The practice of reporting suspected irregularities aims to protect organizations and enable the market to become more sustainable and transparent among the market agents and society, and may therefore be considered a Corporate Social Responsibility practice (Eterović et al., 2015). Reports of suspected irregularities can occur on a variety of topics, from money laundering and cartel crimes to practices of environmental contamination, exposure of employees to chemical agents, falsification of products, and the use of damaged or prohibited inputs. These types of misconduct occur for the benefit of persons or organizations that aim to obtain economic advantages or that accidentally leave their control mechanisms precarious, and may even be caused by employees to the detriment of their organizations. The costs of repairs and penalties can sometimes result in major harm to the company, and could even lead to bankruptcy.

Whistleblowing is the main organizational mechanism for identifying fraud and other illicit acts. The sooner these acts are reported, the less damage to the organization they generate (ACFE, 2022). Reporting suspicious and illegal acts can contribute to the achievement of the Sustainable Development Goals (SDGs), such as Goal 8, which highlights economic growth by mitigating risks and financial losses, both directly and indirectly, as well as Goal 16.5, which deals with reducing corruption and bribery in all its forms (Lubas et al., 2021).

There are different ways to approach the underlying reasons for behavior depending on the geographical location where it takes place, such as in the USA (Lee et al., 2021), Africa (de Maria, 2005; Owusu et al., 2020; Soni et al., 2015), Europe (Bogdanovic & Tyll, 2016; Oelrich & Erlebach, 2021), Asia (Mansor et al., 2020; Park et al., 2008; Tudu, 2021; Zhang et al., 2017), and Latin America and the Caribbean (Alleyne et al., 2016; Maragno, 2019), which has the highest proportional value of fraud losses when compared to GDP per capita (ACFE, 2022). In this region, there exists a large network of corrupt companies and government officials, reaching as far as Argentina, Bolivia, Brazil, Colombia, Cuba, Ecuador, Guatemala, Mexico, Panama, Peru, the Dominican Republic, and Venezuela (Maragno & Borba, 2019; Sallaberry et al., 2020).

The whistleblowing literature does not address the differences in the behavior of employees in small and medium-sized enterprises (SMEs), even though they represent the majority of market organizations (Bikefe et al., 2020). Nonetheless, other variables in the literature can contribute towards understanding what might happen in these small companies. The variable representing the size of the organization was not significant in the internal channel. However, it was negative for the reports in external channels. In other words, professionals from smaller companies are more likely to whistleblow (Alleyne et al., 2018). The organizational support variable can be related to the size of the company and was also not significant for the internal channels, and was negative for the reports in external channels (Alleyne et al., 2016). Presenting another perspective, Reichelt and Wang (2010) showed that SMEs, among the comparative advantages, possess the highest degree of knowledge about their customers and their business.

Faced with this divergence and seeking a better understanding of whistleblowing in small and medium-sized accounting firms, the research aims to analyze the relationships between attitudinal, normative, and control beliefs on the whistleblowing of accounting professionals, distinguishing the relationships between small and large accounting firms in light of the Theory of Planned Behavior (TPB). By understanding how the beliefs of accounting professionals from small and large companies influence their behavior, the results may contribute to the development of controls, strategies, and supports that could promote more effective reporting of suspicious events, particularly in SMEs. Losses, damages, and fraud would therefore be identified earlier, reducing economic losses for accounting firms and their clients. In addition, the clients of small accounting firms are more likely to also be SMEs, and a larger governance in SME accounting has a greater potential impact on the business environment (Chan et al., 2022).

Hence, a survey was conducted in order to identify whistleblowing beliefs and behavior, and, consequently, to analyze all the indicators through structural equations, using the partial least squares (PLS-SEM) method. The article is composed of this introduction, followed by the theoretical development and a section on the methodological procedures, which includes the indicative of the sample of respondents, the construction of the collection instruments, and the analysis technique. The second part of the article consists of the data analysis, the validation and discussion of the model and its structural relationships, and the interpretation of the results, ending with the final considerations.

