE scale includes three components that capture the heterogeneity of the OSE concept: social change intention, commercial activity, and inclusive governance. The OSE scale is developed and tested through a sample of 182 nascent social enterprises from 55 different countries the world and then revalidated using a second sample of 263 mature social enterprises from six European countries. Results suggest that the scale items exhibit internal consistency, reliability, construct validity, and nomological validity.


INTRODUCTION
In the past years, there has been an increased academic interest in social entrepreneurship and the organizations that engage in this practice (Dorado & Ventresca, 2013;McMullen & Bergman, 2017;Meyskens, Robb-Post, Stamp, Carsrud, & Reynolds, 2010;Miller, Grimes, Mcmullen, & Vogus, 2012;Miller & Wesley II, 2010;Nicholls, 2010;Zahra, Gedajlovic, Neubaum, & Shulman, 2009;Zahra, Rawhouser, Hawe, Neubaum, & Hayton, 2008). Although the exact nature of social entrepreneurship has been widely discussed over the past decade, we still lack quantitative instruments that allow researchers to measure it (Short, Moss, & Lumpkin, 2009). This is partially because the term is used quite loosely to describe phenomena at three different levels of analysis. Firstly, at the personal level, the term is sometimes used to describe individuals who change institutional settings. For example, Bill Drayton, the founder of Ashoka, highlights Florence Nightingale, an English social reformer and the founder of modern nursing, and William Lloyd Garrison, an American abolitionist and proponent of women's suffrage, when he talks about historical examples of social entrepreneurs (Drayton, 2002). Secondly, at the societal level, the term is used to describe social change such as that caused by social movements (Alvord, 2004;Rao, 2004) in order to fill institutional voids (Mair & Marti, 2009;Mair, Martí, & Ventresca, 2012;Tobias, Mair, & Barbosa-Leiker, 2013). Thirdly, at the organizational level, the term is used to describe the activities of charitable social purpose organizations that use innovative market-based approaches to create social impact following the principals of cooperative rather than shareholder governance (Dees, 2012;Defourny & Nyssenes, 2012).
Following Mair and Marti (2006) this article conceives of social entrepreneurship as a process. Moreover, it focuses on the organizational level only. We propose the term organizational social entrepreneurship (OSE) to delimit our construct from other interpretations of social entrepreneurship. OSE in this paper is defined as the actions of organizations aimed primarily at creating social impact for beneficiaries by engaging in commercial activity while employing cooperative governance mechanisms that uphold the primacy of the beneficiary.
At a theoretical level, this paper identifies the ontological nature of OSE, while exploring its most important dimensions (i.e., social change intention, commercial activity, and inclusive governance) and sub dimensions (i.e., social mission adhesion, change making intent, earned income generation, salaried employees, democratic decision making, and stakeholder participation). Methodologically, our research advances a validated set of measures that allow conceiving of OSE as a third order formative measure. We expect this measure to be useful for future research into both the antecedents and outcomes of the OSE dimensions and sub dimensions.
Since such a fine-grained instrument did not exist before, research into OSE had to rely on more generalized measures, which do not reflect the full breadth of the construct. Finally, at an empirical level, we contribute evidence that the measure advanced here matches the self-evaluation of both nascent and mature social entrepreneurs.
Why do we need a measure for organizational social entrepreneurship? Such a measure will allow studies on the effect of activities promoting OSE, such as for example public policies (Rawhouser, Cummings, & Crane, 2015), educational initiatives (Miller, Wesley, & Williams, 2012) as well as the effect of so called impact investors (Geobey, Westley, & Weber, 2012;Ruttmann, 2012). A clear measurement model will help study the usefulness and cost effectiveness of such promotional activities. It will also allow comparing incidences of OSE in different countries or in different organizational forms.
We purposefully avoid the term social enterprise 1 for two reasons. Firstly, the expression tends to refer to organizational forms rather than organizational processes. What is considered a social enterprise thus varies widely between countries depending on the legal frameworks. In Finland, for example, since 2004, any firm that has at least 30% of disabled or long-term unemployed people among its employees can be registered as a social enterprise with the Ministry of Labour (Pättiniemi, 2006). In the UK, on the other hand, the social enterprise mark requires organizations to have an asset lock in place and to guarantee to reinvest at least 50% of its profits in line with its social mission (Ridley-Duff & Southcombe, 2012). In the U.S., the legal definition of a social enterprise even differs from state to state (Brakman Reiser & Dean, 2015;Rawhouser et al., 2015).
Secondly, by trying to define which organizational form counts as a social enterprise one implicitly excludes a large number of organizations. Such a limitation of the term ignores the fact that OSE is actually a process that can take place in any type of organization. We would argue that OSE describes a set of activities that although particularly prevalent in certain organizational forms, might in fact be found in some traditional charities as well as in profit seeking businesses that would not at first glance be considered as social enterprises. In fact, we conceive of OSE as a continuum on which organizations can be placed. Over time as their behavior changes, these 1 Please note that as used by us, the term enterprise refers to it being an organization. We acknowledge that the term enterprise can have different meanings and could also be used in the sense of "readiness to engage in daring or difficult action" (Merriam-Webster, 2018). To avoid confusion, we will in this paper use the term exclusively to refer to the first definition. However, by this we do not deny the fact that it might be legitimately used differently by other authors.
organizations' positioning on that continuum changes too. This viewpoint even allows the question of whether some organizations that are labeled as social enterprises actually engage in OSE activities even though their organizational form is considered a social enterprise. In return, we can ask whether OSE practices might not also be found in traditional businesses or charities as well. 2 One approach to measuring OSE could be to develop a unidimensional construct such as organizational hybridity. Thus, the more an organization mixes activities of both charities and forprofit organizations, the more it could be considered to engage in social entrepreneurship. On the other hand, the more their behavior aligns with only one of these two types, the less this would be the case. There have been academic efforts to develop unidimensional construct to measure social entrepreneurship (Carraher, Welsh, & Svilokos, 2016). However, such an approach would ignore the conceptual richness of the construct (Brinckerhoff, 2000;Choi & Majumdar, 2014;Mair & Martí, 2006;Mort, Weerawardena, & Carnegie, 2003;Peredo & McLean, 2006;Young & Lecy, 2014), which is too complex to be operationalized as unidimensional (Weerawardena & Mort, 2006).
Far from being homogeneous, social enterprises come in a bewildering array of configurations. So far, quantitative studies in the field of social entrepreneurship (Battilana, Sengul, Pache, & Model, 2015;Liu, Eng, & Takeda, 2013;Mair, Mayer, & Lutz, 2015;Meyskens et al., 2010;Stevens, Moray, & Bruneel, 2015) have utilized mainly homogenous samples of social enterprises and have not been able to differentiate between various configurations of social 2 For the purpose of this paper we conceive of traditional charities as organizations that can receive taxdeductible donations, and of profit seeking businesses as all organizations that are run for the purpose of creating shareholder value. enterprises. Therefore, this paper proposes a multi-dimensional measure for OSE. We argue that OSE is best represented by three components.
In refining a multidimensional construct, scholars can take two possible paths. The more often selected approach is to think of the construct as reflective (Coltman, Devinney, Midgley, & Venaik, 2008). In a reflective measurement model, we assume that the indicators of a construct are caused by a construct. They can thus be expected to be correlated with each other and they thus should exhibit internal consistency reliability (Jarvis, MacKenzie, & Podsakoff, 2003). A formative measurement model, on the other hand, hypothesizes that the individual indicators influence (or form) the latent construct. While they may be correlated, this is not a requirement.
Indeed indicators of a formative measure can be completely uncorrelated (Jarvis et al., 2003). We propose that the extent to which the components of OSE are to be found is contingent on the context, such as the sector, or internal factors, such as the organizational structure. Thus, we assume that the components of organizational social entrepreneurship may vary independently, depending on the context. This is consistent with Gartner's view that traditional entrepreneurship is also a multidimensional phenomenon forming "complex and unique combinations" (Gartner, 1985: 697).
Our intention in advancing the OSE scales is not only to allow scholars to study variations in the components of OSE, but we also want to encourage more quantitative work regarding the antecedents of OSE behavior and the question of how they differ in terms of economic, social, and environmental outcomes. By this, we are responding to the call by Short, Moss and Lumpkin (Short et al., 2009) for more quantitative research in the field of social entrepreneurship.
Hence, our research questions deal with the following two issues: What are the major components that define organizational social entrepreneurship? How can these components be operationalized and measured? As a first step towards answering these questions, this paper develops and validates measures of three second order components of social entrepreneurship, namely: social change intention, commercial activity, and inclusive governance (see figure 1 for an overview about the underlying formative measurement model for organizational social entrepreneurship).

