Introduction

To understand racisms’ roots in the United States, a longstanding business ethics issue, academics have sought to historically explain it, for example, in the ways American capitalism and slavery are interwoven (Beckert & Rockman, 2017; Baptist, 2016; Williams, 1994) and in how racism continues to be widespread in advertising through the cultural production of whiteness (Davis, 2018; Mitchell, 2020). Recent business studies on racism have been set within critical race theory (CRT), which seeks to identify the structural factors contributing to racial inequalities in business settings (Gold, 2016; Poole et al., 2021). Basic tenets of CRT propose that both race and racism are socially constructed, rooted within underlying institutions so that racism becomes ‘the usual way society does business’ (Delgado & Stefancic, 2017). CRT has been suggested as a factor in education inequities where institutional and structural racism are built into our educational systems (Ladson-Billings & Tate, 2016).

Others suggest a more subtle form of marginalization may be rooted at the micro-level in the business academy (Buttner et al., 2007). Aversive racism is “a modern form of prejudice that characterizes the racial attitudes of many whites who endorse egalitarian values, who regard themselves as non-prejudiced, but who discriminate in subtle, rationalizable ways” (Dovidio & Gaertner, 1996, p. 55). Whereas CRT portrays racism embedded in the structure as durable features of social life (Crockett, 2022), aversive racism takes an approach where the evolving socio-cultural process can be altered. The consequences of aversive racism (e.g., the restriction of economic opportunities due to race) are as pernicious as overt acts.

In this study, we propose that marginalization does not always emanate from a negative position but can also come from a place of neutrality. This means that the business academy, through solipsistic practices, sanctions this diminishment. ‘Solipsism’ is defined as social cognitive tendencies by an individual to focus on one's own internal states, goals, motivations, and emotions (Kraus et al., 2012). Like aversive racism, whether intentional or not, the racially minoritized become ‘othered’ because of the focus on self and ingroup members.Footnote 1 All others are in the outgroup (Messick, 1998). Unlike aversive racism, which is based on a framework of underlying subconscious negative feelings toward racialized minorities (Dovidio & Gaertner, 2004), solipsism is a neutral cognitive trait “lead[ing] to disengagement from the world and a privileging of self over others” (Gardiner, 2018, p. 31). In a white solipsistic world, marginalized populations, when recognized, are compared to the white norm, and inevitably become viewed as lacking (Moon, 1999). More often their existence is simply ignored. This ‘white’ view of the world can be detrimental to ethical action (Gardiner, 2018) and is often referred to as ethical solipsism.

While actively admonishing aversive racism, the business academy often rewards white solipsistic behavior. The focus on the white racial majoritarian in business ethics studies in the United States sets an ontological foundation for theoretical advancement and pedagogical methods that is white. This results in the white population of managers and business owners becoming the hegemonic category against which all other categories of managers and business owners are compared. This white solipsistic foundation establishes a biased epistemological approach influencing the ways academics pose research questions regarding racially minoritized populations, and how they teach business in universities (Hunter, 2002).

To explore these ethical issues, this empirical study is set in entrepreneurship, as it is an area that is typically viewed to be a ‘white endeavor’ (Ahl, 2004). We see possibilities of marginalization, the exclusion or ignoring, especially by relegating to the outer edge of a group (El-Bassiouny, 2014), represented in the racial make-up of business ownership in the United States where approximately 18.7% (1.1 million) of all U.S. businesses are minority-owned. This figure is significantly below the 40% representation of minorities in the overall population (United States Census Bureau, 2021). The dominant entrepreneurial narrative in the business literature portrays that of the heroic White man (Achtenhagen & Welter, 2011; Ahl, 2004) and under-represents racially minoritized entrepreneurs, which, when compared to this archetype, are presented as ‘deficient.’ In this study, we question: How does the business academy practice white solipsistic behavior in studies of entrepreneurship thereby contributing to the archetype of the ‘less-than’ under-represented racially minoritized (URM) entrepreneur? How is the identity of the racially minoritized reflected in their own discourse and how do universities (mis)represent them within its entrepreneurship discourses? Can altering the discourse of universities affect the entrepreneurial interest of nascent entrepreneurs?

Majszak (2019) describes white solipsism as a ‘I-it’ notion, a simple knowledge of others (‘it’) only in relation to oneself (‘I’). Thus, to empirically test for evidence of white solipsistic behavior calls for macro–micro-level analyses. Thus, to answer these research questions, we methodologically ground this study within van Dijk’s (1993) critical discourse analysis (CDA) framework for studying racism. The CDA approach bridges the gap between the macro-level (re)production and challenge of dominance in social structures with the micro-level cognition of individuals. The structure and property of text and images disseminated by symbolic elites, who determine what is published, act as communicative events that contribute to the production of dominance over the racially minoritized (see Fig. 1).

Fig. 1
figure 1

Operationalization based on van Dijk’s model of dominance through discourse structures

van Dijk (2015) identifies the professors who control scholarly discourse as ‘symbolic elites,’ and universities as having the ability to flex social power and manipulate beliefs through discourse. “As the producers, managers or brokers of knowledge, scholars are among the most prominent symbolic elites of contemporary society” (van Dijk, 1993, p. 158). Thus, in study 1, to elucidate the role of scholars as symbolic elites, a critical lens was applied to the literature to examine how collectively it portrays the racially minoritized entrepreneur. Within CDA, ‘critical’ should not be perceived as criticizing or being negative. Instead, it means being “self-reflexive in one’s research, and through these processes, making opaque structures of power relations and ideologies manifest” (Amoussou & Allagbe, 2018, p. 12).

In study 2, we conduct a content analysis of twelve spoken and written corpuses to differentiate the language patterns of outgroup racially minoritized entrepreneurs from ingroup racial majoritarian entrepreneurs. We also critically analyze the website content of more than 200 university entrepreneurship programs. In study 3, to explore the impact of university discourse on personal cognition, we employ an experimental design manipulating exposure to four different websites that vary in terms of the dominant language and imagery. With this experiment, we test the relationship between a white hegemonic presentation of entrepreneurship and the entrepreneurial interest of the nascent racially minoritized entrepreneur. We suggest a correlation between the academy’s portrayal of the racially minoritized entrepreneur and the entrepreneurial interest of these populations. Using the multi-method approach of van Dijk (2015) makes it possible to empirically identify how the academy may take white solipsistic approaches to promote a power base of privilege and the ethical implications that result.

