Abstract
To fully understand legitimacy as a complex construct, it is necessary to capture both collective perceptions and individual judgments. Much of the empirical research on legitimacy has focused on measuring the collective perceptions of groups of evaluators or critical institutions. This research develops and validates a psychometric measure of individual perceptions of pragmatic, moral, and cognitive legitimacy. Across seven studies, we demonstrate consistent reliability and scale structure, as well as convergent, discriminant, nomological, and predictive validity. We further show the generalizability and robustness of the measure across a variety of organizations and industries. This measure will advance empirical research on legitimacy by enabling researchers to capture the perceptions of individual evaluators directly and permit the comparison of results across studies.
Introduction
Legitimacy has been recognized as a core construct in the strategic management literature (Suddaby et al., 2017) and as an essential gateway to the resources necessary for the survival and growth of organizations, particularly new ventures and entrepreneurial organizations (Aldrich and Fiol, 1994; Zimmerman and Zeitz, 2002). While legitimacy has been examined from a social psychological perspective (Tost, 2011), most researchers adopt an institutional theory approach that views legitimacy as “a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions” (Suchman, 1995: 574). Researchers have conceptualized legitimacy as a theoretically complex, multi-level construct that exists as an interaction between a somewhat fluid perception held by individuals and a more concrete and stable social fact (Bitektine and Haack, 2015; Cording et al., 2010; Suddaby et al., 2017).
Our view corresponds to the perspective of Suddaby et al. (2017) that legitimacy is most appropriately conceptualized as a property, a resource, or capacity held by an entity; as a process, an interactive mechanism through which an entity legitimates; and/or as a perception, a socio-cognitive evaluation or judgment of an entity’s appropriateness that occurs between the individual and the collective. Legitimacy judgments occur across multiple levels both simultaneously and recursively; individual perceptions of categorical norms, themselves socially constructed, provide shared cognitive templates with which agents make legitimacy judgments relative to certain structures, symbols, processes, and entities. In the aggregate, the summation of these individual level judgments as a set of socially constructed shared beliefs (Berger and Luckmann, 1967; Suchman, 1995) both reflects and reinforces the institutional order, (Bitektine, 2011; Hoefer and Green, 2016; Suddaby and Greenwood, 2005). This is in keeping with the broader perspective that individual perceptions and judgments form the microfoundations of the organizational-level and institutional-level constructs at the core of management theory (Felin et al., 2015; Felin and Foss, 2005; Hambrick, 2004).
Extant research has primarily focused on legitimacy as either a property or a process, which has resulted in the proliferation of studies using proxy variables to capture the relative degree of legitimacy an organization or industry may possess (Deephouse and Suchman, 2008). As with many organizational-level constructs, much of the empirical research on legitimacy has focused on measuring the collective perceptions of groups of evaluators or critical institutions, such as government regulators and the media (Bitektine, 2011; Deephouse, 1996; Deephouse and Suchman, 2008). This is not problematic per se, but it does highlight the fact that there is more than ample opportunity to research legitimacy as a perception to gain meaningful insights that may translate into valuable contributions to the field. To date, there have been only five attempts to measure individual legitimacy perceptions (Díez-Martín et al., 2013; Elsbach, 1994; Foreman and Whetten, 2002; Humphreys and Latour, 2013; Pollack et al., 2012), and each has developed an idiosyncratic measure unique to a specific research context. The absence of a generalizable measure from the extant literature provides a disincentive to research on legitimacy as a perception. Furthermore, the lack of a useful measure may also lead to flawed research and inaccurate conclusions regarding the antecedents and consequences of legitimacy perceptions or its dimensions. To fully understand legitimacy as a complex construct, it is necessary to capture individual judgments as well as collective perceptions (Bitektine and Haack, 2015; Tost, 2011), which requires a measure that possesses a high degree of validity and can be easily adapted for research in a number of contexts.
The contribution of this article is to develop and validate a psychometric measure of individual perceptions of pragmatic, moral, and cognitive legitimacy (Suchman, 1995). The measure is developed by compiling items from the five previous attempts to measure individual legitimacy perceptions as well as items developed by representatives of the general population of consumers. Four studies purify the scale across a variety of organizations and industries using both student samples and participants from Amazon’s Mechanical Turk survey pool. Three additional studies validate the scale and demonstrate consistent reliability and scale structure, as well as convergent, discriminant, nomological, and predictive validity. The final scale is a three-dimensional, first-order reflective dimension set that can be used to measure these three dimensions of organizational legitimacy either individually or simultaneously.
A valid, generalizable measure of individual legitimacy perceptions will be valuable for use in survey and quasi-experimental studies, as well as a potential dependent measure in experimental studies, which have been suggested as necessary methodological tools to further the development of this stream of legitimacy research (Suddaby et al., 2017). Furthermore, a perceptual measure will promote better exploration of the process by which individual perceptions of legitimacy might ladder up to broader collective assessments of legitimacy as a property of an organization or field. For example, such a measure may enable researchers to examine how collective perceptions influence the formation of individual judgments and reference points, and how these individual perceptions then either reinforce or change the collective view. Finally, having a generally accepted instrument permits the comparison of results across studies, eventually leading to a critical mass of empirical results from which more generalizable inferences can be drawn.
Conceptual background
While researchers have agreed upon a conceptual definition of legitimacy, it has been conceptualized with different and potentially unrelated dimensions within the construct domain (Cording et al., 2010). Bitektine (2011) identifies 18 different types of legitimacy that have been proposed and examined in different contexts. The lack of consistent terminology and agreement on the dimensionality of the construct make the development of a valid measure particularly challenging (Hill et al., 2012). Thus, it is necessary to first determine which dimensions of legitimacy to capture within the development of an individual-level perceptual measure of legitimacy.
In reviewing this literature, a common theme emerges: nearly all of this research has distinguished between an active form of legitimacy evaluation or judgment—one that requires deliberate attention to compliance with either norms or laws—and a more passive, subconscious legitimacy evaluation corresponding to the “taken-for-grantedness” of the object under evaluation. For example, Aldrich and Fiol (1994) distinguish between the active judgment of sociopolitical legitimacy, which “refers to the process by which key stakeholders, the general public, key opinion leaders, or government officials accept a venture as appropriate and right, given existing norms and laws,” (p. 648) and the passive assessment of cognitive legitimacy, which they discuss as the taken-for-grantedness and familiarity that ventures or activities achieve as they become institutionalized over time. More recently, Zimmerman and Zeitz (2002) divide the active sociopolitical legitimacy judgment into two components: regulatory legitimacy, derived from rules and regulations created by governing bodies and other influential organizations, and normative legitimacy, derived from the norms and values of society.
