Effect of the usage of data analytics in audits and the professional scepticism of the auditor

(1)

1

Effect of the usage of data analytics in audits and the

professional scepticism of the auditor

Name: Mahmoud Diouri Student number: 11362855

Thesis supervisor: Dr. E.E.O. Roos Lindgreen Date: 24-05-2018

Word count: 14917

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

2 Statement of Originality

This document is written by the student Mahmoud Diouri who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

The professional scepticism of auditors obtained attention in the media caused by audit scandals like Enron and WorldCom (Tran, 2002). Today these scandals are widely debated. From the investigation of fraud-related restatements by the U.S. Securities and Exchange Commission (SEC), it was concluded that auditors failed to find material misstatements because of an inadequacy of sufficient professional scepticism at the time of the audit. Other articles (Noviyanti & Winata, 2015; Hurtt, 2010) suggest that there are multiple variables that influence the professional scepticism.

A possible explanation, which could lead to the lack of professional scepticism could be the change in the audit profession by the usage of data analytics performed in the audit. The findings described in this research strongly suggest that there is a positive association between the usage of data analytics and the professional scepticism of the auditor. This suggests that auditors are not certain about themselves and the data when the audit is performing using data analytics. It seems that less tangible data makes the auditors less certain about the statements they could make. Because of their higher scores, it would seem that auditors would do more research in their audit when using data analytics as a method.

The research described in this paper is believed to contribute to the accounting profession and academic literature in the following ways. The auditors in the population surveyed show a natural sceptical attitude. When using data-analytics, the respondents become even more sceptic and may even spend more time in the audit. The contribution of this thesis is that it identifies a new determinant of professional scepticism, namely the usage of data analytics. Keywords: Scepticism, professional scepticism, data analysis, data analytics, classical analysis, classical analytics, experiment

(4)

4

1. Introduction

1.1 Background

1.1.1 Changes in financial profession by technological developments

A report written by Verbeeten, Kolthof and Steenwijk (2017), which was recently distributed by the VU of Amsterdam and the Dutch Association of Accountants, also known as the ‘NBA’ in Dutch, states financial professionals underestimate the impact of technology in their work field. Before, a focus on increasing efficiency of transactions was sufficient. However, in modern days these professionals should also focus on the development of robotics and data analytics. The research states that many organizations are already investing lots of time to standardize financial processes. If financial professionals want their job to survive, they should also invest time in technical and personal competences.

There are five reasons why financial professionals should develop their skills regarding technological developments (Verbeeten et al., 2017). First, robotics and data analytics are important in all duties of the financial professionals. Second, more organizations are compliant with the constraints regarding these developments. Third, efficiency and effectivity rise because of robotics and data analytics. Organizations that implement robotics have lower costs for the financial function and organizations that implement data analytics are more effective in their financial function. Fourth, the majority of the financial professionals underestimate the impact of the technological developments. And last, competences which are important for the financial function in the digital world need to be developed quickly. For example, data analyses and leadership must develop further and faster.

1.1.2 Professional scepticism

There are multiple definitions for professional scepticism according to relevant literature. Nelson (2009) defines professional scepticism “as indicated by auditor judgements and decisions that reflect a heightened assessment of the risk that an assertion is incorrect, conditional on the information available to the auditor”. The American Institute of Certified Public Accountants, also known as AICPA, defines professional scepticism as “an attitude that includes a questioning mind and a critical assessment of audit evidence” (AICPA, 2002). Hayes et al. (2014) defines professional scepticism as “an attitude that includes a questioning mind, being alert to conditions which may indicate possible misstatements due to error or fraud, and a critical assessment of evidence”.

Hurtt (2010) mentions in her article that professional scepticism is a multidimensional individual-characteristic. Professional scepticism can be a trait (a relatively stable, enduring aspect of an individual) and it can also be a state (a temporary condition that is awakened by situational variables). Hurtt also posits that there are six characteristics that determine the degree of

(7)

7 professional scepticism: a questioning mind, search for knowledge, suspension of judgement, self-determining, interpersonal understanding and self-confidence.

The professional scepticism of auditors has gained more attention in the media as a result of the audit failures that were highly visible in the past years, which caused accounting scandals like Enron and WorldCom (Tran, 2002). From the analysis of fraud-related restatements by the U.S. Securities and Exchange Commission (SEC), it has come to conclusion that auditors failed to detect material misstatements due to a lack of sufficient professional scepticism when the audit was performed.

A possible variable which could lead to the lack of professional scepticism may be the change in the audit profession by the usage of data analytics in performing the audit. The subject investigated in this thesis is whether the usage of data analytics in audits influences the professional scepticism of an auditor.