These analyses provide three primary theoretical contributions: (1) elucidating how the racially minoritized are often egregiously ignored by the academy, and when compared, are portrayed as inferior, unworthy, and less successful compared to the White-man business owner who is valorized as superior; (2) proposing a racial epistemology that recognizes the unique identities of the US-born racially minoritized who often approach entrepreneurship from a communal or spiritual perspective with goals of family inheritance, mentorship, and community empowerment; and (3) demonstrating how white solipsism, a** non-racist, neutral social perspective, contributes to the marginalization of racially minoritized populations. From a methodological perspective, we demonstrate an operationalization of van Dijk’s macro–micro critical discourse analysis framework to explore how marginalization might emerge within an institution.

These findings call for greater reflexivity in business ethics (and entrepreneurship) research. Although reflexivity has been used as a tool for addressing power differences between the researcher and the test subject, it can also enable researchers to think critically about their own power relationships to the topics being studied (Awkward, 1995). “Epistemologies, by their nature, are hard to see beyond. Researchers then must redouble their efforts to illuminate the spaces they may inadvertently occlude. This allows them to discover the questions they are not asking, the categories they are not using, and interpretations they may overlook” (Hunter, 2002, p. 133).

Theoretical Foundation

We first clarify our use of the term ‘under-represented racially minoritized (URM).Footnote 2’ Cornell and Hartman (2006, p. p.25) write that ‘race’ represents “a group of human beings socially defined on the basis of physical characteristics.” Determining which characteristics constitute the race—the selection of markers and therefore the construction of the racial category itself—is a choice human beings make. Race as a social construct becomes difficult to define because it is not a scientific identification and is continually evolving in popular culture, history, and politics. To illustrate, in 1930, ‘Mexican’ appeared as a racial category on the census but due to political pressure from Mexico it was removed, and a Hispanic racial category has never reappeared (Demby, 2014). The 2020 census lists six options: White, Black or African American, American Indian or Alaska Native, Asian, Native Hawaiian and Pacific Islander, and ‘Other.’ ‘Hispanic’ is defined as an ethnicity, not a race.

The entrepreneurship literature often refers to the ‘ethnic minority entrepreneur’ (Aldrich & Waldinger, 1990; Bates, 2011; Volery, 2007) conflating the immigrant entrepreneur and the native-born URM entrepreneur, who may or may not identify with the familial ethnic culture into which these individuals were born. Depending on the standing within a society as an immigrant, a migrant, an indigenous natural-born or native-born business owner, political, spatial, economic, and regulatory contexts vary considerably (Kloosterman, 2010). Unlike the native-born ethnic entrepreneur, immigrants may have unique complications due to the strains of settlement and assimilation, further aggravated by government policies that constrain or hinder resource acquisition for immigrant entrepreneurs (Aldrich & Waldinger, 1990).

In this study, ‘under-represented racially minoritized’ refers to non-immigrants who self-identify as ‘non-White,’ including Hispanic or Latino, Asian, Indian or Native American, African American or Afro-descendants. ‘Racial majoritarians’ refers to ‘White’ individuals as defined by the US Census Bureau.Footnote 3 Both labels reflect a social construction of race that is dynamic. It is important to note that this entire racial framing is deeply embedded within the American racial context. The categories used in this paper reflect this setting and carry with them the biases and structures associated with discussion of race within the United States. We also interchangeably use ‘business owner’ with ‘entrepreneur’ following on Gartner (1988).

Role of White Solipsism in Marginalized Populations

The theory of white solipsism, originating from sociology, focuses on how subconscious racial habits by an individual possibly lead to marginalization of other populations. Overt racism and white solipsism both result in the same outcomes: construction of homogeneous groups, naturalization of cultural differences between groups, hierarchization and negative evaluation of the racialized, and legitimization of power differences between groups. Overt racism by an individual comes from a place of negativity, whereas white solipsism practiced by an individual comes from a place of indifference.

Sullivan (2006) suggests that race is socially constructed partially through white solipsism. She theorizes that white domination becomes constructed, maintained, and protected because of an individual’s subconscious bias toward one’s white space in society. Instead of acknowledging others’ particular interests, needs, and projects, racial majoritarians tend to recognize only their own place in society where ‘whiteness’ is perceived as a normative and universal condition (Sullivan, 2006). This is not a neutralization of race (Hunter, 2002) but an erasure of race by ignoring its existence.

An outcome of white solipsistic practice is ingroup/outgroup behavior (Messick, 1998). With ingroup favoritism/bias, a tendency to treat members of one’s own racial identity more favorably than non-members, emerges. Outgroup members are perceived as more homogeneous by ingroup members. As the marginalized population becomes the ‘outgroup’ there is less exposure to them, thus, they become stereotyped not intentionally, but because they are on the ‘outside’ of the institution. It is human nature to focus on ‘privileging of self over others’ so that those in the ‘outgroup’ are essentially non-existent. This study uses discourse analysis to identify which populations emerge as the ingroup and which as the outgroup to reveal whether the academy practices white solipsism.

Marginalization Theory in Business Ethics

In 1992, Stella Nkomo called for a rewriting of race in organizational theory. She argued the business academy had “amassed a great deal of knowledge about the experience of only one group [White men]” (Nkomo, 1992, p. 489). Organizational studies based on samples of White men managers do not routinely note that results should be considered as only valid for that group. Yet, the results of a study on racially minoritized managers add contingency disclaimers, providing minimal relevance for advancing organizational knowledge. “Thus, instead of race being an analytical category critical to the fundamental understanding of the organization, it is marginalized” (Nkomo, 1992, p. 490).

More recently, Alm and Guttormsen (2021) ground marginalization of populations in ignorance embodied through a failure, particularly by academic institutions, to embrace marginalized populations’ critical agency, or people’s ability to critically analyze their own social circumstances in ways that empower them to act and transform the situation (see Giovanola, 2009; Sen, 1985). Morris (2017) suggests that the White faculty predominating the academy constitute gatekeepers limiting epistemological research on race and racism. The exclusion of marginalized populations’ voices and topics in leading business and business ethics journals create a self-fulfilling prophecy in which the few published studies set a precedence suggesting a lack of interest in the subject or lack of need to explore the subject as a separate research topic. Smith (1999) refers to this as ‘conceptual imperialism.’