Distinguishing between active and passive dimensions of legitimacy parallels Suchman’s (1995) conceptualization of legitimacy as a three-dimensional construct with pragmatic, moral, and cognitive dimensions. The first two dimensions involve the active assessment of both a focal organization’s expected value (pragmatic legitimacy) and compliance with social norms, ethics, and values (moral legitimacy). The third dimension (cognitive legitimacy) involves a passive assessment of the organization’s taken-for-grantedness and comprehensibility. While these dimensions by no means exhaust all possible types of legitimacy, they capture the majority of the conceptual domain of the construct and represent the most frequently measured dimensions of legitimacy (Díez-Martín et al., 2013; Elsbach, 1994; Foreman and Whetten, 2002; Humphreys and Latour, 2013; Pollack et al., 2012). Based on our review and following the well-accepted typology proposed by Suchman (1995), we therefore suggest that legitimacy as an individual perception is best conceptualized and measured as a three-dimensional construct. We define the three dimensions as follows.
Pragmatic legitimacy is an active assessment of the expected value that an organization will provide its primary stakeholders. This type of legitimacy “rests on the self-interested calculations of an organization’s most immediate audiences” (Suchman, 1995: 578) and is considered the most tangible of the three types. Dart (2004) describes pragmatic legitimacy as a judgment that if stakeholders receive anything of value from the organization, either directly or indirectly, then it is considered pragmatically legitimate.
Moral legitimacy is an active assessment of the degree to which an organization adheres to social norms and shared values in a manner that promotes societal welfare. According to Suchman (1995), “moral legitimacy reflects a positive normative evaluation of the organization and its activities” (p. 579). Moral legitimacy differs from pragmatic legitimacy in that it is based on an evaluation of the propriety of the organization’s activities relative to external norms, rather than an assessment of the benefits received by specific stakeholders (Dart, 2004).
Cognitive legitimacy is a passive assumption that an organization is simultaneously comprehensible and necessary or taken-for-granted. Consistent with prior conceptualizations (Suchman, 1995; Tost, 2011), we define cognitive legitimacy as a passive evaluation, distinct from the more effortful, elaborative judgments required to assess pragmatic and moral legitimacy. Within a shared cultural-cognitive schema, cognitive legitimacy represents the extent to which an organization’s activities can be “classified, understood, and integrated” by the observer (Humphreys and Latour, 2013: 774).
These different dimensions of legitimacy may be of varying relevance to researchers depending on several non-mutually exclusive factors, including the evaluating audience, the features of the organization being evaluated, the type of cognitive processing the evaluator undergoes in their legitimacy judgments, the degree of benefit the evaluator associates with the organization, and the mechanism of compliance (Bitektine, 2011). This is further complicated for researchers who wish to study the legitimation process of new ventures, as the relevant audience may change as the venture evolves and must cross various thresholds of legitimacy (Fisher et al., 2016). To be useful for researchers, a perceptual measure of legitimacy must therefore be able to capture one or multiple dimensions, depending on the needs of the researcher and the appropriateness to the focal organization.
Previous measures of legitimacy
In keeping with the emphasis on collective evaluations of legitimacy, there have been few attempts to empirically measure legitimacy as an individual perception. Prior empirical research examining legitimacy has relied heavily on the use of proxy variables to capture the relative degree of legitimacy an organization or industry may possess (Deephouse and Suchman, 2008). Early empirical studies of legitimacy focused on the evaluations of critical institutions, using proxy variables such as the approval of regulatory agencies, ratings of accrediting organizations, public endorsements, and media coverage (Deeds et al., 2004; Deephouse, 1996; Deephouse and Carter, 2005; Westphal et al., 1997). Legitimacy has also been examined indirectly, through analysis of legitimacy signals used by the firm to communicate with external stakeholders (Deeds et al., 2004; Khoury et al., 2013; Pollock and Gulati, 2007; Tornikoski and Newbert, 2007), legitimating activities or behaviors engaged in by the firm (Laïfi and Josserand, 2016; Low and Johnston, 2008; Schembera and Scherer, 2017; Tornikoski and Newbert, 2007; Zamparini and Lurati, 2017) or legitimating or delegitimating events in the organization’s history (Rutherford and Buller, 2007; Sinha et al., 2015). Researchers have also inferred legitimacy from key informant interviews with internal stakeholders (Hudson and Okhuysen, 2009; Human and Provan, 2000) or from secondary measures such as economic performance and organizational survival rates (Delmar and Shane, 2004; Kennedy and Fiss, 2009; Kuilman and Wezel, 2013; Staw and Epstein, 2000).
Our review of the extant empirical research reveals only five studies attempting to directly measure perceived legitimacy. The earliest example was a study conducted by Elsbach (1994) of the legitimacy of the California cattle industry. This study was conducted prior to the typology advanced by Suchman (1995) and therefore treated legitimacy as a unidimensional construct, labeled “normative” legitimacy. The measure consists of 12 items designed to assess normative perceptions of legitimacy on a 7-point Likert scale. Items included statements such as “The general public approves of the organizations’ operating procedures,” “The organization is viewed by business writers as one of the top firms in the cattle industry,” and “The organization has one of the lowest rates of employee turnover in the cattle industry” (Elsbach, 1994: 87). Foreman and Whetten (2002) provide the second instance of legitimacy measurement, measuring members of rural cooperatives’ perceptions of the pragmatic and cognitive legitimacy of their field. A total of 13 items were developed to capture perceptions of field-level legitimacy, such as “In general, I believe that co-ops are the lifeblood of the rural community” (cognitive legitimacy), “In general, I believe that co-ops have staff and managers with superior professional skills” (pragmatic legitimacy), and “In general, I believe that co-ops have outlived their usefulness” (cognitive legitimacy, reverse-coded).
Pollack et al. (2012) develop a five-item measure of cognitive legitimacy in order to examine the mediating effect of cognitive legitimacy between entrepreneurial preparedness in new venture pitches and funding outcomes. After one round of preliminary refinement studies, the final number of items was reduced to three: “I envision this business receiving high-profile endorsements in the future,” “I envision this business receiving favorable press coverage in the future,” and “I envision this business having a top management team that will benefit the organization.” To facilitate the measurement of legitimacy in a survey context, Díez-Martín et al. (2013) developed a 21-item measure of perceived pragmatic, moral, and cognitive legitimacy from a variety of perspectives filtered through the survey respondent’s perception, for example, “You (LI01), your colleagues (LI04), your bosses (LI07): believe that all your organization’s actions help it achieve its goals,” “The community (LE01), clients (LE02), allies (LE03), and government (LE04): consider that the organization’s actions are beneficial to them.”
Finally, Humphreys and Latour (2013) provide the most recent study measuring legitimacy. In this study, the authors examine the effects of media framing on consumer perceptions of the online gambling industry. The authors adopt the measure originally developed by Elsbach (1994) for normative legitimacy. In addition, this study measures cognitive legitimacy in a unique fashion. Rather than a psychometric instrument, the authors infer cognitive legitimacy from response time latencies in an implicit association test. The logic for this novel measurement technique is that cognitive legitimacy involves classifying an object into a preexisting cognitive category; therefore, individuals perceiving an organization as possessing a higher degree of cognitive legitimacy are able to categorize it more quickly, thus exhibiting faster response times (Humphreys and Latour, 2013).