Whistleblowing behavior can be defined as the disclosure by organization members (former or current) of illegal, immoral, and illegitimate practices under the control of their employers to parties and organizations that may be able to effect action (Near & Miceli, 1985), revealing unauthorized information that provides evidence of the violation (Vinten, 1992), of which the whistleblower is neither a direct participant nor a direct victim (Gottschalk & Asting, 2022). Whistleblowing is an action based on a highly complex psychological process and depends on the trust or the belief that wrongdoings will be corrected (Gundlach et al., 2003; Soni et al., 2015). It is considered a positive form of behavior and is encouraged in the workplace (Park & Blenkinsopp, 2009). To modify behavior, interventions can be directed towards the individual's beliefs when these have control over their behavior (Ajzen, 2016).

The application of behaviorist theories has proved itself as a useful theoretical lens for predicting ethical and unethical behavior (Park & Blenkinsopp, 2009), such as dishonest actions (Beck & Ajzen, 1991), the unauthorized copying of software (M. K. Chang, 1998), the intention to report irregularities in the medical profession (Randall & Gibson, 1991), in consumer behavior (Fukukawa, 2002), in the intention to pay taxes (Bobek & Hatfield, 2003), in the intention of drivers to commit traffic violations (Parker et al., 1992), and behavior concerning waste (Teo & Loosemore, 2001).

The subject of whistleblowing does not have an applicable general theory, which was initially considered a significant problem (Miceli & Near, 1988), in both a theoretical and a practical sense (Park & Blenkinsopp, 2009). Over the last decade, the TPB has established itself as the most widely applied theory about the relationships between attitudes, intentions, and behaviors, including whistleblowing (Latan et al., 2018). Due to the need for regulators to encourage whistleblowing and to the difficulties with directly studying such behavior, researchers had previously relied on indirect measures: first they focused on the attitudes, and then on the intentions, as a substitute measure of whistleblowing behavior (Park & Blenkinsopp, 2009).

Behaviorist theories about whistleblowers’ beliefs arrived from the field of the social sciences. Their origins and references are derived from the work of Gustave Le Bon on the psychology of crowds and from Freud on group psychology. Later, Kurt Lewin, at the Massachusetts Institute of Technology, founded the American School of Social Psychology in 1933, where the theory of rational action (TRA) was developed in the 1960s, as well as, later on, the theory of planned behavior (TPB). The theoretical framework assumes that people tend to behave rationally and to systematically use the information made available to them when deciding whether to act or not (Ajzen & Fishbein, 1980; Fishbein & Ajzen, 1975). In this context, behavioral intentions are determined by the attitude towards behavior and the subjective norm, still considered the best intention predictor (Ajzen & Kruglanski, 2019).

Behavioral beliefs attach the behavior of interests to the probable outcomes of the behavior, and also to the assessments of these outcomes (Y. Chang et al., 2017; Park & Blenkinsopp, 2009). The attitude in relation to a behavior represents the degree to which the behavior performance is valued positively or negatively (Ajzen, 1991). The attitude is considered to the extent in which it agrees or disagrees with the result of the behavior, and whether this action causes an adverse effect, or whether the individual will be reluctant to report it (Tarjo et al., 2019).

Normative beliefs represent the behavioral expectations perceived by the individual from social referents, who are people considered to have a relevant opinion, such as bosses, teachers, family members, etc. (Ajzen, 1991). This normative belief considers the force corresponding to the weighting of how much these social referents' opinions are important for the individual when deciding how to act (Ajzen, 1991; Mesmer-Magnus & Viswesvaran, 2005). Being accepted by your referents is an important goal (Tarjo et al., 2019). Thus, if the accounting professional considers that his social referents believe that suspicious illicit acts must be reported, the individual may consider the relevance of the referent's opinion. If the weighting is superior, the influence considered will be the denunciation, otherwise the influence will lead to the individual choosing to remain silent (Park & Blenkinsopp, 2009; Trongmateerut & Sweeney, 2013).

