Entrepreneurship is a field of research whose concept is constantly evolving (Lemos, 2016) and bifurcated, in its trajectory, in an economic order, and another, behavioral one, which seeks to clarify the nature of the decision-making process.

Concerning the economic current, according to Lemos (2016), studies on entrepreneurship are focused on decision making and the attempt to predict the future and look for opportunities in it. It encompasses, in this way, the unfolding of economic theory (such as consumption choices, without and with risk; statistical decision theory; and game theory, for example), considering that the individual must have as a purpose the intention of a better alternative than the east for opportunities, based on the development of specific characteristics (Lumpkin & Dess, 1996, 2001; McGrath & MacMillan, 2000).

The behavioral approach of entrepreneurship, on the other hand, involves the daily life of the organizational life and the individuals because they understand that decision-makers often have incomplete data, which leads them to choices based on maximizing the subjective reality (Gartner, 1988).

From an economic or behavioral perspective, anyone can own their own business, as long as they have a purpose and conditions to put it into operation (Wang et al., 2015). In addition, according to the criterion of equal opportunities, all individuals start their businesses from a common point, obtaining success on their own merits (Krueger et al., 2000). However, in societies that protect historical inheritances from overcoming one group over the other, the concepts of equality and meritocracy are opposed (Lawton, 2000), which makes the study of Afro-entrepreneurship relevant, for example, since it is a way to understand the entrepreneurial process from the perspective of individuals segregated in society.

Unlike populations originating in other parts of the world, who sought migration as a chance to escape from an unwanted environment or even to glimpse the opportunity of success on distant lands, the vast majority of the Afro-descendant population was forced for several centuries to migrate from inhumane forms (Russell-Wood, 1978).

The look on the ethnic aspects of entrepreneurship, however, is not recent nor restricted to Brazil. In the United States, longer-lived affirmative policies have had moderate effects on Afro-entrepreneurship (Boston, 1998), following the warning given by Green and Pryde (1989), that the African American population, in addition to suffering from socioeconomic and historical conditions, responsible for keeping their income below that of the white people, facing the highest unemployment rates, having more difficulty in obtaining resources to invest and, consequently, reaching levels of well-being perpetuated at levels below the average of the American population.

The act of entrepreneurship has always been present in the daily lives of Brazilian Afro-descendants for reasons of survival and necessity (Ribeiro, 2018). According to Paixão (2015), there are numerous studies on ethnic discrimination present in the labor market and access to public services. Such studies, together with human capital theory, offer interpretations of the factors considered to be the main determining factors for social or wage inequality (Śliwa, 2007). The possession of personal social capital, such as schooling or length of experience in the profession, also influences the determinants of the occupational and remuneration trajectories of individuals, linked to their collectivities, usually discriminated as: women, blacks, indigenous people, and other ethnic groups (Paixão, 2015).

Recently, Afro-entrepreneurship has gained theoretical notoriety due to publications in journals of great repercussion in the economy (Kopkin, 2017), and for representing a legitimate socioeconomic demand, spearheaded by the “Black Money” movement (Berth, 2018), whose central proposal is to make money circulate among the Afro-descendant population, establishing a strategic reversal of the meaning of power, based mainly on strengthening the entire community and the black population. Thus, both the product (since its planning) and the advertising are aimed at black audiences, who cannot identify with what is offered to them.

Responses to this movement seek to lead non-white characters, or outside the dominant aesthetic standards, in the advertising of companies. An example of this is in the “Black Lives Matter” movement and how it has affected the media in the United States and, consequently, worldwide, exploring the image of hip- hop artists (Harlow & Benbrook, 2019), among others.

Thus, in conjunction with other actions, Afro- entrepreneurship enables the rise of people of African descent as a consumer, capable of contributing to the success of affirmative social inclusion policies in Brazil, such as the Quota Law (Law 12.771/12, 2012).

It is worth mentioning that Afro-entrepreneurship exists long before its concept (Ribeiro, 2018). Today, it surpasses the frontiers of subsistence because of the search for improving skills and competencies related to the entrepreneurial attitude. As a result, people of African descent are betting on creating, opening, and managing their businesses, facing prejudices in different places on the planet (Kopkin, 2017).

Given the disastrous legacies of discrimination, this population tends to have less economic power, being socially segregated and dependent on public policies that establish equal opportunities in the economy (Guimarães, 2004). As a result, the non-white entrepreneur tends to start a new business more out of necessity than through opportunities (Bosma & Kelley, 2019). In this sense, it is argued that, while opportunity entrepreneurship is based on meritocracy, the necessity is based on the inequality of the initial conditions (Lawton, 2000).