Table 1 presents a comparison of these previous efforts to measure legitimacy perceptions. It is noteworthy that while there seems to be consensus that researchers are seeking to measure Suchman’s (1995) dimensions of pragmatic, moral, and cognitive legitimacy, there have been no efforts to develop a valid, generalizable measure that can be used across contexts. With the exception of Humphreys and Latour’s (2013) use of the Elsbach (1994) scale, each study developed its own idiosyncratic measure that is specific to the field, industry, or context in which the study was being conducted. While this enables a level of precision that is valuable for the in-depth study of a phenomenon, such measures limit the ability of researchers to generalize and compare findings across contexts (Cording et al., 2010). To be useful to researchers, a measure of legitimacy perceptions should be able to be easily adapted to a specific study as well as applied across contexts to enable the synthesis of accumulated knowledge (Campbell and Fiske, 1959; Churchill, 1979) and facilitate convergence in the discussion of this complex construct (Hill et al., 2012). The construction of such a measure requires a rigorous validation process across a variety of organizations, contexts, and samples. While each of the existing studies that sought to develop a new measure used some form of factor analysis to test the scale structure, none of these measures have been through such a validation process, and none used more than one sample in order to validate the measure. Thus, the existing measures fail to address the difficulties associated with measuring a complex construct such as legitimacy.
Comparison of previous perceived legitimacy measures.
This research will therefore contribute to the broader literature on legitimacy by developing a generalizable measure of all three dimensions of legitimacy that can be used to capture an individual perception of a focal organization. The first two dimensions involve the active assessment of both a focal organization’s expected value (pragmatic legitimacy) and its compliance with social norms, ethics, and laws/regulations (moral legitimacy). The third dimension involves a passive assessment of the organization’s taken-for-grantedness and comprehensibility (cognitive legitimacy). To be useful to future researchers, these dimensions can be measured separately or simultaneously, depending on the research context and goals. The development of a three-dimensional, valid, generalizable scale that can be used across contexts will better enable the comparison of results across studies and the testing of more general theories of legitimacy.
Scale development
To develop and validate a measure of the three dimensions of perceived legitimacy, we followed the methodological procedures suggested by Nunnally (1978), Churchill (1979), and Anderson and Gerbing (1988). To begin, we compiled the items from the five previous attempts to measure the three dimensions of legitimacy and assessed them to determine which items could be revised to apply to organizations in general, as opposed to the specific contexts in which they had been developed. These items were revised as needed and used as the initial basis for the pool of scale items. As our goal is to develop a measure that can capture the legitimacy perceptions of individuals, and not critical institutions, we then expanded this pool using representatives of the general population of consumers.
We first consulted seven graduate students who were from different demographic backgrounds, did not have any particular expertise in legitimacy research, and were familiar with the process of scale development in general. Each participant was provided detailed conceptual definitions of the three dimensions of legitimacy and asked to generate a list of as many items as possible representing each of the three dimensions. The seven lists generated a total of 105 scale items representing the three underlying dimensions. Highly similar items created by more than one participant were considered to be duplications and combined into a single item for the next phase of the scale development process.
Next, we provided three of the graduate students with the complete list of items they and their colleagues had generated, as well as the items compiled from the existing measures in the literature, and asked them to conduct a sorting task matching each item with its corresponding dimension. The goal of this was to determine whether all three participants would view each item as measuring the same underlying dimension as the individual who had originally generated the item. If two or more participants in this second stage matched an item to a different dimension than had been originally intended by the item’s author, the item was eliminated. Items that were matched to two or more different dimensions by the participants were also eliminated. This resulted in a pool of 60 items on which the participants agreed (19 pragmatic, 23 moral, and 18 cognitive).
This reduced item pool was then evaluated by three faculty members selected for their familiarity with and knowledge of the legitimacy literature. All three faculty members were also familiar with the scale development process and had experience developing scales. Each faculty member was consulted individually, provided with the conceptual definitions of the three dimensions of legitimacy and the pool of items, and asked to revise and refine the items for content validity. Their feedback led to the generation, deletion, and revision of items, resulting in a final pool of 44 items (12 pragmatic, 16 moral, and 16 cognitive; see Appendix 1). We then conducted four studies to refine and purify the scale items and three studies to validate the scale.
Scale purification
The scale purification studies followed a common procedure. Individual participants were first provided a brief description of an organization, selected to represent a variety of organizational forms, industries, sizes, and ages. Descriptions ranged from 400 to 700 words and were carefully designed to provide only basic information regarding the organization’s products and services. This was done to ensure at least a minimal degree of familiarity with the organizations, but not to provide enough information to potentially bias any responses. This would force participants to rely primarily on their generalized perceptions, assumptions, and expectations corresponding to the dominant institutional logic of the focal organization’s field. Following the descriptions, the scale items were presented in a random order. Participants were asked to evaluate each scale item for the focal organization and responded to each item on a 7-point Likert-style scale with endpoints strongly disagree/strongly agree. A between-subjects design was used in which each participant evaluated one organization. The data were analyzed both at the level of the individual organization and pooled across organizations. For brevity, only the pooled results are discussed.
One of the challenges in capturing individual, and not institutional, legitimacy perceptions is in determining the proper sample. Studies of individual and consumer behavior have historically relied on student samples, which have raised concerns that the homogeneity of the sample might influence the results (Lynch, 1982). Indeed, a second-order meta-analysis found that the responses of student samples are relatively more homogeneous than those of nonstudent samples, and that effect sizes can differ across student and nonstudent samples (Peterson, 2001). Researchers have argued that relatively homogeneous samples such as student samples are desirable for theory development and testing, as their use minimizes extraneous variability that might unduly influence the results (Calder et al., 1981; Calder and Tybout, 1999). However, concerns about validity and generalizability still lead researchers to recommend the use of both student and nonstudent samples (Peterson, 2001).
The second common source of participants for individual and consumer research has become Amazon’s Mechanical Turk survey pool. Participants in this pool have been found to be more demographically and geographically diverse (Buhrmester et al., 2011; Mason and Suri, 2011) and more representative of the general population (Paolacci et al., 2010) than typical American student samples. While concerns have been raised about the misrepresentation of identity among MTurk participants (Wessling et al., 2017), on the whole data collected from this source has been found to be of high quality (Buhrmester et al., 2011; Goodman and Paolacci, 2017). Given the benefits of and concerns about both of these sources, we have taken a conservative approach and used a combination of student samples and Amazon MTurk panel members in the scale purification and validation process.
Studies 1–3: scale purification
Study 1 was designed to assess the reliability and initial structure of the scale. The 44-item scale was administered to 364 undergraduate students at a large Midwestern university in exchange for extra credit in a business course. Participants were provided descriptions of three relatively large, well-known organizations (Walmart, the Cleveland Clinic, and FanDuel). These organizations were selected based on informal a priori expectations that they should exhibit variance on the three proposed dimensions. For instance, we expected that due to the relative newness of the daily fantasy sports industry and the organization itself, FanDuel should exhibit lower composite scores for cognitive legitimacy than either WalMart or the Cleveland Clinic. Similarly, we expected that the Cleveland Clinic’s nonprofit status would lead to higher composite scores for moral legitimacy than either WalMart or FanDuel.