SOCIAL ENTREPRENEURSHIP LITERATURE
A first step towards developing an OSE scale is an appreciative reading of prior literature on social entrepreneurship as it relates to organizational behavior. From an extensive review of possible sources, articles were selected according to the following procedure.
In selecting the literature to be considered in this analysis we had several options. One way would have been to select the most cited articles and/or to cover only contributions published in journals that had a minimum journal rank. We were worried that such an approach would reduce diversity of ideas and concepts. Moreover, such publications although undoubtedly influential are not always the original source. In fact, they often repackage ideas developed earlier expounded in book chapters or in working papers. Our goal was to achieve conceptual exhaustiveness and to attribute new concept to the authors who first advanced them.
To achieve this we have begun our process from one of the earliest texts (Dees, 1998) (Dees, 1998: 4-5). Next we have also considered a later text by Dees in which he further develops his notions of social entrepreneurship clarifying and extending the 1998 piece (Dees & Anderson, 2006). From these two foundational definition papers we have derived the three main components which underpin our model (social change intention, commercial activities, and inclusive governance).
In a next step we analyzed papers that, although building on this foundational contribution, have advanced it either through more nuanced definitions or have added substantial new concepts. Following the tenets of theoretical saturation, we have not included texts that simply rephrased definitions that were essentially covered in earlier contributions.
This has led to the following papers being retained. Firstly, we considered Austin et al. (2006) because they build on Dees (1998) by very clearly stating that "social entrepreneurship […] can occur within or across the nonprofit, business, or government sectors" (Austin et al., 2006: 2). This opens up social entrepreneurship to one of the central tenets of this paper namely that OSE can be found in many organizations and is not exclusive to only one type of legal organizational form.
Following these two initial readings we have then followed up on the three themes deduced from the original Dees text. Firstly, this has led us to consider several papers that further refine the social change intention component of social entrepreneurship.
The two early contributions lacked an element that is critical to entrepreneurship research namely the idea that once opportunities have been recognized they tend to change institutional systems. The first to make this construct part of their social entrepreneurship definition were Mair and Marti (2006)  This notion is further detailed by Hockerts (2007) who draws on a Schumpetrian perception of entrepreneurship when he elaborates that the "discovery […] of opportunities to create public goods […] is usually done through the generation of disequilibria in market and non-market environments" (2007: 422). He later expounds on this idea when he proposes that it is the conversion of antagonistic assets into complementarities that is at the heart of the social entrepreneurship process (Hockerts, 2015). They also extend on Dees' original inclusive governance component by specifying that social enterprises favour a "high degree of stakeholder participation […]. Customer and stakeholder interests are accounted for in decisions, and the management style is democratic" (Borzaga et al., 2008: 32). Zahra et al. (2009)  The above mentioned papers were added following a logic of theoretical sampling. They were chosen to increase the diversity of the definitions brought forward. Papers were added only until all additional articles kept surfacing already previously established concepts again, a point at which we considered theoretical saturation to have been reached. Table 1 represents the twelve articles that we retained eventually for developing the OSE scale items. We include not just the original contributions of each text but also cover elements of the social entrepreneurship definition that echo earlier contributions. Together we feel that these texts exhaustively cover the domain with each paper providing at least one original nuance or contribution to the definition of social entrepreneurship.
Through the above described process we have aimed to cover all major definitional streams. Our goal was to achieve variance in terms of geographical areas (covering the predominant Anglo-Saxon contributions as well as varied continental European voices). We have also aimed to include contributions that were centered around the individual (the social entrepreneurs), the organization (the social enterprise), and the societal change (social innovation).
We are aware that the results of our review are eclectic, but we maintain that they are collectively exhaustive and do tend to pinpoint emerging ideas to the authors who first advanced them.  Defourny and Nyssens (2010). However, they have added a third element, which we term the democratic governance component. Drawing inspiration from the cooperative movement, Defourny and Nyssens (2010) argue that democratic organizational governance is more likely to achieve an equitable distribution of profits and assets than would be achieved through capital ownership governance. We accordingly propose that inclusive governance should also be a component of organizational social entrepreneurship.