Suppose the social conditions, inherited by historical factors such as slavery, affect the act of undertaking (Kopkin, 2017). In that case, the Afro-entrepreneur is expected to be less subject to the detection of opportunities, reflected in income. In this way, entrepreneurship by necessity, aligned with lower conditions of employability, denotes more a state of self- employment than the choice of a potentially profitable business alternative (Margolis, 2014). In this sense, Hypothesis 1 is proposed to test whether ethnic condition determines the entrepreneur’s income.

H1: The entrepreneur’s ethnicity influences income; thus, the income of white entrepreneurs is higher than that of non- whites.

In this article, we tried to identify ways to be chosen by entrepreneurs characterized as non-white to mitigate the situation, based on their behavior, being relevant, in this context, using the effectuation theory.

The effectuation theory, proposed by Sarasvathy’s (2001) thesis, guided by Herbert Simon, was based on: (a) Knight’s contributions (1921) on the entrepreneur’s role for the company’s profit, in times of uncertainty; and (b) by Weick (1979), about the central role of the entrepreneur in the process of evolution of the organization; as well as (c) in the argument of Cyert and March (1963), that it does not make sense to make a decision today, in objective terms, because the results are known only in the future, being, therefore, subject to oscillations resulting from the uncertainties ; (d) in the trade-off between the exploration and the extraction of knowledge (March, 1991); and by the reasoning of Mintzberg (2000), that strategy does not necessarily mean strategic planning.

Thus, Sarasvathy (2001) proposed that most businesses do not originate from pre-established plans, but from the entrepreneur’s cognitive reactions, before and during the process of developing their business, molded between their intuition and their ability to assess risks and uncertainties.

According to the effectuation theory (Sarasvathy, 2001), two logics are predominant in the entrepreneur’s behavior: (1) causation, based on the prediction of a specific effect, and on the selection among the personal means (resources and skills) to create it; and (2) effectuation, centered on a set of means and the choice between its possible effects, created from what entrepreneurs are, know and how their contacts can contribute to the business.

As much as they differ from each other, these logics do not represent two opposites of the same line, given their orthogonal characteristics, which allow the combination of more rational elements (causation) with more intuitive ones (effectuation) (Matzler et al., 2014). In addition, causation can address the orientation towards opportunities within the effectuation theory, recognizing the path of entrepreneurship studies constituted until the turn of the millennium.

However, more recent research indicates that these logics are not limited to new businesses but also occur in situations of uncertainty (Yang & Gabrielsson, 2017). Even when entrepreneurs cannot design scenarios, they can establish plans to minimize the effects of what they cannot control. In this sense, the effectuation theory gains space not only in the discussion of how business emerges (small companies or multinationals) (Dash & Ranjan, 2019) but in how they are consolidated in complex scenarios (Johansson & McKelvie, 2012).

In this way, the entrepreneur who elaborates adequate planning and has a clear objective to be reached will obtain better returns to their business according to the logic of causation. In doing so, a future to be pursued is envisioned (Heracleous & Jacobs, 2011; Sarasvathy, 2001), regardless of the size of the business (Shi et al., 2013).

So, even if the development of a plan does not guarantee success, in terms of the ability to assess the perception of risk (Frishammar & Andersson, 2008), the entrepreneurs who best define the horizon to minimize uncertainty (Welter & Kim, 2018), will tend to achieve more satisfactory financial results.

In addition, the causation logic can have different effects on performance, depending on the level of technological intensity of the business (Kristinsson et al., 2016). Thus, entrepreneurs with more characteristics of causation behavior tend to achieve higher performance because they plan better. Better planning, in turn, allows the entrepreneur to seek previously crucial resources, knowledge, and partnerships to assist in the performance of the business. Therefore, it is expected that causation behavior will have a positive effect on the entrepreneur’s income. Based on this reasoning, Hypothesis 2 was formulated.

H2: The entrepreneur’s causation behavior has a positive effect on income, regardless of socioeconomic condition

Even though the definition of plans and goals collaborates to develop more successful strategies, what often brings success is the ability to change plans (Einhorn & Hogarth, 1987). In addition, most micro and small entrepreneurs, especially those who start a business out of necessity, rarely plan their actions (Block & Sandner, 2009). According to Chandler et al. (2011) and de Vasconcellos et al. (2019), in the effectuation logic, opportunities do not pre-exist; they arise from the knowledge built along the entrepreneur’s trajectory and need to be constantly changed, to innovate, including the business model (Snihur & Tarzijan, 2017).

Thus, opportunities are developed in an intense communication and negotiation process between those involved, in the internal or external environment, to practice concrete aspirations, values, products, services, and institutions, resulting from interpersonal connections built over time (Peng & Luo, 2000).