Surprisingly, few business studies, particularly those focused on ethics, have sought to understand the marginalization of the racially minoritized in white entrepreneurial capitalistic systems (Lee & Rodríguez-Pose, 2021). In the Journal of Business Ethics, marginalization has been tangentially referenced through diversity on boards and in workplaces (Buttner et al., 2007; Harjoto et al., 2015; Rabl et al., 2020) and in affirmative action debates (Libertella et al., 2007; Shaw, 1988). It has directly addressed racism in advertising (Canedo et al., 2014; Shabbir et al., 2014) and corporate responsibility (Azmat & Rentschler, 2017). Chowdhury (2021) calls for a more racially aware theory of the marginalized stakeholder in Western multinational corporations. The journal has tangentially addressed ethics in entrepreneurship with a special issue on social entrepreneurship (André & Pache, 2016; Bacq et al., 2016; Chell et al., 2016; Dey & Steyaert, 2016; Smith et al., 2016). This lack of studies on the occurrences and impact of marginalization demonstrates a white dominant paradigm prevailing in business ethics research, which is explored further in study 1.

Study 1: Construction of the Racially Minoritized Entrepreneur in the Business Academy

van Dijk (2015) suggests institutions’ social structures give power to specific individuals (symbolic elites) that may (re)produce social dominance through communicative events influencing the social attitudes, ideologies, and knowledge impacting the personal and social cognitions of individuals within the institution (see Fig. 1). Discourses occur at the micro-level of the social order. Marginalization, dominance, and inequality occur at the macro-level as supported by the social structure of the institution.

To understand how the racially minoritized entrepreneur is constructed in the business academy (an institution), we conducted a review of the business ethics literature. In it we found no studies on the intersection of entrepreneurship and race. Consequently, our review required looking beyond business ethics. A Google scholar search identified peer-reviewed academic papers published from 2000 to 2020 on the URM in entrepreneurship. A snowball method evaluating citations for each paper ultimately identified 36 different journals with more than 200 articles (see Table 1 for a list of journals and keywords searched). Only empirical studies (both quantitative and qualitative) where the unit of analysis was the US-based URM firm or URM individual were included. We excluded studies of immigrant populations or that co-mingled immigrant entrepreneurs with native-born entrepreneurs. Studies using race only as a control variable were also excluded as the sample sizes of the URM were typically small, resulting in little power to make conclusive findings. Forty-five studies fit our criteria. Most journals had either one or no publications except for the Journal of Developmental Entrepreneurship with nine papers, Small Business Economics with six, and the Journal of Small Business Management with three. This paucity of studies is not a reflection of the importance of these entrepreneurs on the U.S. economy. The U.S. Census of 2020 reported that 1.1 million racially minoritized owned businesses contributed more than $14 trillion in annual receipts to the economy (United States Census Bureau, 2021).

Table 1 Study 1—Literature review criteria

To analyze how the URM entrepreneur is portrayed in the literature, we utilize Gartner’s (1985) framework for describing new venture creations by identifying (1) individual characteristics/behaviors; (2) organizational structure; (3) impact of environment; and (4) process of new venture creation. Ahl’s (2002) approach for gender studies further subdivided these four factors into ten categories to construct how the business academy depicts racially minoritized entrepreneurs. Several of the studies took a race–gender intersectional perspective, making it useful to divide them into a separate category (see Table 2).

Table 2 Literature summary of US-based race and gender–race studies compared to racial majoritarians entrepreneurs/business owners

Individual characteristics indicate that the racially minoritized are more likely to start businesses than the racial majoritarians (Edelman et al., 2010) with Black women starting businesses at a faster rate than Black men (Edelman et al., 2010; Fairlie, 2004; Gibbs, 2014; Lofstrom & Bates, 2013; Sabbaghi, 2018). No motivational differences in racially minoritized and racial majoritarians to start a business appear; however, racial majoritarians exhibit higher motivations to grow their businesses (Edelman et al., 2010). Several studies suggest racially minoritized are ‘pushed’ into entrepreneurship for higher wages, lack of other opportunities, and desire to be their own bosses (Singh et al., 2008; Smith‐Hunter & Boyd, 2004; Wingfield & Taylor, 2016). In contrast, Carter et al. (2002) posited that African Americans are ‘pulled’ into entrepreneurship for self-realization, recognition, and a desire to innovate. African American women set high social and civic-responsibility goals with business ownership being a spiritual calling for many (Jones, 2017; Robinson et al., 2007). Both Hispanic (Liu, 2012) and Black women (Carpenter, 2011; Jones, 2017) see themselves as community role models.

Regarding organizational and managerial practices, URM owners have fewer employees, lower sales, are less profitable, and are more likely to be sole proprietor entities with service sector businesses (Bitler et al., 2001; Edelman et al., 2010; Fairlie et al., 2022; Freeland & Keister, 2016; Shelton & Minniti, 2018). Although Black men generate higher revenues than Black women (Gibbs, 2014), women demonstrate higher business survival rates (Robb, 2002). Latinas also underperform Latinos (Greene et al., 2003; Robles & Cordero-Guzman, 2007; Zuiker et al., 2003). URM women are often ‘doubly disadvantaged.’ Despite access to individual-level resources (i.e., human, social, psychological, and financial capital), external factors such as dynamism or hostility more negatively impacts their performance (Juma & Sequeira, 2017). African American women often define success not just on financial terms but more holistically (family wealth, community give back, serving customers, and mentoring) (Robinson et al., 2007).

From an environment perspective, it remains uncontested that the racially minoritized face higher barriers to capital access and start businesses with less personal capital, fewer community resources, and little external financing (Fairlie & Robb, 2010; Gibbs, 2014; Köllinger & Minniti, 2006). These barriers cause the racially minoritized to be less likely to seek capital (Fairlie et al., 2022; Neville et al., 2018) and more likely to use personal savings and familial funds (Freeland & Keister, 2016; Rhodes & Butler, 2004; Smith‐Hunter & Boyd, 2004). Both Hispanic women and African American women rely on familial commitments to support their endeavors (Chang et al., 2009; Ortiz-Walters et al., 2015; Robinson et al., 2007; Smith‐Hunter & Boyd, 2004).