Social Change Intentions Component
The social change intentions component plays a central role in both Dees and Anderson (2006) and Defourny and Nyssens (2010). Often, this process is also referred to as social innovation or change making (Bornstein, 2007;Drayton, 2002). This component draws strongly on the Schumpeterian view of entrepreneurship as the discovery and profitable exploitation of business opportunities by generating market disequilibria (Drucker, 1986;Shane, Venkataraman, & Shane, S., & Venkataraman, 2000), which in turn allow the generation of so called entrepreneurial rents (Alvarez & Barney, 2007). However, as Schumpeter (1934) points out entrepreneurial profits are only transient. Competitors are likely to emulate the innovation strategy of a successful entrepreneurial venture, thereby competing away the entrepreneurial profit. At this point, most entrepreneurial enterprises become merely "optimizing firms" (Schumpeter, 1934:133) unless they can identify a new entrepreneurial opportunity and exploit it.
Following Hockerts (2007Hockerts ( , 2010, this paper accordingly posits that the social innovation component of organizational social entrepreneurship entails the creation of market and non-market disequilibria through the discovery of opportunities for generating social impact and the identification of a mechanism to do so in a financially sustainable 3 way. However, just as with traditional for-profit innovators, the disequilibria created are not permanent. As their activities are observed by market and non-market players, the successful ideas of social enterprises will be emulated. This can have two transformative consequences. On the one hand, non-market players (i.e. charities, NGOs, governments) are likely to mimic the behavior of the social venture aiming 3 We conceive of financial sustainability as follows: Any situation in which an organization acting upon the discovered opportunity can create enough income to cover their operating costs without having to diminish their capital stock. to improve their social impact. On the other hand, for-profit firms will attempt to capture part of the economic rent created by the social venture. In adopting certain practices of the social venture, both market and non-market players propagate the innovation and increase the social impact generated (Hockerts, 2010).
Organizational social entrepreneurship with a predominant social change intention component aims to transform methods of social value creation by shifting resources to previously neglected areas, ultimately leading to higher yield for society (Dees & Anderson, 2006;Leadbeater, 1997). Organizations engaging in such activities aim to create lasting systemic change through the introduction of new ideas, methodologies, and changes in attitude (Kramer, 2005). The emphasis is more on outcomes produced at the societal level than on the performance of a single organization. Based on this discussion, we advance the first component of social entrepreneurship to be social change intention, defined as the creation of market and non-market disequilibria through the discovery of opportunities to generate social impact and the promotion of these practices among an organization's peers and the sector in general.
The question can be raised whether we are here conflating social impact and social innovation. We would argue not. Our line of reasoning here follows Schumpeter (1934) who differentiates between the entrepreneur (who creates market disequilibria) and the optimizing manager who creates performance without being innovative. Schumpeter considers optimizing managers to be legitimate market participants. However, he does not consider them to be acting entrepreneurially.
Transferred to our field of interest this means that a completely non-innovate social sector organization with a high social impact would not exhibit social entrepreneurship behavior. Within the Schumpeter definition (and our view) such an organization might well be considered a social economy organization, it would just not be behaving entrepreneurially and hence would not be exhibiting social change intentions which we stipulate to be part of the OSE scale.