It is understood, then, that the effectuation logic is established in the changes that occur outside the traditional and rational model and is defined by the search for the effect, not dependent on the cause (Sarasvathy & Dew, 2005), based on how much the entrepreneur is willing to risk this change of path (Kalinic et al., 2014) and the perception they have about how much they are capable of assuming as a loss.

According to Faia et al. (2014), the effectuation logic is related to the exploration of opportunities in new markets with a high degree of uncertainty, favoring an experimental process and interactive learning, which allow the entrepreneur to discover information about the future, considering the environment as dynamic, non-linear. As the future is intrinsically immeasurable and liable to transformation, its design becomes unnecessary (Sarasvathy, 2001).

The effectuation theory (Sarasvathy, 2001, 2008) assumes that entrepreneurs start their businesses based on the science of three premises: who they are, what they know, and whom they know to enhance their business. So, their traits, tastes, and skills; their knowledge bases; and the social networks of which they are part support the possibility of entrepreneurship (Sarasvathy et al., 1998; Venkataraman et al., 2012).

In contrast to the logic hitherto prevalent in studies on entrepreneurship, in effectuation, entrepreneurs: (a) stipulate an acceptable loss amount, determined by the limitation of resources; (b) seek to experiment with other strategies, considered feasible; and (c) seek to build strategic alliances, formal or informal, through their networks of contacts, to improve their performance in the market. Thus, difficulties and limitations are transformed into opportunities (Kalinic et al., 2014; Sarasvathy, 2001) and, little by little, as they start a business, they build their future, shaping, during their entrepreneurial action, the strategies emerging (Heracleous & Jacobs, 2011; Mintzberg, 1987).

As the effectuation theory matured, it was noticed that the logic, first aligned with the new businesses, also made sense for long-standing ventures, gaining greater attention from researchers, for agreeing with the perception that the future is built daily (Perry et al., 2012). As a result, it started to be equally used in cases in which there were: (a) difficulty in identifying how to deal with marketing actions (Yang & Gabrielsson, 2017); (b) internationalization in little-known markets (Galkina & Chetty, 2015); (c) situations of uncertainty, experienced by multinationals (Dash & Ranjan, 2019); and, mainly, (d) scenarios where competition is not due to the size of the company, but through the ability to respond, in environments of uncertainty, which require intuition (Brettel et al., 2012), in particular, when entrepreneurial effectuation attitudes are combined with causation management (Berends et al., 2014).

In this sense, responses aligned with the effectuation behavior usually bring better results in stagnant or low profitability industrial sectors. In contrast, in the most robust and stable earnings sectors, causation logic tends to maximize earnings (Futterer et al., 2017).

In societies where race segregation was an element in building their economies (Lawton, 2000), the least favored developed specific skills, such as experimentation, flexibility (Pacho & Mushi, 2020), and the establishment of relationship networks based on sympathy (Miura et al., 2019). These individuals tended to survive in better conditions than those who resigned themselves to political and socially imposed differences due to the skills to deal with uncertainties inherent to the business and to provide answers and better performance in an unstable environment when using the effectuation behavior.

Taking into account that the entrepreneurial behavior, based on the logic of causation (H2) and effectuation, are orthogonal, being able to coexist and bring different benefits to the business (Matzler et al., 2014), it is assumed that the effectuation behavior leads to gains in the income of the entrepreneur, overcoming the socioeconomic conditions that affect it negatively. Given this, Hypothesis 3 is proposed, as follows:

H3: The effectuation behavior of the entrepreneur has a positive effect on income, regardless of socioeconomic condition.

Although the literature on entrepreneurship oscillates between an orientation towards opportunity (Faia et al., 2014; Kropp et al., 2006; Pittaway & Cope, 2007) or as an effect of necessity (Margolis, 2014), it is considered, in this article, that there are socioeconomic factors that determine the act of entrepreneurship. Furthermore, based on the principle of the individual’s self-determination (Krueger et al., 2000), it is argued that certain entrepreneurial behaviors can mitigate the harmful effects of ethnic prejudice, imposed on almost half of the population, both in Brazil and in Brazil. in different places in the world (Santos, 2005).

Although the literature ensures that the best financial performance is associated with the entrepreneur’s behavior, empirical studies indicate that, in entrepreneurship, the ethnicity factor greatly influences income (Bosma & Kelley, 2019; GEM, 2018). So, considering that some characteristics of entrepreneurial behavior can reflect on the business's success, the assumption of certain decision-making logic can mitigate the effects of ethnicity on the income of entrepreneurs, thus moderating the relationship between variables ethnicity and behavior of the entrepreneur.

Previous studies have shown that individuals with management skills related to planning and creating more reliable scenarios tend to have a higher income (Faia et al., 2014; Tasic & Andreassi, 2008). The entrepreneurial behavior based on causation logic, in this sense, offers better possibilities for planning and searching for the necessary resources for business development.