Principal axis factoring with oblimin rotation initially resulted in a six-factor solution, based on a combination of eigenvalues and a scree plot. Upon further examination, several items from the four previous attempts to measure legitimacy and from the item generation process were found to be written in the negative, intended to be reverse coded. Since the scale is intended to measure legitimacy, and not illegitimacy, these items were removed. In addition, a number of items, particularly those retained from the existing scales, were identified as referencing field-level legitimacy rather than organization-specific legitimacy. To be consistent with the intended contributions of the proposed measure, these items were revised for subsequent analysis (see Appendix 2). Items were then removed based on factor loadings. Stevens (1992) recommends a minimum cutoff for factor loadings of .4; however, Tabachnick and Fidell (2007) suggest retaining items with factor loadings of at least .3 when the items have different frequency distributions. At this early stage, the risk of prematurely eliminating items outweighed that of retaining too many items, so we opted to retain items with factor loadings of .3 or greater. This resulted in a three-factor solution that reflected the underlying pragmatic, moral, and cognitive legitimacy dimensions. However, a number of items had high cross-loadings, and several items that had been intended to measure cognitive legitimacy loaded onto either the pragmatic or moral legitimacy factors. All positively worded items were retained in the second purification study as well as the revised items that were measuring field-level legitimacy.
Study 2 followed the same procedure as Study 1, using the revised scale items and different organizations as stimuli. To control for extraneous variance and industry-level effects, three organizations were selected from a single industry of grocery store chains (Whole Foods Market, Giant Eagle, and Save-a-Lot). Following the same procedure as the previous study, the organizational descriptions were provided to a sample of 471 undergraduate students. Principal axis factoring with oblimin rotation yielded a three-factor solution, based on a scree plot and eigenvalues greater than 1, that explained 50% of the total variance and reflected the three predicted underlying dimensions. Items with factor loadings greater than .3 and cross loadings less than .3 were again retained (Tabachnick and Fidell, 2007). While the statistical analyses aided the selection and retention of items from one study to the next, care was taken to err on the side of retaining too many items if the decision was unclear. The resulting three factors demonstrated acceptable levels of reliability for the six-item pragmatic (α = .86), nine-item moral (α = .92), and six-item cognitive legitimacy (α = .79) dimensions. The factor analysis results of Study 2 are summarized in Table 2. Further examination of the cognitive legitimacy dimension suggested that the majority of retained items were primarily capturing the necessity component of the definition of cognitive legitimacy, and not the taken-for-grantedness component. A new item was therefore added in Study 3 that was intended to capture perceptions of taken-for-grantedness.
Preliminary factor analysis results: Study 2.
Study 3 followed the same procedure as Studies 1 and 2 to test the refined set of items using a single organization, the Cleveland Clinic, and a more diverse participant pool. Rather than a student sample, 225 participants representative of the general population were recruited from Amazon’s MTurk survey engine. Analysis of demographic information revealed a representative sample of the general public drawn from 46 states in the United States. Participant ages ranged from 21 to 76 years with an average age of 40 years. Reported income primarily fell into the ranges of US$25,000–US$49,999 (37%), less than US$25,000 (28%), or US$50,000–US$74,999 (17%). Participants were provided with the same description of the Cleveland Clinic used in Study 1 and responded to the reduced set of items from Studies 1 and 2. Participants were also asked to report how familiar they were with the organization prior to the study on a seven-item scale with endpoints not at all familiar/very familiar. The majority of participants (78.7%) reported being at least somewhat familiar with the Cleveland Clinic, with a mean familiarity score of 4.55 and a standard deviation of 2.39.
Principal axis factoring with oblimin rotation was again conducted and items were eliminated systematically as before for low factor loadings or high cross loadings. The resulting 12-item scale produced a three-factor solution, based on a scree plot and eigenvalues greater than 1, explaining 80.2% of the total variance and reflecting the three predicted underlying dimensions. All factor loadings were greater than .5 and no item demonstrated a higher cross loading than .2. All three factors again demonstrated acceptable levels of reliability for the three-item pragmatic (α = .93), six-item moral (α = .94), and three-item cognitive legitimacy (α = .76) dimensions. Three of the revised items from the original Elsbach (1994) scale were retained in the final 12-item scale.
Study 4: confirmation of scale structure
The purpose of Study 4 was to confirm the final structure of the proposed legitimacy scale and ensure that the three factors were capturing separate, distinct dimensions of legitimacy. Following the same procedure used in Studies 1–3, 159 participants recruited from Amazon MTurk were presented with the description of FanDuel used in Study 1 and responded to the 12-item, three-dimensional scale. We selected FanDuel as the focal organization because at the time of the study, the organization was at the center of a small controversy and had received significant media coverage that presented both positive and negative aspects of their business practices. As a result, we expected participants to have similar levels of familiarity with the organization, but to exhibit variance in their perceptions of its legitimacy. We further expected this organization to score particularly low on cognitive legitimacy due to the newness of both the organization and the industry, as well as its varying legal status across states. This enabled us to test the ability of the scale to capture judgments of organizations that have questionable legitimacy. Participants were again asked to report how familiar they were with the organization prior to the study on a seven-item scale with endpoints not at all familiar/very familiar. The majority of participants (64.2%) reported being at least somewhat familiar with FanDuel, with a mean familiarity score of 3.23 and a standard deviation of 2.08. Examination of demographic data showed a sample drawn from 49 US states, with participants ranging from 21 to 65 years of age with an average age of 39 years and a representative range of income demographics: less than US$25,000 (25%), US$25,000–US$49,999 (34%), US$50,000–US$74,999 (25%), US$75,000–US$99,999 (10%), over US$100,000 (6%).
Principal axis factoring with oblimin rotation of the 12-item scale produced a three-factor solution, based on a scree plot and eigenvalues greater than 1, explaining 72.8% of the total variance and reflecting the three predicted underlying dimensions. All factor loadings were greater than .65 and no item demonstrated a higher cross loading than .2. All three factors again demonstrated acceptable levels of reliability for the three-item pragmatic (α = .89), six-item moral (α = .93), and three-item cognitive legitimacy (α = .89) dimensions. To address possible concerns about the ratings of those participants who were unfamiliar with the organization, the analysis was re-run excluding those participants who answered “not at all familiar.” The results revealed the same three-factor solution explaining 71.8% of the variance and with the same pattern of factor loadings.
A confirmatory factor analysis was conducted to test the three-dimensional structure of the scale using Stata 14.1. Overall goodness-of-fit statistics suggest that our data fit the model well (χ2(50) = 96.82, p < .001; root mean square error of approximation (RMSEA) = .08; confirmatory fit index (CFI) = .97; and standardized root-mean-square residual (SRMR) = .05). Values of up to .08 for RMSEA and SRMR and greater than .90 for CFI are considered acceptable (Bagozzi and Yi, 1988; Hu and Bentler, 1999). The three factors were correlated with each other at r = .24 for pragmatic and moral legitimacy, r = .08 for pragmatic and cognitive legitimacy, and r = .65 for moral and cognitive legitimacy. The factor analysis results of Study 4 are summarized in Table 3.