Commercial Activity Component
Rather than relying on charity or government grants to solve social challenges, organizational social entrepreneurship has been described as using strategies to create revenue from market operations to achieve a social mission to create impact for a defined group of beneficiaries (Robinson & Klein, 2002). This is referred to as earned income by Dees &Anderson (2006) as well as Defourny & Nyssens (2010). As traditional routes of funding from donors and public social welfare projects have dried up, many non-profits are adopting a new strategy of developing a revenue stream to sustain the organization (Frumkin, 2009;Skloot, 1983). Earned income generation may also be preferable to the time-consuming processes of application and reporting required to attract and retain grants and donations (Frumkin, 2009;Skloot, 1983).
The aim of the earned income component is to provide a solution to imperfect matches between private resources and public needs (Frumkin, 2009). This component can include accepting government contracts for delivering social services, charging corporations for doing work, charging the end customers of the organization for services that used to be free previously (Dees, 1998), or sale of products and services to supporters for revenue generation (Kerlin, 2006).
The commercial activities could be closely related or unrelated to the social mission of the organization. Drawing on the literature discussed above, we thus propose commercial activity as the second component of organizational social entrepreneurship. We define the variable as the degree to which an organization engages in commercial operations to meet its social mission by generating revenues through the selling of goods and services on private and public markets.

Inclusive Governance Component
The first two components of organizational social entrepreneurship tend to be present in most previous conceptualizations in the existing literature to a greater or lesser degree. However, the third element advanced here is less explicit in many definitions. Actually, the degree to which stakeholders (Freeman, 1994) and in particular beneficiaries are part of an organization's decision making processes has been primarily advanced by European social entrepreneurship scholars such as Defourny and Nyssens (2010). Rather than basing decision power exclusively on capital ownership or an all-powerful charity board, Defourny and Nyssens argue that democratic decision making (Low, 2006) is a third formative component of social entrepreneurship. They base their argument primarily on literature tracing the origins of modern social entrepreneurship to the cooperative movement: "[A] democratic control and/or a participatory involvement of stakeholders reflects the quest for more economic democracy in the tradition of cooperatives" (Defourny & Nyssens, 2010).
Why is this third element needed? Doesn't it suffice that an organizations aims to create positive social change using a variation of an earned income strategy? Actually, in the absence of the inclusive governance element it is difficult to differentiate social entrepreneurship from corporate social responsibility (CSR), which tends to be conceptualized as the actions of firms to contribute towards societal welfare (Cornelius, Todres, Janjuha-Jivraj, Woods, & Wallace, 2008).
What sets social entrepreneurship apart from CSR is intentionality. In CSR, shareholder value creation remains the primary motivation. Social issues are considered either as antecedents of profit creation or as limitations to profit seeking resulting from ethical norms and standards. In social entrepreneurship, creation of social value for beneficiaries tends to be the primary motivator, with shareholder value creation being either a minor motivation or not present at all. The degree to which stakeholders and in particular beneficiaries are involved in organizational governance is a proxy for the question of intentionality. 4 Inclusive governance has been identified as an important element of social entrepreneurship helping organizations deal with conflicting social and financial objectives (Ebrahim, Battilana, & Mair, 2014;Reay & Hinings, 2005). Research in Europe (Borzaga & Defourny, 2001;Mair et al., 2015;Pestoff & Hulgård, 2015) as well as in the U.S. (Young & Lecy, 2014) has found democratic governance to be a key component of organizational social entrepreneurship. The component is particularly strong among cooperative societies and organizations comprising the social economy in Europe Evers & Laville, 2004). Organizational social entrepreneurship thus extends traditional stakeholder theory (Freeman, 1984(Freeman, , 1994 into the realm of collective social purpose organizations (Huybrechts, Mertens, & Rijpens, 2014) by advocating the involvement of multiple stakeholders and in particular the local community (Low, 2006) as well as the intended beneficiaries (Yunus, Moingeon, & Lehmann-Ortega, 2010). 4 We use the term stakeholder in the sense that it refers to "any group or individual who can affect or is affected by the achievement of the organization's objectives" (Freeman, 1984: 46). By beneficiary we refer to those type of persons that the OSE behavior intentionally aims to aid and support as part of the organization's social mission. This means that beneficiaries are one type of stakeholders among many. Other stakeholders could be supporters and employees of the organization, investors, the government, or social organizations that work on behalf of the beneficiaries.
Inclusive governance in organizational social entrepreneurship can include beneficiaries utilizing the services of the social enterprise, employees, shareholders, foundations, customers, donors, and governments (Low, 2006). In particular, stakeholder participation in social entrepreneurship has been suggested to help avoid mission drift (Cornforth, 2014;Ebrahim et al., 2014;Ramus & Vaccaro, 2014). Democratic governance is characterized by an independent and participatory nature in the daily operations and decision making of the organization (Low, 2006).
Enhancing the representation of beneficiaries and all stakeholders in the decision making of the social entrepreneurship is also important for maintaining accountability (Ebrahim at al., 2014).
Consequently, we posit that the third component of organizational social entrepreneurship is inclusive governance, which we define as the degree to which stakeholders and specifically beneficiaries are included in decision-making, particularly pertaining to decisions about the social mission and use of profits generated.