In the context of Afro-entrepreneurship, causation behavior can be used to plan and anticipate the problems faced in the new business (in particular, prejudice), enabling the development of actions to mitigate its effects. In this way, practices related to causation behavior provide a higher income to the entrepreneur, mitigating the impact of their ethnicity. This argument is linked to the perception that more organized management practices reduce the socioeconomic problem and allow more significant gains. Thus, Hypothesis 4a of this study is proposed, assuming that entrepreneurial behavior, in the logic of causation, potentially reduces the negative effect rooted in Brazilian society concerning the impact of the entrepreneur’s ethnicity on their income.

H4a: The causation logic of entrepreneurial behavior moderates the relationship between ethnicity and the entrepreneur’s income, mitigating the effects of prejudice by ethnicity.

Although the practices of causation logic are efficient, the set of sub-dimensions of effectuation logic can also contribute to mitigating ethnicity's effects on the entrepreneur’s income. The literature on entrepreneurship, in effectuation logic, argues that one of the practices adopted, both when starting their business (Sarasvathy, 2001) and in situations of uncertainty, faced by more experienced entrepreneurs (Welter & Kim, 2018), is the option for flexibility when changing the business model to maintain earnings (Reymen et al., 2017). In addition, mistakes are taken as learning to reconfigure businesses, showing the practice of experimentation (Chandler et al., 2011).

On the other hand, the premise that the entrepreneur, by the effectuation logic, seeks acquaintances and establishes pre- agreements before closing deals as a way of circumventing the impossibility of planning (Chandler et al., 2011; Kalinic et al., 2014).

Despite these three dimensions getting closer to intuition and improvisation (Evers & O’Gorman, 2011), the entrepreneur who follows the effectuation logic is careful with the financial limits assumed, which characterizes the under-dimension of bearable losses (Chandler et al., 2011; Fiedler et al., 2017). Therefore, in this study, it is proposed that, regardless of ethnicity, the logic of effectuation behavior can contribute to the entrepreneur’s income. In this case, the practices associated with the effectuation logic can reduce the income gap of non-white entrepreneurs, leading to Hypothesis 4b.

H4b: The effectuation logic of entrepreneurial behavior moderates the relationship between ethnicity and the entrepreneur’s income, mitigating the effects of prejudice by ethnicity.

The following (Figure 1) is a summary of the relationships proposed by this study.

Figure 1 Figure 1 Research Model Figure 1 Research Model Note: The authors (2021).

In short, first, it is proposed that: (H1) there is a relationship between the individual’s ethnicity and their income as an entrepreneur; (H2) entrepreneurs, whose behavior is aligned with planning (causation), tend to have a positive effect on their income; (H3) the use, by entrepreneurs, of effectuation characteristics, such as flexibility, experimentation, loss tolerance, and the building of relationships, capable of establishing pre- agreements, also promotes positive effects on their income; (H4a and H4b) these behaviors (causation and effectuation) can mitigate the impact of ethnic prejudice on income.

The procedures adopted to test the hypotheses are described in the next section.

The questionnaire, used for data collection, was previously evaluated by doctors and entrepreneurs, who validated the Portuguese version of the instrument, later sent via email and social networks (Facebook, Linked In, and WhatsApp) through link sharing the respondent to Google Forms.

In addition to sending, via researchers’ networks, some entities representing Afro-entrepreneurship in Brazil (who requested discretion about their data and demanded care not to reuse the list), as well as groups linked to female entrepreneurship, were invited to distribute the questionnaire to achieve a homogeneous sample.

According to Hair et al. (2005), researchers generally want to know if two or more variables are associated and if they are also related when there is covariation. The correlation coefficient is then used to assess such an association. If significant coefficients are presented during the analysis, it is an indication that there is a high covariation and a strong correlation. When two variables change together on a reliable basis, it is possible to make predictions.

Although the correlation between variables contributes to understanding the sample composition, it is necessary to apply regression analysis to verify the effects of one variable on another (Maroco, 2010). So, even if the cross-sectional data have limited power to reveal the impact, it is possible to determine associations that, interpreted and compared to other studies, infer the dependence relationships between the variables (Dancey & Reidy, 2006).

To obtain these data, identify correlations, and perform regression analysis, Excel, SPPS (version 25), and Smart PLS (version 2) were used.

The first 20 questions were grouped into five dimensions, along with their respective factorial loads. The first dimension, with seven items, refers to the causation approach, and the others, to effectuation, with 13 items, divided between experimentation (four), bearable losses (three), flexibility (four), and pre- agreements (two).