Confirmatory factor analysis results: Study 4.
To further assess the appropriateness of the three underlying dimensions, we compared the goodness of fit of the proposed three-factor model to a single-factor model and a series of two-factor models representing all possible combinations of the three dimensions. Goodness-of-fit statistics for all five models are presented in Table 4. The three-factor model demonstrated a significant improvement in overall fit over the single-factor model (
Comparison of three-factor, two-factor, and one-factor models: Study 4.
df: degrees of freedom; RMSEA: root mean square error of approximation; CFI: Confirmatory Fit Index; SRMR: standardized root-mean-square residual.
We further computed the average variance extracted for each composite factor, all of which exceeded the recommended cutoff of .5 (Fornell and Larcker, 1981). The results are presented in Table 5. Discriminant validity between the three dimensions of the scale was assessed according to the well-accepted criterion proposed by Fornell and Larcker (1981), whereby the average variance extracted for each factor should exceed the squared correlation between the focal construct and each of the other latent constructs. Results of the discriminant validity analysis are presented in Table 5, with the square-root of the average variance extracted presented on the diagonal of the correlation matrix. As these results suggest, each factor explains more unique variance than shared variance with any of the other factors, providing evidence of discriminant validity between the three dimensions in the scale.
Composite descriptive statistics and discriminant validity: Study 4.
SD: standard deviation; AVE: average variance extracted.
Square root of AVE for each factor displayed on the diagonal of the correlation matrix.
Scale validation
Study 5: comparison to previous measures
Study 5 compares our measure of the dimensions of perceived legitimacy to previous measures of legitimacy. If this is a valid measure, then it should be significantly correlated with previous measures of the same legitimacy dimensions. However, to be an improvement over these previous measures, the new scale should also demonstrate better psychometric properties than the earlier measures. Study 5 tests this by comparing the new scale to two previous measures of legitimacy dimensions: Elsbach’s (1994) measure of normative legitimacy and Foreman and Whetten’s (2002) measure of pragmatic and cognitive legitimacy. While Elsbach’s scale predates Suchman’s (1995) three-dimensional conceptualization, the items suggest that it most accurately maps onto moral legitimacy.
Following the same procedure as in the previous studies, we presented 118 undergraduate students at a Midwestern university with a description of SoFi, a new online personal finance company that specializes in consolidating and refinancing student loan debt. The description was adapted from the organization’s Wikipedia page and website. This organization was selected because of its high relevance to the focal population and the fact that the newness of both the organization and the online personal finance industry would be likely to create variance in perceptions of its legitimacy. Participants then responded to the 12-item, three-dimensional scale along with adaptations of Elsbach’s (1994) 12-item measure of normative legitimacy and Foreman and Whetten’s (2002) 7-item measure of pragmatic legitimacy and 6-item measure of cognitive legitimacy. As both of these previous measures were developed for specific industries (cattle and farming co-ops), the items were adapted to be appropriate to this context (e.g. “the cattle industry” became “the financial services industry,” and “small/large farmers” became “small/large borrowers.”). In addition, it was noted in Study 4 that one of the items in the pragmatic dimension that had been adapted from an earlier measure retained the word “shareholders” rather than “stakeholders.” This phrasing was changed and retested, and it was not found to affect the loading of this item on its relevant factor.
As expected, composite scores on the pragmatic dimension of the scale were significantly correlated with the pragmatic dimension of Foreman and Whetten’s (2002) scale (r = .27, p < .005), composite scores on the moral dimension of the scale were significantly correlated with Elsbach’s (1994) scale (r = .63, p < .001), and composite scores on the cognitive dimension of the scale were significantly correlated with the cognitive dimension of Foreman and Whetten’s (2002) scale (r = .24, p < .01). All three dimensions of the scale were also significantly correlated with each other (pragmatic and moral r = .54, p < .001; pragmatic and cognitive r = .36, p < .001; and moral and cognitive r = .53, p < .001). This supports both the convergent validity of the scale and its conceptualization as three correlated dimensions. The pragmatic dimension of the scale was also found to have a comparable reliability (α = .86) to the pragmatic dimension of Foreman and Whetten’s (2002) measure (α = .82), and the moral dimension of the scale was found to have a comparable reliability (α = .89) to Elsbach’s (1994) measure (α = .87). The cognitive dimension of the scale had a slightly lower reliability (α = .72) than the cognitive dimension of Foreman and Whetten’s (2002) measure (α = .77). However, this may be in part due to the sensitivity of Cronbach’s alpha to the number of items in a scale.
To test the structural properties of the three scales, exploratory and confirmatory factor analyses were run on each of the three scales separately. To be consistent with the original conceptualizations, the new scale should load onto three factors, Foreman and Whetten’s (2002) measure should load onto two factors, and Elsbach’s (1994) measure should load onto one factor. Principal axis factoring with oblimin rotation revealed that the new scale loaded onto three factors explaining 62% of the variance. Foreman and Whetten’s (2002) scale loaded onto four factors explaining 52% of the variance, and Elsbach’s (1994) scale loaded onto two factors explaining 47% of the variance. Confirmatory factor analyses were then run on all three scales using Stata 14.1 and the original theorized factor structure for each scale. Our scale was found to have better overall goodness of fit (
Study 6: nomological and discriminant validity
Study 6 was conducted to test the nomological and discriminant validity of the new scale in two ways. First, if the scale is valid, it should be related to, but distinct from, theoretically related constructs. Deephouse and Suchman (2008) argue that legitimacy is highly related to, but a distinct construct from, organizational reputation and trust in the organization. Aldrich and Fiol (1994) suggest that legitimacy can act as an antecedent or prerequisite for the development of trust in interorganizational relationships. Zimmerman and Zeitz (2002) further suggest that perceptions of legitimacy can lead external stakeholders to feel that a venture is generally competent, effective, and worthy, suggesting that legitimacy perceptions may also be related to overall attitudes toward the organization. Thus, if this scale is truly capturing perceptions of legitimacy, responses on the scale should be correlated with, but distinct from, responses to measures of organizational reputation, trust in the organization, and attitude toward the organization. Distinction from attitude toward the organization may also rule out the possibility that the scale is capturing a halo effect in which organizations that are more well-liked are rated as overall more legitimate.
To test this, we measured organizational reputation, trust in the organization, and attitude toward the organization in addition to our scale. Organizational reputation was measured using the 15-item corporate reputation scale developed by Walsh et al. (2009). Trust in the organization was measured using an adaptation of the three-item trust in the brand scale developed by Sheinin et al. (2011). Attitude toward the organization was measured using a three-item measure with endpoints favorable/unfavorable, good/bad, and positive/negative. As theoretically related, yet distinct constructs, we expect the composite scores for each dimension of legitimacy to be significantly correlated with reputation, trust, and attitude toward the organization, yet not so highly correlated as to suggest the measures are capturing the same underlying construct.