1.2 Research Question

As stated in section 1.3 (next paragraph), there is limited research regarding factors that may affect the professional scepticism of an auditor. Therefore, the following research question will be examined:

How does the usage of data analytics in performing the audit affect the professional scepticism of an auditor?

1.3 Contribution and relevance

According to Hurtt (2010), there has been limited research about professional scepticism and what could influence this particular phenomenon. Hurtt states that professional scepticism is an important concept in academic literature. The purpose of this thesis is to investigate whether data analytics as a variable has an effect on the professional scepticism of an auditor. Prior literature focuses more on the effect of auditor independence on the quality of an audit. This thesis examines one specific auditor capacity, namely the professional scepticism of the auditor.

As mentioned earlier, the literature on professional scepticism is rather limited. Noviyanti and Winata (2015) have examined the influence of sceptical attitude, tone at the top, and knowledge of fraud on auditors’ professional scepticism. The contribution of this thesis is adding another possible determinant of professional scepticism, namely the usage of data analytics.

(8)

8 Hopefully this thesis will also be a contribution to society. Auditors are the stewards of the capital market; their reason for existence is to give assurance to stakeholders, in order to facilitate the appropriate operation of these markets. In order to fulfil this role, the auditor must possess the important capability of professional scepticism. This research contributes by giving insight on how professional scepticism is influenced.

1.4 Structure of the thesis

The second chapter contains the literature review and theory. In the third chapter, the hypothesis development is discussed. In the fourth chapter, the research method and design will be reflected upon. Following the empirical results of the experiment. This will be debated in the fifth chapter. The thesis ends with the sixth chapter, which contains the discussion and conclusion of this research.

(9)

9

2. Literature review and theory

2.1 Role of the auditor

In every firm, stakeholders have different objectives. Management may prefer higher salaries and benefits (expenses for the firm), while investors want higher returns and dividends. Investors and creditors are depending on fair reporting of financial statements. Auditors play a role here to give all stakeholders confidence on the fairness of these reports with an independent and expert opinion, called an audit opinion (Hayes et al., 2014).

Hayes et al. (2014) state that the function of auditors is to give credibility to the financial statements. It is the responsibility of management to develop these statements, while it is the auditor’s responsibility to lend them credibility. With this audit process, the auditors enhance the usefulness and value of financial statements. The auditor also increases the credibility of other non-audited information that was given by management.

Hayes et al. (2014) mention that there are two different types of auditors: independent external auditors and internal auditors. Internal auditors are employed by organisations to “investigate and appraise the effectiveness of company operations for management” (Hayes et al., 2014). They mostly work on appraising their internal controls. Internal auditors have two primary effects on a financial statement audit: First, their existence and work can influence the nature, timing and extent of audit procedures. Second and last, the external auditors may use them, them being the internal auditors, to aid in performing the audit. The external auditor must test the competence of the internal auditor and his objectivity. Independent auditors have a primary responsibility over the performance of audit functions regarding the published financial statements of publicly traded and non-public companies.

According to Huibers (2013), the role of an auditor can be divided into three parts. Those parts are (1) assurance, (2) consulting and (3) participative roles. He also states that there are three types of audit that can be classified. These types are (1) audits of financial statements, (2) operational audits and (3) compliance audits. To have reasonable assurance that financial statements and other financial reporting don’t have material misstatements, the auditor needs to collect appropriate audit evidence (Hayes et al., 2014).

According to Hayes et al. (2014), there are two types of assurance engagement under the International Framework for Assurance Engagements, namely a reasonable assurance engagement and a limited assurance engagement. The objective of a reasonable assurance engagement is a cutback in assurance engagement risk to a low level, which is acceptable as the basis for a form of expression of the conclusion of a practitioner which is positive. The objective of a limited assurance engagement is a cutback in assurance engagement risk which is acceptable in the

(10)

10 circumstances of the engagement. However, the risk will be greater than for a reasonable assurance engagement, as the basis for a negative form of expression of the conclusion of a practitioner.

The provision of assurance is important because of the following two reasons: The first reason is readers of the financial reports put an appreciable certainty on the quality of the audited financial statements. The second reason is the assurance that an auditor provides a certain security role for the principal (e.g. shareholders) (Watts & Zimmerman, 1986).

Hayes et al. (2014) state that opinions expressed in the auditor’s report may be one of four types: unmodified, qualified, adverse or disclaimer of opinion. To form an unmodified opinion, the auditor shall conclude as to whether he has obtained reasonable assurance about whether the financial statements are free from material misstatement, even if due to fraud or error.