Such questions were developed by the study by Chandler et al. (2011), cited and used by Faia et al. (2014), and adapted by de Vasconcellos et al. (2019). A Likert scale - recommended for application in countries where the dominant logic is the metric (Awang et al., 2016) - of ten points was developed to obtain the answers, based on the respondent’s agreement, being | 1 | total disagreement, and | 10 | maximum agreement.

As for control variables, specific questions were included, in which respondents should identify, according to IBGE criteria, ethnicity, gender, age, level of education. Respondents should also answer about their previous experiences as entrepreneurs and how long they have been undertaking (the instrument in its entirety, the description of the five dimensions, the factorial loads, the details of the mean, the standard deviation, kurtosis, and asymmetry are detailed in Appendix A).

The sending of the questionnaire started on August 10 and ended on September 2, in the year 2018. In total, 2,721 contacts received the questionnaire and up to three emails (three waves), reaching 107 valid responses from entrepreneurs of 14 states of Brazil, which represented a response rate of 4%, denoting the degree of difficulty in data collection.

For statistical validation, initially, the existence or absence of absent or atypical values was evaluated through visual inspection of the scatter plots; after that, the following constructs reliability tests were performed - extracted variance, composite reliability, and Chronbach’s alpha (Table 1).

Table 1 Table 1 Robustness test Table 1 Robustness test

Table 1
Robustness test
	Extracted variance	Composite reliability	Chronbach’s alpha
Causation	0.621573	0.919565	0.899133
Effectuation	0.360511	0.868947	0.834668
Note: The authors (2021).

Note: The authors (2021).

The data allowed to confirm the robustness of the sample and the statistical models. Although the variance extracted from effectuation was below the acceptable limit (0.5) (Hair et al., 2005), the construct was validated, as it is a subjective variable, acceptable for a value above 0.32 (Tabachnick et al., 2001), and also for having obtained adequate indexes for the compound reliability and Cronbach’s alpha (above 0.7) (Hair et al., 2005).

In addition, asymmetry and kurtosis were evaluated, confirming the normality of the data. Values greater than | 3 | are expected to be obtained for asymmetry; and a | 10 | for kurtosis (Hair et al., 2005) (results of asymmetry and kurtosis tests available in Appendix A).

Finally, the robustness of the model was tested, and it was observed that the three models are significant at p <0.001 and are close to | 2 |, by the analysis of Durbin Watson.

In terms of gender, it was observed that the sample is uniform: the male represents 50% of the answers, and the female, 49%, while the others or who prefer not to identify their gender represent 1%. For data analysis purposes, the female gender was represented by | 1 |, and the male | 0 |.

Regarding ethnicity, of the 107 respondents, 60 declared themselves to be white (56% of the sample), and 47, black/brown (44%). Blacks and browns were grouped in the research as non- whites, in alignment with what Degler (1986) describes as a characteristic of the slave-owning heritage in Brazil, which also affects the fruits of miscegenation. For the analysis, non-whites were assembled and represented by | 1 |, and whites by | 0 |.

The collection of descriptive data is helpful to test whether socioeconomic factors should be controlled and to evidence whether they could interfere in the comparative analysis between the means. The synthesis of the sample’s descriptive data is presented below (Table 2).

Table 2 Table 2 Sample Table 2 Sample

Table 2
Sample
Variable	Min.	Max.	Mean	Std. Dev.
Age	22	63	41,2	11,16
Years of Experience	0,5	40	9,14	8,42
Business Experience	1	5	1,93	1,11
Income	1	5	2,64	1,14
Education	1	8	6,28	1,68

Note: The authors (2021).

Of the 60 entrepreneurs who identified themselves as white, 58% are men and 42%, women; and 47 non-white entrepreneurs, 40% are men and 60% are women.

Regarding the geographic distribution of the entrepreneurs, responses were obtained from 14 units of the federation, validating the national scope of the collection, which corresponds to 81% of the GDP and 79% of the population of Brazil (IBGE, 2020).

The mean age of the respondents was 41 years, making the sample homogeneous concerning age. The age difference between white and non-white men is less than half a year, while the age difference between white and non-white women is less than one year.

Income was categorized, according to IBGE criteria, into five ranges: (1) up to two minimum wages (SM); (2) above two and up to four MW; (3) above four and up to ten MW; (4) above ten and up to 20 MW; and (5) above 20 MW.

Based on the IBGE criteria, eight ranges were considered for the education variable, from incomplete elementary education to complete graduate school.

The interviewees were also asked about their experience in entrepreneurship, in terms of whether or not they had already constituted other businesses, being able to respond in ranges - from one to four businesses, and above five; and, for control, over the time, in years, they have been an entrepreneur.

Before analyzing the relationships between the independent, dependent, and moderating variables, it was checked whether there was a correlation between the control variables. Spurious data would not compromise the results (Hair et al., 2005).