Second, if the scale is valid, then it should be able to capture differences in legitimacy perceptions across organizations that are generally viewed as more or less legitimate by the public. Participants were asked to evaluate one of three organizations that were from the same industry but expected a priori to vary in their levels of pragmatic, moral, and cognitive legitimacy. Comparisons of the mean scores on the three legitimacy factors for the three organizations should therefore provide evidence of the content validity of the scale items. We chose to hold the industry constant to control for industry level effects and examine the unique variance on each of the legitimacy dimensions. The airline industry was chosen as the focal industry because most individuals are familiar with airlines, the industry itself is considered legitimate, and according to the Reputation Institute’s 2015 customer satisfaction rankings, many airlines demonstrate fairly modest customer satisfaction ratings, therefore controlling for any potential contamination of our analyses which may be caused by attitudinal halo effects.
Three airlines were selected based on differences in several key attributes we hypothesized, which would result in different mean levels of legitimacy on each of the three dimensions. We chose American Airlines, a large, well-established airline, JetBlue Airways, a medium-sized airline with the highest customer satisfaction rating of the major airlines considered, and Baltia Airlines, which is still in the start-up phase. As the oldest and most established of the three airlines, we postulated that American Airlines should demonstrate higher pragmatic legitimacy than either JetBlue or Baltia (Aldrich and Fiol, 1994). However, given the reputation of the large airlines in this industry at the present time, we expected that they might score lower than the other two on moral legitimacy. Because JetBlue is relatively new compared to other airlines and positions itself with many consumer-friendly policies such as low fares, free inflight WiFi and television, and low baggage/cancellation fees, our a priori prediction was that JetBlue should be perceived as higher in moral legitimacy than the other two. Finally, while Baltia possesses some legitimacy in that it was founded in 1989 and is a registered business traded in OTC markets (ticker: BLTA), at the time the study was conducted it had no flights in operation. We therefore expected Baltia to exhibit substantially lower cognitive legitimacy as compared to either of the other two airlines.
Following the same procedure as in the previous studies, we sampled 396 participants from Amazon MTurk and randomly assigned each to one of the three airlines. Once again, demographic descriptive statistics revealed a representative sample of the general public, drawn from 49 US states. Participants ranged from 19 to 76 years of age with an average age of 37 years and represented a wide range of income demographics: less than US$25,000 (25%), US$25,000–US$49,999 (28%), US$50,000–US$74,999 (25%), US$75,000–US$99,999 (9%), and over US$100,000 (13%). Participants responded to the 12-item, three-dimensional scale, corporate reputation scale, trust in the organization scale, and attitude toward the organization scale for their assigned airline.
Table 6 presents descriptive statistics and reliability for each dimension of legitimacy, reputation, trust in the organization, and attitude toward the organization, and Table 7 presents a summary of the correlations between these variables. As expected, all three dimensions of legitimacy are significantly correlated (p < .001) with the composite measures of organizational reputation (pragmatic r = .40, moral r = .68, and cognitive r = .52), trust in the organization (pragmatic r = .40, moral r = .59, and cognitive r = .42), and attitude toward the organization (pragmatic r = .20, moral r = .68, and cognitive r = .39). As reputation is a multidimensional scale, correlations were also calculated for each of the five dimensions with the three dimensions of legitimacy, and all three dimensions of legitimacy were found to be significantly correlated (p < .001) with all five dimensions of reputation. Overall, the magnitude of these correlations suggests that the scales are measuring distinct but related constructs. It is also worth noting that the correlations among the three dimensions of legitimacy demonstrate a slightly different pattern across studies, suggesting that the relationships between the dimensions of legitimacy may vary across different organizations and industries. This further supports the conceptualization of these three aspects of legitimacy as distinct dimensions, as opposed to being reflective of a single higher order construct.
Descriptive statistics and Cronbach’s alpha of legitimacy dimensions and related constructs: Study 6.
SD: standard deviation.
Correlation coefficients of legitimacy dimensions and related constructs: Study 6.
p < .001.
We next compared the mean values for the three airlines on the three dimensions of legitimacy using a multivariate analysis of variance (MANOVA). The results for all four omnibus tests yield identical and significant p values (p < .001), so individual analyses of variance (ANOVAs) were performed for each of the three legitimacy dimensions. Figure 1 presents a summary of the means on all three dimensions. A one-way ANOVA on pragmatic legitimacy revealed a significant effect of organization (F(2, 394) = 36.27, p < .001). Planned contrasts revealed that as expected, the older and more well-established American was perceived to be more pragmatically legitimate than both JetBlue (American M = 5.44, JetBlue M = 5.10, F(1, 394) = 5.63, p < .05) and Baltia (American M = 5.44, Baltia M = 4.26, F(1, 394) = 68.36, p < .001). JetBlue was also perceived to be more pragmatically legitimate than Baltia (JetBlue M = 5.10, Baltia M = 4.26, F(1, 394) = 33.98, p < .001), which is not surprising given Baltia’s current start-up status and lack of flights.
Legitimacy means by airline: Study 6.
A one-way ANOVA on moral legitimacy also revealed a significant effect of organization (F(2, 395) = 47.77, p < .001). Planned contrasts revealed that as predicted, American was perceived to be less morally legitimate than both JetBlue (American M = 4.15, JetBlue M = 5.27, F(1, 395) = 64.59, p < .001) and Baltia (American M = 4.15, Baltia M = 5.37, F(1, 395) = 78.22, p < .001). JetBlue and Baltia were not significantly different in perceived moral legitimacy (p > .10). Finally, a one-way ANOVA on cognitive legitimacy also revealed a significant effect of organization (F(2, 394) = 9.76, p < .001). Planned contrasts revealed that as expected, Baltia was perceived to be less cognitively legitimate than both American (Baltia M = 3.91, American M = 4.61, F(1, 394) = 19.34, p < .001) and JetBlue (Baltia M = 3.91, JetBlue M = 4.30, F(1, 394) = 6.46, p = .01). American and JetBlue were not significantly different in perceived cognitive legitimacy (p > .05). Collectively, these results support the nomological validity of the new scale.
Study 7: predictive validity
Study 7 was conducted to test the predictive validity of the scale. Legitimacy is widely acknowledged to benefit organizations, particularly new ventures, by increasing their ability to attract resources (Aldrich and Fiol, 1994; Zimmerman and Zeitz, 2002). This includes the capability of persuading others to provide you with financial resources. For example, Tornikoski and Newbert (2007) find that new ventures that have established strategic legitimacy are also more likely to have made their first sale and/or received external financing. This may be due to the legitimation of the firm enabling it to engage in activities such as completing product development, communicating with customers, and initiating marketing and promotion efforts (Delmar and Shane, 2004). Perceived legitimacy also appears to have an effect on the likelihood of attracting funding from investors. Pollack et al. (2012) find that increased cognitive legitimacy perceptions of an entrepreneur pitching to investors significantly predicts the amount of funding that the entrepreneur receives. Signaling organizational legitimacy has also been shown to increase the amount of capital raised in an initial public offering (Deeds et al., 2004; Khoury et al., 2013), which in turn increases the firm’s ability to attract resources beyond their IPO, such as strategic alliances (Pollock and Gulati, 2007).