In the subsequent decades of the 1980s and 1990s, the theoretical platform had evolved by incorporating control beliefs into a new theory, the TPB (Ajzen, 1991), which improved the predictive capacity of the whistleblower’s intention (Cheng & Lam, 2008; Tudu, 2021). These control beliefs, dimensioned by their capability and strength, refer to the difficulty, the facility, or even the impediment perceived by the individual for behaving in a certain way (Ajzen & Kruglanski, 2019), and are related to the perceived presence of certain factors that may facilitate or restrain the performance of a behavior.

These control elements derive from negative or positive beliefs about reporting. Among the negative ones are the impossibility of correcting the wrongdoing and the concern about retaliation, which can be considered as organizational obstacles, including other antecedents such as past experiences, third-party information on hindering or ignoring reports, difficulties in presenting the complaint, and the organizational impossibility of correcting mistakes (Ajzen, 1991; Mesmer-Magnus & Viswesvaran, 2005; Miceli & Near, 1992; Park & Blenkinsopp, 2009). Positive beliefs include organizational support (Latan et al., 2018; Mustafida, 2020), rewards (Maragno, 2019), whistleblower protection, and company support (Y. Chang et al., 2017; Gorta & Forell, 1995).

Control beliefs are added to attitudinal and normative beliefs to explain a particular behavior. The TPB later incorporated real control over behavior (Ajzen, 2002), which refers to the individuals' perceptions of their real capacity and possibility in performing a certain behavior. The intention proxy is an indicator of the promptitude of performing a certain behavior (Park et al., 2008). The intention is considered an immediate antecedent of behavior, and, to the extent that actual behavioral control exists, could serve as a proxy and contribute to the prediction of the behavior (Fishbein & Ajzen, 2011).

The analysis of the data was performed using the partial least squares structural equation modeling (PLS-SEM) method. The application of structural equations is understood to be more appropriate when there are interdependencies or simultaneous causes on the observed response variables, and when there exist important unobserved or omitted explanatory variables (Jöreskog & Sörbom, 1982).

At the same time, the technique allows to estimate a series of separate, but interdependent, multiple regression equations by specifying the structural model (Dijkstra, 2010; Hair et al., 2017; Ringle et al., 2015). The choice of the PLS-SEM method instead of one based on covariance (CB-SEM), stems from the aim of the research (structure or prediction), whose focus consists in predicting behavior instead of simply studying the structural equation models with observable variables (Jöreskog & Wold, 1982; Ringle et al., 2015). In accordance with Hair et al. (2017), PLS-SEM models are preferable in predicting driver constructs, exploratory research, and the extension of an existing structural theory.

By analyzing whistleblowing behavior in Brazilian SME accounting firms, it was possible to identify that employees cognitively process intention and effective behaviors in different ways. Large companies positively supported their reports based on the behavioral results and the respondents’ gender characteristics, although, in general, the intention to report did not reflect on effective behavior, which denotes the impulsive and intuitive aspect of reporting suspicious events. In SMEs, beliefs that affect reporting intentions and behaviors were demonstrated to be different from those that influence individuals in large companies.

For intention and whistleblowing behavior, the perception of attitudinal beliefs is indifferent in SMEs, which differs from the results found by Alleyne et al. (2018), Brown et al. (2016), Chwolka and Oelrich (2020), Dalan et al. (2019), Kashanipour et al. (2020), Lee et al. (2021), May-Amy et al. (2020), and Park and Blenkinsopp (2009). These findings demonstrate the absence of clear and direct links between the perception of attitudes and their results, denoting the possibility of automated or intuitive (Courtois & Gendron, 2020), but not utilitarian, behavior.