FORMATIVE VERSUS REFLECTIVE CONCEPTUALIZATION OF ORGANIZATIONAL SOCIAL ENTREPRENEURSHIP
In their study of measurement, models presented in marketing journals Jarvis et al. (2003) found that up to a third of reflective measurement models should actually have been specified as formative. This indicates that traditionally a rather large amount of misspecification has tended to exist. It is a central contention of this paper that OSE should be modeled as a formative construct.
In coming to this conclusion, we have followed a purposeful evaluation process as set out by previous literature (Coltman et al., 2008;Jarvis et al., 2003) that identifies four distinct steps: directionality, interchangeability, covariance, and the nomological net.
A first step to decide on whether a construct should be specified as formative or reflective has to do with directionality. In other words, are the indicators defining characteristics or manifestations of the construct? For a construct to be reflective, any change in the construct would be considered to cause change in all its indicators. On the other hand, in a formative model one would expect the indicators to cause change in the latent construct. Using formative measures implies that the causal relationship goes from the lower-order latent variables to the formative construct. Thus, not every increase in the higher-order formative construct implies a commensurate increase in all its components. In other words, the indicators form the latent construct, and the changes to the indicators cause changes in the latent construct. Practically, the components sum algebraically to give the higher-order construct (Diamantopoulos, 2008;Diamantopoulos & Winklhofer, 2001;Jarvis et al., 2003).
Secondly, a decision on whether to model a construct as reflective or formative requires an answer to the question whether indicators have the same or similar content and thus are interchangeable. Formative constructs have several dimensions, each embodying a significant element of the construct. As a result, one cannot drop or interchange variables from a formative construct without reducing content validity (Bollen & Lennox, 1991).
Covariance of the indicators is a third pointer. A reflective measure would imply that change in one of the indicators would be associated with commensurate changes in the other indicators. A lack of covariance, on the other hand, would hint at a formative measure. The fourth and final criterion suggested by Jarvis et al. (2003) concerns the nomological net of the construct indicators. In a reflective measure, all indicators would be expected to have the same antecedents and consequences.
A closer look at the directionality of the OSE construct reveals that its components (social change intentions, commercial activities, and inclusive governance) are actually the source of the OSE construct rather than its result. It is, for example, not earned income that is the result of organizational social entrepreneurship. Instead, it is an earned income strategy that causes an organization to be perceived as engaging in OSE.
Let us look at directionality in some more detail. Proponents of the view that OSE is a reflective measure might argue that the decision to start a social enterprise goes along with the selection of a social enterprise legal organizational form. In other words, the choice of legal form causes a set of behaviors. This would point towards a reflective nature. We see two problems with this argumentation. Firstly, the empirically observed organizational forms identified as social enterprises are varying widely. In the Finish law on social enterprises it is specified that any such firm has to have at least 30% of disabled or long-term unemployed (Pättiniemi, 2006). If we adopt a reflective interpretation of OSE this would mean that having disabled employees is reflective of being a social enterprise and that any reduction in this one indicator would go along with commensurate reductions in other indicators (such as for example earned income). This does not sound convincing, particularly once we realize that in the UK a community interest company (CIC) is completely differently specified and does not require the employment of disabeled people.
Which view then is the right one?
It is our suggestion that the selection of organizational form is not a good starting point for identifying OSE behavior. Actually, we would argue that quite a number of organizations that have selected an organizational legal form, that is supposedly a social enterprise do not exhibit all types of behavior that we would argue should be identified with OSE.
We would submit that OSE behavior might very well be found in many organizations that have not selected that legal form. For example, we are convinced that even in traditional for profit companies we may find some (limited) type of OSE behavior in one of their business units. On the other hand, when studying the list of CICs in the UK, we find that many flagship social enterprises have decided against adopting a CIC form, whereas many organizations that are perhaps part of the social economy have adopted CIV forms, but they do not exhibit all the characteristics that the literature above suggests should be found in social entrepreneurship.
A key advantage of considering OSE a formative behavior is that it allows us to look at organizations with a social enterprise form and to follow over time the degree to which their OSE behavior varies. This would allow for the possibility that the changemaking or the inclusive governance component of a social enterprise falls as social enterprises mature.
Summing up, we maintain that it is not just possible but actually necessary to conceive of the directionality for OSE behavior as running from the items to the construct. This establishes OSE behavior clearly as a formative measure. We do, however, realize that a different measure is conceivable in which the elements of the legal form are taken as the starting point and conceived of as reflective. However, such a measure would in our view not be presentative of organizational behavior and thus not be useful for this paper." In a second step, we have evaluated interchangeability of the components of OSE. It is obvious that dropping one of the components of OSE would significantly affect its content validity.
This again points at a formative measure. Thirdly, a quick study of the empirical reality shows wide variations in how OSE is practiced. On the one hand, there are organizations engaging in social entrepreneurship, such as ChildLine India Foundation, that tend to prioritize social change, disregarding commercial activities and inclusive governance as afterthoughts. Social economy organizations, on the other hand, tend to favor earned income and community-driven governance (Haugh, 2007). This indicates that covariance between the components of social entrepreneurship is limited again speaking in favor of a formative construct.

Change Intentions
In all the literature concerning social enterprises social mission is considered an essential factor. It can be conceived of as a declaration of an organization's core purpose and focus that unambiguously states the organization's social value creation intention and the beneficiaries that are expected to accrue that value. Hence, the first component that we propose here is the social mission. As social organizations adopt earned income strategies (and possibly even engage in dividend seeking), they are at risk of losing sight of their social mission. This kind of "mission drift" (Christen & Cook, 2001) should be avoided. Hence, an organization strong in socially entrepreneurial behaviour will have the expectation that it will help safeguard the organization's social mission.
A criticism of this approach could be that we are not actually measuring the content of the social mission with these items. The reason for this lies in our conviction that whether or not an organization has a social mission is largely socially constructed. Thus, some might consider an abortion clinic aiming to provide services to marginalized women to have a social mission.
Abortion critics would disagree with that view and consider the organization socially irresponsible.
For this reason, we do not try to capture the actual content of the mission but instead use the strength of commitment to that mission as our proxy.
The second component of social change intentions is the degree to which a social enterprise wants to create change in the existing system. In other words, to what degree do social enterprises want market and non-market competitors to adopt their innovative ideas? Prior literature suggests that organizations engaging in social innovation will tend to aim at putting themselves out of business (Tafel, 2011;Thorpe, 2015) by transforming the market and nonmarket environments they operate in until their innovation has become the quasi-standard (Hockerts, 2010).
Given the goal of changing practices among peers and in the sector, social innovation is more likely to be associated with organizational behaviour that is significantly different from the status quo (Gundry, Kickul, Griffiths, & Bacq, 2011). In particular, it requires that social entrepreneurship leads to the identification of previously unidentified opportunities (Corner & Ho, 2010) to help beneficiaries and create social impact.