Next (Table 3), there is the presentation of the data to observe the representativeness of the research, if compared to others, the limitations of the analysis. For example, in the survey, the female gender is represented by one and the male gender by zero; the negative correlation indicates that, in the sample, the respondent men are older, more experienced, in terms of years, as entrepreneurs and have an income 27% higher than that of women. The positive correlation between ethnicity - non-whites | 1 | and whites | 0 |, and gender indicates that, in this sample, most women entrepreneurs are not white.

Table 3 Table 3 Correlations Table 3 Correlations

Table 3
Correlations
	Mean	Standard deviation	1	2	3	4	5	6	7	8	9
1 Gender	0.500	0.502	1
2 Age	41.21	11.162	-0.304 (0.001)	1
3 Ethnicity	0.44	0.499	0.366 (0.000)	-0.266 (0.006)	1
4 Education	6.28	1.681	0.169 (0.082)	0.032 (0.743)	-0.171 (0.079)	1
5 Income	2.64	1.144	-0.274 (0.004)	0.293 (0.002)	-0.444 (0.000)	0.147 (0.131)	1
6 Business experience	1.93	1.113	-0.136 (0.164)	0.209 (0.031)	-0.263 (0.006)	0.052 (0.597)	0.304 (0.001)	1
7 Years as entrepreneur	9.14	8.4202	-0.446 (0.000)	0.526 (0.000)	-0.453 (0.000)	0.015 (0.877)	0.429 (0.000)	0.349 (0.000)	1
8 Causation behavior	6.711	2.054	-0.149 (0.126)	0.247 (0.010)	-0.136 (0.163)	-0.019 (0.847)	0.387 (0.000)	0.003 (0.977)	0.159 (0.101)	1
9 Effectuation behavior	6.907	1.519	-0.150 (0.123)	0.191 (0.049)	-0.184 (0.058)	-0.023 (0.817)	0.281 (0.003)	0.232 (0.016)	0.186 (0.055)	0.627 (0.000)	1
Note: The values in parentheses represent the p-value when assessing the significance of the correlations. The authors (2021).

Note: The values in parentheses represent the p-value when assessing the significance of the correlations. The authors (2021).

Still, concerning gender, there was no significant correlation with education, experience in other businesses, and characteristics of entrepreneurial behavior. Thus, initially, there is an indication that the sample, in terms of gender, does not compromise the analysis of the relationship between ethnicity and income.

The correlation analysis shows that, in the sample, the positive correlation between an experience (measured by the number of previous ventures and time as an entrepreneur) is more intense with effectuation than with causation. As they create businesses and gain experience, entrepreneurs apply less management and planning practices associated with causation behavior and take advantage of what they learn from effectuation behavior.

Among the control data considered, there is a negative correlation between age and ethnicity: non-white entrepreneurs are younger than whites. The sample size does not indicate whether this is decisive in the entire universe of entrepreneurs in Brazil, represented by the sample. However, GEM data (IBQP, 2018) is in line with this finding, pointing out that the average age of non-white entrepreneurs in Brazil is lower than that of whites, which ensures the reliability of the information collected.

There is also a correlation between age, experience in years, and previous experiences of entrepreneurship. Again, this is an expected fact, as entrepreneurs tend to start their businesses in the first years of adulthood and, with older people, they have an income higher than that obtained by the youngest, which can be justified by the continuity of their performance as an entrepreneur and for earning gains from experience.

In addition, there was a correlation between age and behavioral characteristics in the causation logic, although effectuation is also likely to improve with age but with less intensity. This finding is in line with the theory, which predicts an increase in the network of contacts of more experienced people, who learn from their mistakes, try alternatives and have a better sense of how much they can risk their business (Chandler et al., 2011).

Based on previous research (IBQP, 2018), it was possible to ensure that the sample is normal compared to data published by GEM, referring to the age and gender of entrepreneurs by need. In combination with the 2018 unemployment data (IBGE, 2020), it is possible to infer that, in recent years, non-whites are being attracted by entrepreneurship as self-employment (Margolis, 2014) due to the difficulty of repositioning themselves in the labor market. After a long period of job deficit, mainly in the non-white population. With this, the thesis of entrepreneurship by necessity is evident since many people invest in their businesses to survive.

In the sample, the education variable showed a positive correlation with gender and ethnicity, indicating that women are more highly educated than men. A negative correlation with ethnicity, pointing out that the educational background of whites is superior to that of non-whites, aligning the research with information from the Brazilian Institute of Geography and Statistics (IBGE, 2020).