The subjective norm, in its descriptive form, showed a positive relationship with the intention to report, as was contextually evidenced by Dalan et al. (2019), Kashanipour et al. (2020), Lee et al. (2021), May-Amy et al. (2020), and others. This demonstrates that, in SMEs, individuals follow what others are doing, and not just with words. Although this affected the intention to report, it did not affect the final behavior based on past experiences. Moreover, the other subjective beliefs proved to be indifferent to intention and behavior.

The control beliefs, which would be the evolutionary differential of the TPB, were indifferent in influencing reporting behavior in small and large companies, and showed a negative relationship with the individual's intention in SMEs. Thus, the existence of mechanisms that facilitate reporting can mitigate the intention to report, which may be due to a greater empowerment combined with a greater acceptance or appetite for business risks, similar to Chwolka and Oelrich (2020), but divergent from most of the research (Anggraini & Siswanto, 2016; Brown et al., 2016; Cheng & Lam, 2008; Kashanipour et al., 2020; among others).

The research leaves several gaps that can be expanded on and explored in the whistleblowing and SME literature. The evidence obtained contributes to investigating an unexplored context in the SME literature, which, in the accounting segment, depends on a greater participation in reporting illicit acts (Sallaberry & Flach, 2021). The intention to report presented by the collaborators showed the absence of traditional influences and a divergence in the controls, revealing that intentions are sustained from more creative and intuitive cognitive processes, and from the attitudes that need to be supported by the organization for its protection, even in large companies with robust systems.

The absence of influence among the attitudinal beliefs reinforces the perception of integrity in relation to the possibility of corruption, along with the importance of whistleblowing for the preservation of patrimony. The influence of descriptive beliefs demonstrates a possible insensitivity on the training and promotion of explanatory information, but which can be reinforced by the example of other accounting SMEs. In addition, it contributes to the theory by demonstrating how the whistleblowers' relationships can be perceived in the light of the TPB in SME accounting firms.

Family management, as a determinant of control, can influence the omissions on reporting, since it stems from ties that transcend the work environment. These ambiguities are also perceived in the gender of the employee. In the sample, men showed less intention to report, as did the women who assume the role of promoting justice or the investigation of facts in SME accounting offices, since they usually face several challenges in the work environment. In larger companies, this perception turns out to be the opposite: men are the biggest contributors to whistleblowing, since it is a more hierarchical environment and has formalized structures.

These results allow a better understanding of the behavior of SME accounting professionals in regard to the practice of whistleblowing, which can reduce financial losses and organizational risks. This evidence contributes to professional practices and demonstrates the need for SME offices to invest in the formalization and structuration of control systems, raising awareness of its usefulness and importance in protecting the organizations, which can also be developed and promoted by professionals and governmental entities. Thus, these instruments can accomplish the proposed intention and allow the actual behavior to be more stable and planned. The evidence can help the accounting regulators to establish communication channels aimed at SMEs, which would reinforce whistleblowing among accounting professionals. The non-significant relationships demonstrate the gaps for professionals that must be clarified, especially regarding organizational benefits and the mechanisms and tools that facilitate the reporting.

The research has a limitation that derives from the survey’s own nature, since it endorses its findings and conclusions in standardized and quantified answers, without retaining the feelings that transcend each individual. The criterion for selecting small accounting offices might be another limitation considering that the presence of only a few employees does not always reflect the structure, organization, and capital of a company, although we used a criterion already mentioned in the literature. New qualitative research is suggested to identify the perceptions and feelings behind these statistical relationships, along with the application of the survey in other cultural environments. This would enable the verification of which relationships are innate in the context of the accounting business, and which emerge from social, cultural, and ethnic influences.

The authors would like to thank the participants of the IV Congreso Iberoamericano de Investigación sobre MIPYME for their criticisms and suggestions, and to the translator Ms. Amanda B. Soczek for her invaluable support in the translation process.