Although Gartner (1985) identified six common steps entrepreneurs perform to create a new venture few studies addressed process. In step one of the process, opportunity identification, Black entrepreneurs particularly men are more likely to start companies with a clear need (Singh & Gibbs, 2013). Lofstrom and Bates (2013) observed that even with limited access to capital, Black business owners can more easily enter certain low-barrier industries.

Study 1 Summary

To summarize, major gaps exist in the literature regarding U.S.-based racially minoritized business owners, and how they start and build their companies. Published studies are biased to small business owners and none reported on the high-growth technology enterprises started by racially minoritized entrepreneurs. The results conclusively indicate that the URM will be under-funded, have smaller businesses, and underperform their racial majoritarian counterparts. Evidence suggests that even controlling for factors such as education, age, and industry, the URM still face disadvantages, and URM women experience a ‘double disadvantage.’ Racialized minorities, however, may regard success as community impact rather than through financial motivations. Several studies suggest that URM women start businesses with high familial obligations, social impact missions, and civic-responsibility goals with business ownership frequently grounded in religious faith, although this process is under-researched. Logically, the true motivations and measures of success of the racially minoritized in business ownership are largely invisible in the studies of entrepreneurship and business ethics.

This literature review provides insights into our first research question as to whether the academy contributes to the archetype of the ‘less-than’ URM entrepreneur. The answer appears to be an unequivocal ‘yes.’ The racially minoritized entrepreneurs are framed as ‘other’ in contrast with the dominant, normalized racial majoritarians. It seems likely that the white solipsistic epistemology of the academy contributes to this narrative. The lack of articles discussing the URM entrepreneur also reflect the role of academic elites gatekeeping to prevent focus on an outgroup.

Study 2: Entrepreneurship Discourse Analysis

In study 2, as shown in Fig. 1, we move from a discussion of institutional factors to an analysis of discourse structures within the context of communication events. This will then lead to study 3, which explores how altering these discourse structure may lead to cognitive changes to individuals exposed to these messages. More specifically, in study 2, we ask ‘how is the identity of the racially minoritized entrepreneur reflected in their own discourse and how do universities represent this group within its entrepreneurship discourse?’ By capturing the settings, identities, and the voice of the racially minoritized outgroup (Alm & Guttormsen, 2021), it becomes possible to evaluate the appropriateness of university discourse in reaching this population. Media, such as website content, calls for multimodal analysis to evaluate both the text and images (van Leeuwen, 2008). Following van Dijk’s (1991) systematic analysis of the content and structure of racist reporting in the press, a discourse analytical approach using a combination of qualitative and quantitative analyses is taken.

Although we found no studies that evaluated how language of the entrepreneur may differ by the race of an individual, management scholars provide insights on the possible communal nature of the racially minoritized. In studies, racially minoritized groups, identified as the disadvantaged, scored higher on communal orientation markers such as collectivism and familism (Gaines et al., 1997), whereas racial majoritarians, the advantaged group, scored higher on individualistic tendencies (e.g., Gaines Jr, 1994; Oyserman et al., 1995). In a study by Telzer et al. (2010), Latino study participants showed greater reward activity when they contributed to their families, whereas racial majoritarians participants demonstrated greater reward when they acquired money for themselves. Rucker et al. (2018) conclude that “Whites have been found to be more agentic, whereas Blacks, as well as some other minorities, have been found to be more communal” (p.97). Based on this extant literature and study 1’s revelation of the communal strategies often taken by racially minoritized, we posit the lexicon used by the URM entrepreneur will reflect language that is communal in nature and the lexicon of racial majoritarians will be agentic in nature. Thus:

Hypothesis 1

Racial majoritarian business owners (a) use more agentic language focused on individualistic, goal-oriented, competitive efforts compared to racially minoritized business owners and (b) use less communal language focused on community, family, and cooperative efforts compared to racially minoritized business owners.

The extant literature demonstrates how perceptions of ingroup racial majoritarian entrepreneurs excel, whereas the URM entrepreneur outgroup experiences lower levels of success, fewer opportunities, fewer networking partners, and accumulates lower earnings. When the racially minoritized occupy non-traditional entrepreneurial roles, their legitimacy and credibility are questioned, thereby leading to devaluation of their abilities and predictions of their failure as entrepreneurs (Foschi, 2000). Discursive marginalization of the outgroup occurs when the structure of the dominant talk and presentation of imagery in a discourse focuses on the ingroup, and this biased discourse generates hegemony (van Dijk, 1993). In the field of entrepreneurship, given that scholars typically present racially minoritized as the underperforming ‘outgroup,’ logic suggests that the hegemonic racial majoritarian (Achtenhagen & Welter, 2011; Ahl, 2004) ingroup, and their agentic language would dominate university websites. Thus:

Hypothesis 2

Universities present White men entrepreneurs as the dominant ‘ingroup’ using (a) agentic language and (b) predominately featuring these men in its imagery on their entrepreneurship websites.

Discourse Analysis Methodology

As noted, we first examine the lexical style of the URM entrepreneur to understand their discourse structure setting the context for discourse analysis of universities’ communicative events. Natural language processing represents the standard in text analysis to identify personality traits (Mairesse et al., 2007). To test H1a (racial majoritarians use more agentic language) and H1b (racial majoritarians use less communal language), twelve corpuses were identified where the speakers/authors described their journeys as entrepreneurs. The corpuses included (a) four written self-reflections of participants in a university-sponsored accelerator program; (b) four transcripts of YouTube videos of entrepreneur interviews; and (c) four blogs written by entrepreneurs as posted on Medium, the blog-hosting site. All corpuses were in the English language. A total of 63,724 words were analyzed. The split by attribution was 52.6% women/47.4% men; 40% URM/60% Majoritarian. See Table 3 for further breakdown.