Experience with new businesses and years of experience in entrepreneurship, in turn, showed a positive correlation, indicating the propensity of entrepreneurs, over time, to try new businesses. However, previous experience in other businesses related to effectuation behavior, not causation, is perceived, indicating that entrepreneurs who follow the effectuation logic tend to open new businesses. In contrast, those with higher rates of causation behavior tend to persist in the business created.

The moderating effect of the causation and effectuation behavior variables was tested to assess whether entrepreneurial behavior may be attenuating or contrasting with the differences between ethnicity and income (Table 4)

Table 4 Table 4 Regression models Table 4 Regression models

Table 4
Regression models
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Hypotheses		H1	H2	H3	H4a	H4b
Dependent variable	Income	Income	Income	Income	Income	Income
Control variables	Age Gender Education Other business experience Years as an entrepreneur	Age Gender Education Other business experience Years as an entrepreneur	Age Gender Education Other business experience Years as an entrepreneur	Age Gender Education Other business experience Years as an entrepreneur	Age Gender Education Other business experience Years as an entrepreneur	Age Gender Education Other business experience Years as an entrepreneur
Independent variables		Ethnicity	Causation	Effectuation	Ethnicity, Causation	Ethnicity, Effectuation
Moderators					Ethnicity X Causation	Ethnicity X Effectuation
β Age	0.069 (0.500)	0.068 (0.492)	-0.004 (0.966)	0.049 (0.630)	-0.005 (0.955)	0.052 (0.604)
β Education	0.155 (0.083)	0.102 (0.249)	0.158 (0.058)	0.159 (0.071)	0.102 (0.220)	0.109 (0.216)
β Gender	-0.135 (0.175)	-0.070 (0.483)	-0.110 (0.236)	-0.122 (0.214)	-0.024 (0.716)	-0.069 (0.497)
β Exp. In Other business	0.169 (0.071)	0.139 (0.128)	0.190 (0.030)	0.135 (0.147)	0.165 (0.055)	0.107 (0.246)
β Years as entrepreneur	0.272 (0.018)	0.195 (0.090)	0.261 (0.015)	0.267 (0.019)	0.208 (0.056)	0.188 (0.109)
β Ethnicity		-0.258 (0.011)			-0.234 (0.014)	-0.244 (0.016)
β Causation			0.332 (0.000)		0.303 (0.000)
β Effectuation				0.176 (0.051)		0.165 (0.073)
β Causation x Ethnicity					0.086 (0.298)
β Effectuation x Ethnicity						-0.028 (0.762)
Durbin Watson	1.981	2.069	1.974	1.990	2.064	2.059
R²	0.248	0.295	0.350	0.276	0.396	0.318
F	6.655	6.973	8.987	6.356	8.019	5.716
Variation in F		6,688	15,781	3,904	8,162	1,666
Note: The significance of variables are in parentheses, after the factorial load information. Exp. = Experiência. The authors (2021)

Note: The significance of variables are in parentheses, after the factorial load information. Exp. = Experiência. The authors (2021)

For the regression analysis, six models were tested, all significant and robustness ensured by the Durbin Watson test, indicating values close to | 2 | (Maroco, 2010), as shown in Table 4.

Initially, Model 1 was proposed, with only the control variables to verify whether similar effects and socioeconomic origin and experience could or could not influence income without considering ethnicity. Disregarding ethnicity and the way entrepreneurs behave, it was observed that age has no effect on income, but: (a) experience with previous businesses (β = 0.169, p = 0.071); (b) years of experience as an entrepreneur (β = 0.272, p = 0.018); and (c) the level of education (β = 0.155, p = 0.083).

Subsequently, Model 2 assessed whether income is determined by ethnicity (H1); therefore, the dichotomous variable (non-whites = | 1 |; whites = | 0 |) was included. Compared with Model 1, it is noteworthy that education is no longer significant, as experience with previous businesses. The ethnicity variable is significant (β = -0.258, p = 0.011), showing an inverse relationship with income, ensuring that not being white determines the lowest income for the entrepreneur, but that their experience, however, alleviates this effect, for having a positive relationship.

Therefore, Model 2 allows supporting the argument defended in H1, that socioeconomic conditions, of historical origin, determine that non-whites have less access to entrepreneurship by opportunity, being subject to start their businesses, basically, out of necessity (Kopkin, 2017; Margolis, 2014), facing the adversities inherited over time.

To separately test the effect of causation and effectuation behaviors on the income of entrepreneurs, two models were included, Model 3 and Model 4. Model 3 tested whether income is determined by causation behavior (H2), disregarding the effect of ethnicity. In comparison with Model 1, which tested only the socioeconomic and experience variables, it can be seen that there is a positive correlation with education (β = 0.158, p = 0.058), indicating that individuals who have studied longer tend to start businesses with more planning. It is also noticed that entrepreneurs with more experience, both with other ventures (β = 0.190, p = 0.030) or as time entrepreneurs (β = 0.261, p = 0.015), tend to adopt the logic of causation in their businesses and these factors contribute to a higher income. On the other hand, it appears from the model that causation behavior is not associated with the age and gender of the entrepreneur.