Earned Income and Salaried Employees as Components of Commercial Activities
Financial sustainability is considered as a differentiating factor between social enterprises and charities. By generating income from market sources, they are less or not at all subject to charity legislation and the constraints of grant-givers. Earned income in a social enterprise is usually generated by selling products and services (Defourny & Nyssens, 2010) . Hence, we argue that earned income generation is a first component of commercial activities.
The second component, is derived from extant literature, which suggests that the presence of commercial activities in an organization with a social mission is the results of an increased use of salaried employees (Defourny & Nyssens, 2010). Whereas, many traditional non-profits rely largely on volunteer work (Defourny & Nyssens, 2010), the presence of commercial activities requires the presence of a professional workforce. By offering competitive salaries to their employees, social organizations are more likely to attract employees that will engage in commercial activities (Frumkin, 2009).

Components of Inclusive Governance
For inclusive governance, the two possible components could be considered part of it.
Firstly, drawing on user-driven innovation (Baldwin, Hienerth, & von Hippel, 2006;Urban & von Hippel, 1988;von Hippel, 2001) and open innovation (Chesbrough & Chesbrough, 2006) literature, we can expect that a participatory decision making approach is likely to favour unexpected outcomes and more radical innovation. Based on these reflections, we can thus expect that an organization favoring socially entrepreneurial behaviour will have democratic decisionmaking.
The second factor is the stakeholder participation in the organizational activities. Yunus et al. ( 2010), for example, stress the importance of stakeholder involvement during the early development process of social entrepreneurship. Stakeholders can include individuals who benefit from the social mission of the organization, employees of the organization, and investors in the organization (Low, 2006). Hence, we argue that an organization strong in socially entrepreneurial behaviour will have more involvement of stakeholders in developing its products, services and activities. Prior research, for example, highlights the risk that a "focus on profits [could be] to the

Development of Scale Items
We initially generated a pool of 90 potential items to represent each dimension of social entrepreneurship as well as the outcome variables we planned to measure as part of the nomological validity test. Items were initially based on a review of the previous literature and then refined to identify ambiguous and unclear wording. In particular, items were discussed in detail with five organizations engaging in social entrepreneurship and feedback was used to delete and modify items. Moreover, to assess face validity, this list of scale items was sent to 19 social entrepreneurship researchers who were asked to assess whether items represented correctly the definition of each dimension of social entrepreneurship.

Scale Purification
In order to determine the factor structure of social entrepreneurship and its outcome variables as well as to purify the measurement instrument, the psychometric properties of the scale were evaluated through a pilot survey based on data collected from 143 Scandinavian social enterprises. The initial scale items were modified based on the results of the exploratory factor analysis. An iterative process eliminated and revised items that had factor loadings below .50 and/or cross-loadings above .40 (Hair, Black, Babin, & Anderson, 2009). Each of the survey items has a five-point Likert scale (Hinkin, 1998), ranging from strongly disagree to strongly agree. This iterative process of qualitative discussions with social entrepreneurship practitioners, researchers and pilot testing eliminated 56 items, leaving 20 scale items.

Sample 1
Following the pilot study, the revised questionnaire was sent to a sample of 3145 nascent social entrepreneurs enrolled in a massive open online course (MOOC) that helps them to develop a business plan for their emerging social enterprise. The survey was sent at the end of the 12-week MOOC course and resulted in 182 usable responses from social entrepreneurs in five countries.
The low response of 5.8% is not surprising since the majority of MOOC participants tend to drop out during the course, and only a minority actually finishes the 12-week program. A total of 54.9% of the respondents were female, the mean age of the respondents was 31.75, and the standard deviation was 11.65. Comparing these descriptive statistics with those of the 3145 participants enrolled at the beginning of the course, we do not find any statistically significant difference.

Sample 2
For our second sample of data, we sent the survey questionnaire to the email address of a were from the United Kingdom, 100 were from Scandinavia, and the remaining 51 were from the Netherlands. The average number of years in operation for the social enterprises was 14 years. We tested for non-response bias by comparing latent variable scores for the first quartile of respondents and the fourth quartile. No statistically significant differences were observed (Armstrong & Overton, 1977).
To verify the factor solution of the exploratory factor analysis in the pilot sample, confirmatory factor analyses were conducted on both sample 1 and sample 2. The results presented in Table 2 show that most factor loadings are above 0.7, suggesting that all items fit well with the   (Fornell & Larcker, 1981  Finally, we have also tested for multicollinearity. The values of the Variance Inflation Factor (VIF) vary from 1.100 to 2.0 for sample 1 and vary from 1.1 to 1.5 , which are all well below the recommended cutoff value of 5 (Hair et al., 2009). This indicates that the components of the social entrepreneurship construct do not correlate perfectly. This is crucial as multicollinearity undermines the assessment of a multidimensional construct (Diamantopoulos & Winklhofer, 2001;Hair et al., 2011).
Taken together, the results of the CFA and the further tests suggest that the scale items are good indicators for the first order reflective components. In particular, the components of the organizational social entrepreneurship construct have good reliability and validity.