Table 3 Descriptive statistics and correlations for 12 entrepreneurs’ corpuses

An agentic/communal entrepreneurial dictionary, based on Garcia’s (2022) study on the language of entrepreneurship, was entered into the Linguistic Inquiry and Word Count (LIWC 2015) to analyze the text of the corpuses (see online appendix Table 1A for a list of words). The LIWC results along with corpus characteristics were entered into SPSS to be analyzed using ANOVA. For example, for the YouTube video, ‘Four Successful Entrepreneurs Share Their Best Tips| Women of Impact’ was coded with a word count of 10,519 words, White race, women, verbal medium, year of 2019, 60% agentic, and 40% communal of dictionary words in the corpus. For details on the coding of all the corpuses refer to the online appendix, Table 2A.

To test H2a (university websites use more agentic words) and H2b (university websites feature White men), the 349 websites associated with the university entrepreneurial programs listed in AcceleratorInfo.com (2020) were evaluated. Eliminating duplicates, defunct programs, and programs without an active website resulted in a subset that we then validated for the active presence of a dedicated center for entrepreneurship (not just a major or minor concentration). This resulted in the identification of 212 university programs as sample. A proprietary scrapping tool gathered the text off the website homepages. A cosine similarity score (CSS) to determine the text similarity across the websites was calculated. When CSS is 1, the documents would be exact copies of each other; a value closer to 0 indicates that the two documents have little similarity. A CSSmean = 0.65 and CSSmedian = 0.70 across the 212 homepages showed a high similarity in the language used across the universities. A single pdf of the text from all the websites was compiled. This was entered into NVivo software for analysis with some words being excluded (e.g., proper nouns or stop words). Following the method used in Garcia (2022), we found that discourse structures varied not just by gender as identified in Garcia’s study, but also by race. This observation guided our focus for the subsequent discourse analysis in this study. The entire corpus of text from the websites was next analyzed with linguistic inquiry and word count (LIWC) using the same agentic/communal entrepreneurship dictionary utilized in the discourse analysis of the entrepreneurs. See online appendix Table 3A for the LIWC results. ANOVA was then used to test the differences in the use of agentic and communal language.

To test H2b, a team of trained of outside researchers working independently scanned each of the websites and identified 533 images with people. They then classified every individual in each photo by gender and race. Race–gender of a speaker was also noted if there was one in the photo. A combination of feminine characteristics such as style of dress, facial features, and body positioning were used to identify women. Skins tones (Mitchell, 2020), hair styles, and/or cultural attire were used to identify URMs. Any discrepancies in the identification by reviewers were rectified as a team with the authors.

We also qualitatively reviewed the visual communication from the websites. This analysis was rooted in the work of van Leeuwen (2008), which considers how racism is reflected visually in terms of the ingroup and outgroup categories and representation distance within images. As one example, images can show inclusion or exclusion through the actions captured in the image, the various categories used, and the amount of homogeneity. Relationships also may reflect group membership through the framing of actors in the image, such as using a low angle versus a high angle versus eye level in the picture. van Leeuwen refers to these various visual approaches as contextual strategies. We used this framework to analyze the images (see online appendix Fig. 1A for details). Two trained outside coders reviewed the 533 images to capture potential marginalization components of the images.

Results H1 and H2

Consistent with the previous studies on entrepreneurial language (Garcia, 2022), the text analysis revealed that as a group men use more agentic words than women (F(1) = 8.31, p = 0.016), and women use more communal words than men (F(1) = 6.35, p = 0.03). H1a, which proposed that racial majoritarians used more agentic words than the racially minoritized was supported (F(1) = 7.88, p = 0.02) and H1b, which stated that racial majoritarians use less communal words than the racially minoritized was also supported (F(1) = 6.52, p = 0.03). University websites used more agentic language (µ = 68.4, sd. = 16.9) compared to communal language (µ = 31.6, sd. = 16.9); t(211) = 31.3, p < 0.001) supporting H2a (see Fig. 2).

Fig. 2
figure 2

Mean agentic/communal usage by corpus type

Applying van Leeuwen’s (2008) critical discourse strategies to these images, clear evidence of marginalization practices appeared in images. As one example, multiple images of White men standing over minoritized actors and/or women appeared, often because the main speaker was a White man (see Table 4). It was common for a downward visual perspective on URM or women actors, or to have these groups positioned with their backs to the camera while racial majoritarian men would be facing the camera in the same image. The ingroup versus outgroup phenomenon was frequent with 54% of the images only having racial majoritarian participants and no minoritized individual. 15.5% of the websites had no image on the entire website of an URM participant creating that ‘white space’ where the racially minoritized is literally invisible. Further promoting the ingroup/outgroup presentation, some photos showed White participants wearing the same attire establishing a ‘uniform’ that must be followed to be part of the ingroup. In some cases, the blurring or cutting off the full images of non-majority participants occurred. Another extreme example of ‘othering,’ though likely not the intent, was an image of a woman with her mouth covered by a red hand signaling her speech being blocked. Thus, we found strong evidence that the websites predominately featured White men in their imagery in support of H2b.

Table 4 Image analysis of universities websites (n = 533 photos)

Following van Dijk’s recommendation to examine headlines in news media, the structure and function of the content section headlines of the 212 webpages were analyzed. Headlines have an important cognitive function as they set the context for the viewer, what van Dijk refers to as the ‘model of the situation,’ or the mental model the reader builds for the contextual situation. Predictable words frequently occurring were ‘entrepreneurship’ (observed 139 times), ‘center/s’(123x), and ‘business’ (61x). Infrequent terms were ‘social’ (3x), ‘community’ (10x), ‘women’ (2x), and ‘diversity, equity, inclusion, accessibility’ (2x). There were 57 uses of ‘innovator/innovation,’ often accompanied by adjacent headlines using agentic terms such as ‘competition,’ ‘challenge,’ and ‘prize.’ The combination of ‘small business’ and ‘innovation’ only occurred thrice, implying that small businesses do not innovate. Websites using ‘community’ in one headline were less likely to include references to competitions, pitches, and challenges in other section headlines. When ‘networking’ was used, it was more likely to be accompanied by headlines related to inclusivity of women and diversity. See Table 5 for examples of the co-occurrences of headlines on the same webpage.