Therefore, Model 3 supports the argument proposed in H2, associated with socioeconomic conditions - except for ethnicity - that the causation behavior allows a higher income for the entrepreneur. This information confirms previous studies that linked the ventures’ performance to this characteristic of entrepreneurial logic because of the ability to outline the future (Welter & Kim, 2018).

Model 4 evaluated whether income can be determined by the effectuation behavior (H3), disregarding the effect of ethnicity and causation behavior. By comparing the results with Model 1, it is possible to verify that there is also a positive correlation with education (β = 0.159, p = 0.071), indicating that individuals who have studied longer possibly acquire skills related to flexibility, experimentation, and develop networks of useful relationships to start a business. It is also noticed that entrepreneurs with more experience with entrepreneurship (β = 0.267, p = 0.019) adopt the effectuation logic in their businesses. However, the relationship with the number of previous ventures is not significant, as well as concerning the age and gender of the entrepreneur.

Thus, Model 4 offers conditions to support the H3. Associated with socioeconomic factors - except for ethnicity - the effectuation behavior offers higher income conditions for the entrepreneur. This information is in line with previous studies that linked the performance of businesses to flexibility (Pacho & Mushi, 2020) and the establishment of relationship networks (Miura et al., 2019). In addition, it indicates that these skills are positively associated with education.

Then, the variables centered on entrepreneurial behavior were included to test moderation with the variable ethnicity (also centered). Model 5 tested whether the interaction between ethnicity and causation behavior could mitigate the negative effects of ethnicity on income; and the results indicated the significance of the causation logic in relation to the entrepreneur’s income (β = 0.303, p = 0.000), as well as ethnicity (β = -0.234, p = 0.014). However, it was impossible to verify the full moderation between the variables, for not showing statistical significance in the interaction term, giving partial support to H4a. From this observation, it was possible to hurt that the causation behavior does not significantly and statistically mitigate the negative effects of ethnicity on income. However, it substantially increases the explanation power of the model (from R² = 0.350 to R² = 0.396).

Finally, the causation variable was excluded, and that of entrepreneurial effectuation as a moderator was included (Model 6). The mean also centered the variables ethnicity and effectuation behavior. Similarly to the previous model, Model 6 suggests that the effectuation behavior is significant concerning the entrepreneur’s income (β = 0.165, p = 0.073), as well as the ethnicity variable (β = -0.244, p = 0.016), but without interaction between variables, as it does not reach statistical significance in the interaction term. Again, it appears that the entrepreneur’s behavior, now by the effectuation logic, does not significantly mitigate the adverse effects of ethnicity on income, despite increasing the explanation of the model (from R² = 0.276 to R² = 0.318), giving partial support to H4b. From these results, it was possible to synthesize the results (Table 5).

Table 5 Table 5 Summary of Hypotheses tests Table 5 Summary of Hypotheses tests

Table 5
Summary of Hypotheses tests
Hippthesis	Description	Hypotheses support
+H1	The entrepreneur’s ethnicity influences income; thus, the income of white entrepreneurs is higher than that of non-whites.	Supported
+H2	The entrepreneur’s causation behavior has a positive effect on income, regardless of socioeconomic status.	Supported
+H3	The effectuation behavior of the entrepreneur has a positive effect on income, regardless of socioeconomic condition.	Supported
+H4a	The causation logic of entrepreneurial behavior moderates the relationship between ethnicity and entrepreneurial income, mitigating the effects of prejudice by ethnicity.	Partially Supported
+H4b	The effectuation logic of entrepreneurial behavior moderates the relationship between ethnicity and the entrepreneur’s income, mitigating the effects of prejudice by ethnicity.	Partially Supported
Note: The authors (2021).

Note: The authors (2021).

The results indicate that both ethnicity and characteristics of entrepreneurial behavior, causation, and effectuation, are related to the entrepreneur’s income. However, it is not possible to ensure that the practices adopted in these behaviors can fully mitigate the effect of prejudice by ethnicity on the entrepreneur’s income because, despite being statistically not validated, the effect of the interaction between the entrepreneurial behavior variables, the proposed models increase their explanation significantly, considering the interpretation of alteration of the R². Thus, in Model 2, the explanation was that the variables included up to ethnicity could explain up to 29.5% of the variation in income. When including the causation behavior, this explanation increased to 39.6%; and the effectuation behavior to 31.8%, leading to the understanding that income can be increased if the decision-making practices, aligned with the behavioral logic of entrepreneurship, are intensified.