Higher perceptions of an organization’s legitimacy should therefore be associated with a greater willingness to provide the organization with financial resources, particularly when the individual is given the opportunity to directly contribute to the organization’s success. To test this predictive relationship, we designed an experiment set in the context of a hypothetical crowdfunding campaign. Since this scale is intended to measure the legitimacy perceptions of individual stakeholders, such as consumers, crowdfunding is an appropriate context in which to capture the likelihood of these individuals providing funding to a new venture. To further test the generalizability of the scale, we manipulated the legal form of the focal organization in a one-factor, three-level (for-profit, nonprofit, benefit corporation) between-subjects design. This enabled us to examine whether pragmatic, moral, and cognitive legitimacy perceptions would differ across different organizational types, as well as which of the dimensions of legitimacy would most strongly predict participants’ willingness to contribute resources to the organization.
Following the same procedure as in the previous studies, we sampled 301 participants from Amazon MTurk and randomly assigned each to one of the three legal form conditions. Once again, demographic descriptive statistics revealed a representative sample of the general public, drawn from all 50 US States. Participants ranged from 20 to 80 years of age (average age = 37 years) and represented a wide range of income demographics, with just over half the sample (52.4%) reporting an annual income between US$25,000 and US$75,000. Participants were randomly assigned to one of the three legal form conditions. To ensure a minimum degree of familiarity with their assigned legal form, at the outset of the study they were provided a very brief (100–170 words) description of the legal form. To ensure participants read and understood each legal form description, we included two attention check questions following the description.
After reading the background information on legal form, participants were instructed to imagine a scenario in which they worked a few extra hours the previous month and have an additional budget of US$100 to spend. Participants were then told, You are looking at a crowdfunding website specifically designed for entrepreneurs creating organizations with a social cause or purpose. As you browsed the projects listed on the website, you found a number of them particularly compelling. In the next frame, you’ll find an example of one of the descriptions listed for an organization that you’re considering contributing to.
Next, participants read a description of a fictitious microfinance organization. The same description was presented in all three conditions, with the only difference being the manipulated legal form mentioned at the beginning of the description. The organization description was developed using information from a registered benefit corporation’s “About Us” section of their website.
Participants were then asked whether or not they would be willing to contribute funds to this campaign. If they answer “yes,” they were asked how much they would be willing to contribute on a scale from US$0–US$100. Participants then responded to the 12-item, three-dimensional scale. The three legitimacy dimensions were each found to have acceptable reliability (pragmatic α = .91, moral α = .93, and cognitive α = .82) and confirmatory factor analysis again demonstrated strong overall model fit (
We first conducted a MANOVA where the independent variable was legal form and the dependent variables were specified as each of the three legitimacy dimensions. Omnibus multivariate test statistics showed a significant effect of legal form on legitimacy (Pillai’s Trace = 6.09, p < .001; Wilks’
We next conducted a logistic regression on willingness to contribute to the hypothetical crowdfunding campaign, where the dependent variable was coded as a 1 if the participant decided to contribute, otherwise 0. The results of our logistic regression analysis are presented in Table 8. Both cognitive legitimacy (β = .68, p < .001) and moral legitimacy (β = 1.06, p < .001) were found to significantly and positively predict willingness to contribute funds to the hypothetical crowdfunding campaign. Pragmatic legitimacy does not appear to have a significant impact on the likelihood an individual will contribute funds.
Logistic regression results of legitimacy dimensions and willingness to contribute: Study 7.
Pseudo R2 = .31; log likelihood = –143.73.
Finally, we conducted a linear regression on the amount the individual was willing to contribute using only those participants who had indicated a willingness to contribute to the hypothetical crowdfunding campaign. The results revealed no significant effects of perceived legitimacy on the amount the individual was willing to contribute (ps > .10). This may have been due to an anchoring effect, as the majority of participants chose to contribute US$50 (27.2%), US$100 (11.4%), US$40 (11.4%), or US$20 (11.4%). However, this may also indicate that perceptions of legitimacy play a stronger role in the initial decision to contribute resources to an organization than they do in the decision of how much to contribute to the organization.
Discussion
As a complex, multi-level construct, the measurement of legitimacy provides a challenge for researchers. While the widespread use of proxy variables and inferred measures is appropriate for the assessment of legitimacy as a property, and as an indicator of the assessment of critical institutions and key stakeholders, the measurement of individual perceptions of legitimacy requires a more direct approach. The current research makes a methodological contribution to the ongoing research stream examining perceptions of legitimacy by developing a valid, generalizable measure of individual perceptions of pragmatic, moral, and cognitive legitimacy. It is our hope that an improved perceptual measure will also provide future researchers with a useful tool to understand legitimacy as a process, better enabling a comprehensive understanding of this key success factor.
Across seven studies, we demonstrate that our measure has a consistent three-dimensional scale structure and exhibits convergent, discriminant, nomological, and predictive validity. We assess the generalizability and robustness of the measure by testing its validity using a diverse selection of organizations and industries. Our results show that the structure and reliability of our measure are consistent across organizations of different sizes, ages, and types. We further demonstrate that our measure can be used to compare the legitimacy of organizations across industries or to identify differences in legitimacy among competitors within a single industry. These results suggest that this measure is both general enough to compare across studies and can be adapted to be applied to a wide variety of organizations and future research contexts.
This scale is a first step in addressing the call for expanded methodological approaches to understanding individual legitimacy judgments and perceptions made by Suddaby et al. (2017). The ultimate goal of this research stream is to be able to map the process through which individuals arrive at legitimacy judgments, which is beyond the scope of this research. However, the development of a valid psychometric measure of individual legitimacy perceptions provides a necessary preliminary step in this process. This scale enables the comparison of current legitimacy perceptions across individuals, organizations, or contexts, can act as a dependent measure in experimental research designed to identify antecedents and influences on legitimacy perceptions or responses to specific legitimation strategies and activities, and provides an instrument to study changes in individual legitimacy perceptions over time. This can serve as a benchmark for comparison to the assessment of critical institutions, key stakeholders (such as potential investors), or the general public. In addition, while this measure was developed to assess the legitimacy of an organization, the items can easily be adapted to facilitate the assessment of the legitimacy of an industry or field.
The proposed measure may also be useful in future research seeking to explore how individual legitimacy perceptions might influence the process of legitimation, or how certain events, individuals, or groups might differentially affect the broader institutional environment. As noted previously, legitimacy occurs at multiple levels both simultaneously and recursively. Individual perceptions of legitimacy are based on internal references to external “social facts” that exist, objectively and distinctly, as a property of a focal object (i.e. an organization, policy, or behavior). At the same time, however, individual judgments influence the underlying institutional logics, rendering the collective process of legitimation somewhat fluid. Further complicating the process, certain individuals or critical stakeholders may have more or less influence. Future research may adopt the proposed measure to examine these more complex social processes in either direction. For example, researchers interested in the distinction between legality and moral legitimacy could examine the impact of policy changes on the legitimation of certain social and political norms by measuring individual perceptions of moral legitimacy before and after regulatory change. However, researchers could examine the indirect effect that moral legitimacy judgments have on broader perceptions of what might be considered on a collective level to be illegitimate activities, such as commercial activity in the informal economy (Webb et al., 2009).