Formative Construct Validation
For specifying the second order formative construct of organizational social entrepreneurship comprising of social change intentions, commercial activities, and inclusive governance components, we used the repeated indicator approach mode B (Becker, Klein, & Wetzels, 2012;Lohmöller, 1989). Table 4 and Table 5

Nomological Validity Testing
In a next step we tested the nomological validity of our proposed scale. To do this one has to place the proposed scale into a "nomological net" based on established theoretical constructs within which a proposed new scale is embedded. Nomological validity thus means that a new construct is being analyzed in relation to a measure that is known theoretically to be related to the advanced scale in a way by the theory that the construct is derived from.
In our case we have proposed two such measures. First, we have asked respondents to estimate their social impact. This latent variable was measured through three items (Our organization has a large scale social impact; Our organization has more social impact than other social enterprises in the country; Our organization has more social impact than other social enterprises globally) each on a five-point scale (from strongly disagree to strongly agree; Sample 1: mean =3.57; SD=.85 and Sample 2: mean =###; SD=.##).
The underlying theoretical assumption here is that increased OSE should go along with increased social performance due to adherence to the social change intention, the commercial activities (which allows for profitable growth), and the inclusive governance (which allows for stakeholder inputs).
Our second variable is a self-evaluation in which we asked respondents to what degree they felt that the term social enterprise was a good descriptor for their organization. Our hypothesis here was that organizations with higher OSE would be more likely to agree with the statement. We again measured on a five-point scale (from strongly disagree to strongly agree; Sample 2: mean =###; SD=.##). The second measure was only collected for the second sample of mature social enterprises and was supposed to account for the fact that in our sampling we had purposefully included organizations that had indicated interest in social entrepreneurship but had not necessarily selected the legal form of a social enterprise.
The nomological validity of the scale items were confirmed as we found a positive correlation between the third order organizational social entrepreneurship construct and perceived social impact of the organization in the nascent social enterprise sample (β =0.355; R 2 = 0.126). In addition, for the mature sample, we found a positive correlation between organizational social entrepreneurship construct and the perceived social impact (β =0.314; R 2 = 0.099) as well as the degree to which the organization identified themselves as social entrepreneurial (β =0.246; R 2 = 0.061).

DISCUSSION
It is only in recent years that social entrepreneurship research has been moved from qualitative theory building towards quantitative research. Clearly stated constructs formulated in a style that is comprehensible to social entrepreneurs is a key step in such a process. Extant literature suggests that organizational social entrepreneurship is a complex concept with potentially conflicting components (Choi & Majumdar, 2014;Young & Lecy, 2014). Therefore, authors such as Short et al. (Short et al., 2009) andHoogendoorn et. al ( 2010) have called for more multivariate analysis and hypotheses testing allowing a more in depth analysis of the complexity of the social entrepreneurship field. To do so requires a realization that there are multiple subtypes of social entrepreneurship (Hockerts, 2017). However, to date there are no validated scales to measure the components of this multidimensional construct.
We address this shortcoming by developing a multidimensional construct for organizational social entrepreneurship. In particular, we propose scale items for three central components of social entrepreneurship, namely social change intentions, commercial activities, and inclusive governance. Our scales thus measure the three formative dimensions identified by Dees and Anderson (2006) as well as Defourny and Nyssens (2010). All items in our instrument take previous literature as their point of reference.
The scale items were modified and purified after detailed discussions with social entrepreneurs and social entrepreneurship researchers, as well as an initial test on a pilot sample.
We performed our current study with two separate samples of data, the first one with 182 nascent social enterprises and the second one with 263 mature social enterprises.
We demonstrate that organizational social entrepreneurship can indeed be modelled as a formative construct composed of social change intentions, commercial activities and inclusive governance. Results show that the three components significantly contribute to the formative construct. To the best of our knowledge, ours is one of the first attempts to capture the components of organizational social entrepreneurship in a formative manner. Moreover, both samples support the validity of the instruments developed. Scales are reliable and valid when tested on two separate samples. By using two quite dissimilar samples, we also demonstrate the applicability of the scale in varied contexts. The first sample consisted of nascent organizations from 55 countries and second sample consisted of mature organizations from six countries highlighting the applicability of the model in international contexts. In particular, to prove nomological validity, we tested the correlation between the formative construct and the perceived social impact and the degree to which the organization identified themselves as social enterprise. It proves that the formative construct behave in ways consistent with prior literature.
The scale presented here offers an important new venue for social entrepreneurship theorizing. Firstly, it draws attention away from the eternal question of which organizational form should be counted as a social enterprise and where to draw the line. Instead it allows scholars to take a broad approach towards a diverse field and study OSE behavior in any empirical field where it may occur.
Next, the scales also allow for more focused theorizing. Scholars are held to delve in on the antecedents of all three components presented here as well as to study the different performance effects they have in terms of likelihood to survive, growth rate, or potential to achieve financial sustainability.