Table 5 Headline co-occurrence examples of university websites

Study 2 Summary

These findings, as suggested by the literature review, indicate that the racially minoritized use more communal language when speaking about entrepreneurship. This study may be the first to investigate the language of entrepreneurship used by the racially minoritized, although similar studies have examined gender differences (Ahl & Nelson, 2015; Bird & Brush, 2002; Garcia, 2022). In contrast, the universities websites were found to be agentic in language and featured White men, thus, demonstrating another example of how white solipsism is dominant in the academy’s discourse concerning entrepreneurship. A majority of the images lacked any minoritized races at all and many of the images were drastically dismissive of the URM communities, sometimes to a shocking degree. Overall, the images clearly signaled who should, and should not, be an entrepreneur reflecting a clear, White, men ingroup. Few of the headlines separating content sections on the websites used communal language, but those that did were more inclusive in their images and language clearly targeting a more diverse student population.

Recognizing that the literature mostly ignores the URM, and university communications around entrepreneurship, specifically language and images, are deeply biased toward the racial majoritarian, we continue by exploring the effect of changing these communications. More specifically, study 3 investigates how changing the discourse of university websites may affect the entrepreneurial interests of students.

Study 3: Impact of Discourse on Social Cognition

Universities, through communicative events, have the power to influence the socially shared attitudes, ideologies, and knowledge of individual recipients (van Dijk, 2015). By having the ability to influence people’s minds they indirectly may control (some of) the actions of these people. Thus, to complete the link between social structure–communicative events–cognition, or what van Dijk terms the discourse–power circle, we conduct an experiment testing what happens when the discourse of communicative events are changed. This effort seeks to answer the research question, ‘can altering the discourse of university websites impact the entrepreneurial interests of nascent URM entrepreneurs?.’

As a first step in this analysis, we orient the personal cognition (as referred to in Fig. 1) for the URM students by establishing their entrepreneurial interests. Several extant studies have revealed that the racially minoritized have a higher interest in starting business than racial majoritarians (Köllinger & Minniti, 2006; Walstad & Kourilsky, 1998; Wilson et al., 2004). Minority social networks (Walstad & Kourilsky, 1998; Wilson et al., 2004), familial and community support (Canedo et al., 2014) as well as the potential benefits of financial independence (Wilson et al., 2004) often contribute to the above average levels of confidence and optimism associated with higher rates of early-stage entrepreneurship for the racially minoritized. Consistent with these studies, we expect that this population desires to be entrepreneurs, or in terms of this study, to be a member of the ingroup. Thus:

Hypothesis 3

Racially minoritized students will have higher entrepreneurial interests compared to racial majoritarian students.

We next examine whether the racially minoritized student’s entrepreneurial interest may be affected when encountering white solipsism in entrepreneurship program presentations. Homophily, the tendency for people to seek out or be attracted to those similar to themselves, is well recognized by many disciplines (Currarini & Mengel, 2016; Turner, et al., 1979). People also tend to treat others of shared social identity more favorably (ingroup bias) and discriminate against those not in their social identity group. Labor market stereotyping provides an example of how social identity impacts the choices of decision makers who tend to discriminate in favor of candidates of their own race (Bertrand & Mullainathan, 2004). Thus, we hypothesize that a white solipsistic presentation of a university website with agentic terminology featuring White men will discourage the racially minoritized’s participation in campus programs because they do not align with this ingroup.

Hypothesis 4

White men-dominated university entrepreneurship websites decrease the interest of racially minoritized in exploring entrepreneurship.

Data Source

A student sample population is particularly well suited for this experiment as it enables the ability to define the institutional boundaries and conduct the experiment in controlled settings of a business research laboratory or classroom. A university setting also provides within group equality for entrepreneurial opportunities regardless of gender, race, socio-economic status, education, or region as university-sponsored programs and courses are accessible to any enrolled student. By surveying enrolled students, the cultural bias and barriers known to discourage entrepreneurial endeavors of women and minoritized populations can be minimized.

Experimental Design

Four separate websites (C1:communal-centric featuring URM women, C2:agentic-centric featuring URM men, C3:communal-centric featuring White women, and C4:agentic-centric featuring White men) were designed. Students were randomly chosen to view one of the website conditions. For demographics and a breakdown of number of respondents per condition, see Table 6.

Table 6 Demographics of survey respondents (n = 463)

To help support ecological validity, the logo/brand for the university attended by each respondent appeared on the website. The primary institutional mechanism incorporated into the website was a promotional video featuring two spokespersons that viewers streamed online. Each video featured two spokespersons of the same race–gender intersectionality talking about their enthusiasm for entrepreneurship programs. The images in the websites were changed in alignment with the intersectionality of the video spokesperson. Finally, the language used in the text of the website reflected agency or communality (c.f., Ahl, 2002). Examples of agentic language included, ‘competitive program,’ ‘risk-taking visionaries,’ and ‘revolutionary creativity’ reflecting masculine entrepreneurial language. Examples of communal language on the website included, ‘supportive program,’ ‘building a community,’ and ‘caring business owners’ (see Fig. 3 for an example).

Fig. 3
figure 3

Website condition examples

To test the manipulation whether the images and videos used on the website conditions represented the race and gender of the spokespersons as intended, two separate pre-studies were completed. A sample of 598 Amazon Mechanical Turk respondents correctly identified the race or gender of the spokespersons in the images/video in the majority of cases. In addition, a separate sample of students from a private, Western U.S. university completed an online survey as a requirement in an entrepreneurship course assessing the target market for the websites used in the experiment. Respondents correctly identified the target market in most cases. Results of the manipulation checks on race and gender separately may be found in online appendix Tables 4A and 5A.

With the manipulation check tests successfully passed, the experiment with a sample of students at five universities in the U.S. representing the eastern and western regions of the country was undertaken. Students participated either through an undergraduate marketing lab or part of a marketing course requirement with 468 students completing an online survey. Respondents failing to self-identify on race or gender were dropped from the study, leaving a sample size of 463. The breakdown was 321 respondents from the Western region and 142 respondents from the Eastern region; 240 self-identified as females and 223 self-identified males; 155 self-identified as racially minoritized and 308 as racial majoritarians students.