The proposed instrument also provides researchers a means by which all three of the dimensions of legitimacy advanced by Suchman (1995) can be measured either simultaneously or individually. However, we consider it important to emphasize that our conceptual and empirical approach to the perceived legitimacy construct is as a reflective measure of the three individual dimensions. Across the studies reported here, we tested a number of different first- and second-order reflective and formative specifications, including the alternative formative construct specifications suggested by Edwards (2011). Our empirical analyses consistently revealed that the first-order, reflective dimension set as put forth in this research was the most appropriate structure. Therefore, we do not consider it appropriate to simply sum the three dimensions to arrive at a composite score, as such measures are certain to exhibit substantial variation between the dimensions across industries, organizations, and/or institutional contexts.
Instead, researchers would be well advised to treat the proposed measures as independent, reflective constructs, and to theorize and test the relationships between dimensions accordingly. For instance, substantial debate has surrounded the relationship between cognitive legitimacy and both pragmatic and moral legitimacy (Dart, 2004; Deephouse and Suchman, 2008; Scott, 2013). One could argue that perceptions of an organization as being pragmatically and morally legitimate could lead to the organization becoming taken for granted and, thus, cognitively legitimate (Laïfi and Josserand, 2016). However, it may also be necessary for an organization to be comprehensible in order to assess its pragmatic and moral legitimacy. One interesting avenue for future research that may be facilitated by this measure would be to examine cognitive legitimacy as either an antecedent or consequent of the other two dimensions.
This research does have several limitations. While the simultaneous measurement of all three dimensions provides opportunities that may enrich our theoretical understanding of their structural configuration, the nature of cognitive legitimacy suggests that a self-report scale may not be the ideal way to measure this passive judgment. Previous research has employed implicit means to measure cognitive legitimacy, such as the use of an implicit association test (Humphreys and Latour, 2013). The ability to measure cognitive legitimacy with both implicit and explicit measures would enable this construct to be captured in both experimental and survey settings, expanding its usefulness to researchers. In addition, it would be interesting to examine whether self-report measures of cognitive legitimacy converge with or differ from implicit measures, and which is more predictive of individual judgments and behaviors toward the organization.
As our goal was to develop a measure of individual perceptions, the participants in our study were drawn from samples of students and from Amazon’s Mechanical Turk survey engine. While these populations are generally viewed to be representative of the external audiences who would be evaluating an organization’s legitimacy, such as potential consumers or the general public, it would be beneficial to test the validity of the scale in assessing the perceptions of internal stakeholders as well. In addition, to control for differences in familiarity with the organizations, participants were presented with descriptions of the different organizations prior to making their legitimacy assessments. It would be valuable to test the legitimacy measure among current stakeholders of an organization, who would be able to form their legitimacy judgments based on their personal experiences with the organization. Finally, while we assessed the validity of the scale across a diverse set of organizations, the majority of these organizations were well established. It would be interesting to examine whether legitimacy perceptions differ when individuals are asked to judge nascent ventures as opposed to established organizations.
In addition, it is critical to bear in mind the multi-level nature of legitimacy to avoid the pitfalls of extreme individualism (e.g. Bitektine and Haack, 2015; Jepperson and Meyer, 2011). In other words, legitimacy as a perception exists not only within individual social actors but also as a social fact. Thus, legitimacy at the level of the organization, field, or institution cannot be reduced to solely the summation of individual level judgments. However, if legitimacy is truly a multi-level construct and not merely a social fact, then measurement at the individual level is justified so long as researchers bear in mind that the construct exists at multiple levels simultaneously. As a result, the proposed instrument may prove to be particularly useful in future research examining the interaction of individual legitimacy judgments with the broader forces at play. For example, media accounts have frequently been used as a proxy for legitimacy as conferred by critical institutions (Bansal and Clelland, 2004; Deephouse, 1996; Deephouse and Carter, 2005). One potential avenue for future research would be to examine the manner in which media accounts influence individual legitimacy judgments of focal organizations, industries, or fields simply by measuring the change in mean values for each dimension before/after news releases.
Legitimacy has evolved to be a core construct in strategic management research due to its central role in organizations’ survival and growth. A perceptual measure of legitimacy contributes to the advancement of this literature stream by permitting future researchers to examine the perceptions of individual stakeholders both internal and external to the organization, enabling a richer and more nuanced understanding of the process through which organizations can achieve legitimacy.
Footnotes
Appendix
Revised pool of potential scale items.
| Pragmatic legitimacy |
| Compared to similar organizations, this organization is better managed and operated. a |
| In general, this organization creates value for its stakeholders. |
| Compared to similar organizations, this organization better understands its customers’ needs and concerns. a |
| The policies of this organization cater to the interests of its shareholders. |
| This organization’s practices create value for its customers. b |
| This organization’s policies and procedures are beneficial to its employees. b |
| I believe the activities of this organization benefit their immediate stakeholders. |
| This organization’s procedures create value for its suppliers. a |
| Others who deal with this organization find value in their interactions with the organization. b |
| Compared to similar organizations, this organization is more innovative in developing new products/services. b |
| Moral legitimacy |
| The general public would approve of this organization’s policies and procedures. |
| Most people would consider this organization’s practices to be moral. |
| The way this organization operates promotes the common good. |
| I do not question whether this organization is operating as it should. b |
| This organization is concerned with meeting acceptable standards for ethical behavior in their field. |
| This organization deals with people fairly. b |
| I believe this organization cares about the well-being of their employees. a |
| I believe that this organization is good for the community. b |
| This organization’s policies seem appropriate. |
| The world is a better place because of this organization. a |
| This organization is concerned with “playing by the rules.” a |
| The policies of this organization adhere to acceptable standards as determined by regulatory agencies. b |
| If more organizations adopted policies and procedures like this one, the world would be a better place. |
| I believe this organization cares about the well-being of its customers. b |
| Most people in the general public would approve of these organization’s operating practices. a |
| Cognitive legitimacy |
| I expect to see organizations like this one. a |
| I believe that this organization fulfills a need in the economy. b |
| I tend to take for granted the value this organization provides. a |
| There must be a good reason for the way this organization operates. a |
| I believe that this organization is necessary. |
| In general, organizations like this provide an essential function. |
| The policies of this organization are typical of its field. a |
| I am comfortable dealing with organizations like this. b |
| The practices of this organization are acceptable. b |
| This organization is typical. b |
| The policies and procedures of this organization make sense. b |
| Most people would understand clearly why this organization needs to do what it does. b |
| I would feel comfortable interacting with this organization. b |
| I expect this organization to have specific policies and procedures. b |
| It is difficult to imagine a world in which this organization did not exist. c |
Eliminated in Study 3.
Eliminated in Study 2.
Added in Study 3.
Acknowledgements
This manuscript is based on the first author’s dissertation under the supervision of the second author. The authors would like to thank Sergey Anokhin, Greta Polites, Tatiana Stettler, and faculty and students at Kent State University for their feedback and suggestions on this research.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