LIMITATIONS AND FUTURE RESEARCH
In this study, we propose the concept of organizational social entrepreneurship as multidimensional. This contribution is in line with extant studies that conceptualized other entrepreneurship constructs as formative, such as, for example, effectuation (Chandler, DeTienne, McKelvie, & Mumford, 2011) and entrepreneurial passion (Cardon, Gregoire, Stevens, & Patel, 2013).
The advantage of formative constructs lies in their inherent heterogeneous nature that allows one to represent the multidimensionality and complexity of constructs such as social entrepreneurship. The formative construct of organizational social entrepreneurship in our study consists of social innovation, earned income and democratic governance components. By presenting scale items for these three components, we are offering a tool for measuring various subtypes of social entrepreneurship. In addition by including the democratic governance component (Defourny & Nyssens, 2010), we are extending the discussion in social entrepreneurship beyond social mission and financial revenues.
We are looking forward to the measures developed here being used in cross-sectional as well as longitudinal studies on the processes and outcomes of social entrepreneurship. This will eventually allow us to compare different manifestations of social entrepreneurs, such as for example the 'Ashoka' social innovator, the 'Yunus' social business founder, or the 'EMES' social enterprise leader (Hockerts, 2017). Finally, it is another limitation of our study that we have only considered respondents who in one way or the other are already identified as somewhat involved in social entrepreneurship. With this in mind we call for future researchers to replicate the work done here with either a pure for-profit business sample as well as a pure charity sample. Such work would also allow inferences about how widespread OSE behavior is in the population in general.
We anticipate that the OSE scale will be used by researchers to study when and why OSE behavior occurs in organizations. This will allow social entrepreneurship research to transcend the current focus on organizational forms. As we have argued earlier that we assume OSE behavior to be found in different organizational contexts. We encourage scholars to study both the incidents of OSE as well as its antecedents and performance outcomes.

Social Mission
Everything is subordinate to social mission 0.490 0.593 We will not change our social mission 0.858 0.787

Changemaking
It is an explicit part of our strategy to create change among our peers. 0.799 0.774 We actively promote change in organizations operating in our sector. 0.811 0.764 Through our work, we intend to change the practice in other organizations. 0.826 0.631 Our approach will change the way our sector works. 0.687 0.716

Earned Income
Our organization produces and sells goods or services. 0.756 0.761 We engage in commercial activities to support our social mission. 0.851 0.574 We are generating income from selling products and services. 0.881 0.876 Our revenue relies on earned income from market activities. 0.848 0.649

Paid Employees
All people working in our organization are remunerated by ordinary employment.
0.831 0.812 All people working in our organization receive market rate salaries. 0.865 0.784

Democratic Decision Making
The stakeholders affected by our organization are part of its decisionmaking process.
0.632 0.713 The decision making power is equally divided among individuals 0.808 0.402 Our management involves all internal and external stakeholders in a democratic decision making process.

Social Impact
Our organization has a large scale social impact 0.784 0.655 Our organization has more social impact than other social enterprise in country 0.893 0.821 Our organization has more social impact than other social enterprise globally 0.876 0.822 Table 5 Loadings and weights of indicators on the organizational social entrepreneurship construct for Sample 1

Social Mission
Everything is subordinate to social mission 0.698 0.082 We will not change our social mission 0.526 0.062

Change making
It is an explicit part of our strategy to create change among our peers. 0.652 0.081 We actively promote change in organizations operating in our sector. 0.704 0.088 Through our work, we intend to change the practice in other organizations. 0.639 0.081 Our approach will change the way our sector works. 0.607 0.077

Earned Income
Our organization produces and sells goods or services. 0.537 0.064 We engage in commercial activities to support our social mission. 0.713 0.083 We are generating income from selling products and services. 0.597 0.069 Our revenue relies on earned income from market activities. 0.592 0.070

Paid Employees
All people working in our organization are remunerated by ordinary employment. 0.454 0.059 All people working in our organization receive market rate salaries. 0.407 0.051

Democratic Decision Making
The stakeholders affected by our organization are part of its decision-making process. 0.686 0.083 The decision making power is equally divided among individuals 0.544 0.066 Our management involves all internal and external stakeholders in a democratic decision making process.

Stakeholder Participation
Before we develop new ideas, activities, and services, we interact with many stakeholders in our social sector.
0.726 0.088 Many of our new activities have been developed in teamwork with our beneficiaries. 0.722 0.088 We use co-creation techniques in our product and service development. 0.802 0.095 We develop projects of our social enterprise in collaboration with our beneficiaries. 0.777 0.092 Table 6 Loadings and weights of indicators on the organizational social entrepreneurship construct Sample 2

Social Mission
Everything is subordinate to social mission 0.443 0.107 We will not change our social mission 0.181 0.052

Change Making
It is an explicit part of our strategy to create change among our peers. 0.499 0.118 We actively promote change in organizations operating in our sector. 0.596 0.140 Through our work, we intend to change the practice in other organizations. 0.543 0.132 Our approach will change the way our sector works. 0.467 0.125

Earned Income
Our organization produces and sells goods or services. 0.246 0.054 We engage in commercial activities to support our social mission. 0.313 0.079 We are generating income from selling products and services. 0.309 0.079 Our revenue relies on earned income from market activities. 0.245 0.073

Paid Employees
All people working in our organization are remunerated by ordinary employment. 0.214 0.048 All people working in our organization receive market rate salaries. 0.137 0.035

Democratic Decision Making
The stakeholders affected by our organization are part of its decision-making process. 0.572 0.134 The decision making power is equally divided among individuals 0.340 0.077 Our management involves all internal and external stakeholders in a democratic decision making process. 0.503 0.115 Stakeholders are represented in committees and work groups of our organization. 0.515 0.120

Stakeholder Participation
Before we develop new ideas, activities, and services, we interact with many stakeholders in our social sector.