Dependent Variable

The dependent variable tested entrepreneurial interest within the university context after exposure to the website condition. Entrepreneurial interest was measured with four items on a seven-point semantic differentiation scale from extremely likely to extremely unlikely, asking interest to (a) take a class in entrepreneurship; (b) refer a friend to an entrepreneurship program on campus; (c) graduate with an entrepreneurship degree (major or minor); and (d) seek information about entrepreneurship activities on campus. These items were based on previous empirical studies on college students’ interests in pursuing self-employment (Liñán & Chen, 2009; Souitaris et al., 2007) and were adapted for the context of entrepreneurship education. The Cronbach’s α of the measures was 0.84 well above 0.70, the recommended threshold measure (Straub, 1989). A principle-components analysis using a varimax rotation indicated that all four items loaded well as one single item, with the lowest eigen value for the item ‘likely to refer a friend’ equal to 0.310, well above the 0.1 cutoff (Kassambara, 2017).

Independent Variables and Control Variables

The two independent variables in the two-way ANOVA analysis were race (as self-identified) and website condition (C1, C2, C3, C4). Five controls were tested: year in school (first year = 1, second year = 2, etc.), whether currently an entrepreneur, whether they had ever taken a class related to entrepreneurship, income level, and region in which the school is located (west = 1; east = 2).

Results H3 and H4

A two-way ANOVA tested: (1) difference in entrepreneurial interest by race; (2) difference in entrepreneurial interest by exposure to one of the four websites; and (3) for an interaction effect between race and website condition on entrepreneurial interest. The data were first examined to determine whether they adhered to the assumptions of a two-way ANOVA. The dependent variable, entrepreneurial interest, was normally distributed (p > 0.05) as assessed by Shapiro–Wilk’s test of normality. No outliers were found. There is homogeneity of variances as assessed by Levene’s test for equality of variances with p = 0.34. Several models with the different control variables were then examined. This analysis revealed no direct effect for the control variables ‘year in school,’ ‘entrepreneurial status,’ ‘income’ or ‘region.’ The analysis did find an effect for ‘taken a class,’ which is expected. A student who has previously taken a course in entrepreneurship would be inclined to express a higher entrepreneurial interest when compared to a student who has not. Subsequently, this variable was included as a covariate in the two-way ANOVA analyses.

A univariate general linear model with entrepreneurial interest as the dependent variable was conducted. A Tukey post hoc test revealed that the URM respondents have higher entrepreneurial interest (µ = 4.77, sd = 0.139) compared to racial majoritarians (µ = 4.38, sd = 0.09). Thus, we find support for H3 (F(1) = 246.6, p = 0.040). A subsequent race–gender intersectional analysis showed that racially minoritized females have greater entrepreneurial interests (µ = 4.84, sd = 0.17) compared to racial majoritarian females (µ = 4.20, sd = 0.13) (F(1) = 9.1, p = 0.003).

Main effects for ‘race’ and ‘website condition’ were significant confirming entrepreneurial interest varied by race (F(1) = 5.03, p = 0.025) and by the website condition viewed (F(3) = 4.64, p = 0.003). Across all students the website condition C1, featuring the URM females and communal language, resulted in the highest entrepreneurial interest when compared to the other website conditions (F(3) = 4.86, p = 0.002). An analysis of the interaction between race and the website condition revealed taken as a single group that there was no significant difference in entrepreneurial interest based on website condition for the racially minoritized (F(3) = 0.89, p = 0.447). However, taking a race–gender intersectional analysis, we found that the racially minorized males had the highest entrepreneurial interest after viewing C3: White women (µc3 = 5.22, sd = 0.43) and significantly lower interest after viewing C4: White men (µc4 = 4.30, sd = 0.51). In contrast, racially minoritized females had the highest interest after viewing C4: White men (µc4 = 5.65, sd = 0.40). Thus, there is partial support for H4 that a white hegemonic presentation of entrepreneurship dampens the entrepreneurial interest of the racially minoritized; it is contingent on gender. Surprisingly, for racial majoritarians, the website C1: URM women had significantly higher effects on entrepreneurial interest (µc1 = 5.00, sd = 0.18) compared to the other three conditions (µc2 = 4.22, sd = 0.13; µc3 = 3.83, sd = 0.18; µc4 = 4.23, sd = 0.19). This effect was consistent even after accounting for gender. See Fig. 4.

Fig. 4
figure 4

Graphical presentation of EMMs

Study 3 Summary

Consistent with the literature review, despite the barriers the racially minoritized encounter as entrepreneurs, our sample of racially minoritized university students expressed higher entrepreneurial interest compared to their racial majoritarian counterparts. Using an experiment varying the website discourse presented to students, we found a White men hegemonic website presentation negatively impacted the entrepreneurial interests of students, except for racially minoritized women. They had the highest entrepreneurial interest after viewing this website condition. Our post hoc test revealed that the highest effect on entrepreneurial interest on the racial majoritarian ingroup was from an outgroup minoritized women centric website. The results of this study show that the website communicative events of universities can negatively impact the personal cognition of the majoritarian population not just the minoritized members of their communities.

Discussion

In this study, we asked if the business academy practiced white solipsistic behavior, and if yes, how? We identified three ways that this occurred in entrepreneurship: (1) setting the white-man-led organization as the hegemonic condition against which all other entrepreneurs are compared against; (2) failing to establish racial epistemologies that could account for the different processes by which URMs pursue entrepreneurship; and (3) presenting the discourse of university entrepreneurship websites as agentic and ‘othering’ the racially minoritized, who through the lack of presence on these websites are established as the outgroup.

More specifically in study 1’s literature review, the major themes identified include:

  • A ‘white space’ through the omission and under-representation of URMs as research subjects in entrepreneurship studies, and more egregiously in the business ethics literature;

  • Conflating the entrepreneurial experiences of the ethnic minority immigrant compared to the native-born URM entrepreneur;

  • Outgroup homogenization of the racially minoritized despite huge cultural differences in Black Americans, Hispanic Americans, and Asian Americans;

  • An over emphasis on environmental negativities, such as funding challenges and fewer resources, instead of positive individual traits such as social impact goals and networking strengths that empower the racially minoritized;

  • Entrepreneurship as an elusive dream of business ownership for URM populations.

The narrow epistemic framework that emerges has either occluded the URM entrepreneur’s lived experiences (Alm & Guttormsen, 2021) or primarily reported comparative statistics highlighting their failures (Fairlie & Robb, 2010). In short, academics not only expect the racially minoritized to be different from racial majoritarians; they also expect them to be deficient (Marlow & McAdam, 2013).

A notable omission in the literature is a lack of studies on the entrepreneurial process used by URMs. The entrepreneurial journey of the racially minoritized as they identify business opportunities, secure funding, and scale their organizations (Gartner, 1985) is under-researched. Further, the current epistemological paradigm ignores the reality of intersectionality that URM men and women experience entrepreneurship differently, which may have an impact on strategic orientation and subsequent performance of these distinct groups (Gibbs, 2014; Robinson et al., 2007). Identifying these two diverse groups as a single population does a disservice to both (Crenshaw, 1997).

The discourse of URM entrepreneurs as they speak about their journeys through interviews, blogs, or self-reflections compared to the discourse of university entrepreneurship websites shows a disconnect as university websites use two times more agentic words than communal and website photos showed three times more racial majoritarians than racially minoritized. The headline and image analyses further demonstrated the existence of the strong agentic associations consistent in the language, and the images portrayed the ingroup as the racial majoritarians with the outgroup as the racially minoritized. This provides further evidence of white solipsistic practices. These findings are revealing because the discourse structure of the URM entrepreneur tends to be communal. This communal language is consistent with the literature findings that racially minoritized are more likely to measure success by familial, community and social impact, and not profits. For example, faith, passion, and determination were found to be the most important success factors for African American women (Awadzi, 2019; Ervin, 2014).

By examining the impact of race and gender in university websites on student’s entrepreneurial aspirations, this study reveals how real exclusionary consequences of institutional bias may impact future generations of business ownership.Footnote 4 The students in our sample, except for the female URMs, were negatively impacted by the White, agentic presentation of the discourse. Surprisingly, entrepreneurial interest for racial majoritarians was highest after viewing the website C1: URM women. Subsequent informal discussions with students revealed a ‘novelty’ effect with the URM woman spokespersons since traditionally they are rarely featured in entrepreneurship websites of U.S. universities in the time period the study was conducted; consequently, students indicated they paid more attention to the content. Unique or novel presentations garner more initial attention that then leads to deeper processing of discourse through marketing messages (Till & Baack, 2005). This could also explain why the female URMs in the sample were not more strongly influenced by C1, the content was not novel. This effect is problematic, as it will more than likely wear-out over time, as do most novelty effects (Chen, et al., 2016).

Together our three studies demonstrate that white solipsistic discourse, even though non-racist and neutral in its communicative events, can lead to the marginalization of URMs. Majszak (2019) suggests that because white solipsism limits perception it causes ignorance and epistemic blindness regarding alternative viewpoints. We illuminated how this blindness becomes a subconscious habit of racial privilege (Sullivan, 2007) within the business academy as revealed by the research questions asked and how universities communicate with their constituents.

Ethical and responsible leadership have been suggested as solutions to eliminate marginalization (Knights & O’Leary, 2006; Maak & Pless, 2009; Werhane, 2008). This approach may work if aversive racism is the contributing factor as it can be identified as a vice. Because white solipsism’s pre-occupation with ‘self’ negates the need to consider other perspectives, it cannot as easily be redirected through leadership. Instead, white solipsism calls for individuals to change their mental models by consciously identifying one’s role in marginalization (Werhane, 2008). Reflexivity, which requires the researcher to think critically about their power relationship with other people or institutions, is increasingly used as a tool to reduce bias in research studies (Hunter, 2002). It demands researchers to acknowledge the questions that they are not asking, to consider categories they are not using, and to explore interpretations they have overlooked. Including racial epistemologies in constructing research questions, and not just research on race, is important because it exposes the fact that research questions are not neutral. Eliminating the white solipsistic perspective of the academy will require a heavy dose of self-reflexivity.

Conclusion

With this study, we seek to shift the mental model of racism away from ‘conceptual imperialism’ (Smith, 1999) to one that increases the boundaries of inquiry to be more inclusive of all populations. Business academics seem to have fallen prey to white solipsistic behaviors and now have an ethical responsibility to rewrite the narrative of the ‘disadvantaged’ minority entrepreneur that it has helped to established in society. This is not to suggest that academics distort history or misrepresent the facts, two areas that critical race theorists adamantly fight against. Instead, aligning with Alm and Guttormsen (2021), we want to hear the voice of marginalized populations whose stories are not being told by the academy. For entrepreneurship scholars, this calls for less reliance on U.S. Census data and other statistical archives, acknowledgment of the heterogeneity in the URM populations, and more reporting on the high-growth tech companies started by the racially minoritized. For business ethics scholars, this calls for recognition of racial epistemologies to better understand minoritized populations. Finally, the voices of minoritized academics themselves must be recognized. Historically, it is the White men academics that write about the URM business owner. Their white solipsistic viewpoint, even if it is unintentional, leads to the hegemonic foundation currently reflected in the literature and on university websites. An important finding of this study is that racial majoritarian students were positively impacted by viewing websites featuring communal language and minority women (our website condition C1). This suggests that the academy could move from white solipsistic presentations and embrace diversity in their presentations of the URM entrepreneur without negatively impacting their appeal to racial majoritarian students.

As with any empirical analysis, this study has limitations. It suffers from a binary representation of race and taking race as homogenous for racially minoritized populations. This amalgamation is something we have criticized and represents an area for improvement and future research. Deeper qualitative analysis would help to alleviate this myopic viewpoint, which we must also leave for future research. Additionally, we purposely took a US-centric perspective, thus, the study cannot be generalized for populations outside of the United States. We would, however, expect similarities in results for studies in other countries as hegemony takes many manifestations around the world. Another limitation is in the role of academic websites and how students consume those websites. We recognize that our experiment is focused on the effects of one website on student responses. While this captures one type of discourse students use to learn about entrepreneurship on campuses, it is just one of many possible activities.Footnote 5 Future research might also explore other higher education interventions beyond promotional efforts, such as course descriptions and syllabi, which may also influence students’ intent to study entrepreneurship while on campus.

As critical discourse analysis is meant to be normative, we suggest that future research should refine an epistemological foundation for a racialized theory of entrepreneurship to better understand the processes the racially minoritized use to start and build their companies. The academy should embrace this ethical obligation to transparentize the ‘invisible’ minority entrepreneur and fill the ‘white space’ (Alm & Guttormsen, 2021) by changing the framing and context of business research to be more inclusive.