According to Hayes et al. (2014), an auditor will give a qualified opinion in two different situations. The first situation when auditors give a qualified opinion is when they get sufficient appropriate audit evidence and the misstatements, individual or aggregate, are material in the financial statements. The second situation is when the auditor is unable to obtain sufficient appropriate audit evidence on which they base their opinion, but the auditors conclude that the effects on financial statements could be material but not pervasive.

An auditor will give an adverse opinion when an auditor, having obtained sufficient appropriate audit evidence, claims that misstatements, individual or in the aggregate, are both material and pervasive in the material misstatements (Hayes et al., 2014). Such an opinion is given when the effect of a disagreement is so material and pervasive, that an auditor will state that a qualification of his report is not enough to disclose the misleading or incomplete nature of the financial statements.

The last possible opinion an auditor can give is the disclaimer of opinion. Hayes et al. (2014) state that auditors are unable to obtain sufficient appropriate audit evidence on which to base the opinion. The auditor will state that the possible effects on the financial statements could be both material and pervasive.

The ethical guidance is set out by International Ethics Standards Board for Accountants (IESBA). It reports its recommendations to the IFAC Board after research and appropriate exposure of draft guidance. The guidance is incorporated into the Handbook of the Code of Ethics for Professional Accountants (the Code). The objective of the Code is to serve as a model on which national ethical guidance is based. It sets standards of conduct for professional accountants and

(11)

11 states fundamental principles that should be observed by professional accountants to achieve common objectives (Hayes et al., 2014).

2.2 Professional scepticism

Scepticism is an interesting phenomenon within the psychology literature. Clarke (1972) defines a sceptic as someone who doubts a lot about the information that he or she gets. He states that everyone has a meaning for scepticism. Each person has its own way of defining the context of the received information. He also states that scepticism frees the person from old problems, because a sceptic always sees a newer and more challenging problem that the person obtains from the context and information (Clarke, 1972). It leaves a certain “non-objectivity”, meaning that everyone has its own opinion about the data that is gathered. A question that comes to mind from this passage is: “How does this phenomenon affect the reliability and trustworthiness of a person where his or her opinion matters the most?”

Hurtt (2010) defines professional scepticism in six characteristics of scepticism, namely: “the questioning mind, suspension of judgment, search for knowledge, interpersonal understanding, self-esteem and autonomy”. While McMillan and White (1993) defines professional scepticism as “conservatism bias in audit judgments, referred to as a hypothesis frame”. Furthermore, Shaub (1996) defines professional scepticism as “one who instinctively or habitually doubts, question or disagrees with assertion or generally accepted conclusions’’. Nelson’s (2009) defines professional scepticism as “a heightened assessment of the risk that an assertion is incorrect conditional on the information available to the auditor’’.

Nelson (2009) reviewed the definitions of professional scepticism and during this review he detected the definition of Hurtt to have a neutral perspective. He finds that the definition of Shaub has a certain presumptive doubt interpretation, which means that he was likely uncertain about his way of defining the term ‘professional scepticism’. There is this concept common in all the definitions that professional scepticism is a certain attitude. The auditor must have a questioning mind, a critical assessment of the audit evidence.

According to Hurtt (2010), there are six characteristics of professional scepticism. The first three characteristics relate to the way an auditor examines evidence. All three characteristics show a willingness to search for a fully examine sufficient evidence before deciding. These three determinants are: a questioning mind, suspension of judgement and search for knowledge. The fourth characteristic, interpersonal understanding, analyses the need to also consider the human aspects of an audit when evaluating evidence. As is indicated in SAS No. 99’s (AICPA 1997b) instruction, “the auditor may identify events or conditions that indicate incentives/pressures [on individuals] to perpetrate fraud, opportunities [for individuals] to carry out the fraud, or attitudes/rationalizations [used by individuals] to justify a fraudulent action” (AU 316.31; PCAOB 2006b). The last two characteristics are

(12)

12 self-esteem and autonomy. This focuses on the capacity of the individual to respond on the obtained information (Hurtt, 2010).

Hurtt (2010) has developed a measurement scale in her study, with 30 questions to determine the level of scepticism. She states that professional scepticism can be both a trait and also a state. This scale is designed to ex-ante measure an individual’s level of trait professional scepticism based on characteristics derived from audit standards, psychology, philosophy and consumer behaviour research (Hurtt, 2010).

However, Nelson (2009, p. 11) mentions also the so-called Hurtt-scale in his article. He discusses the fact that the Hurtt-scale does not correlate with the “Wrightman’s Trustworthy or Independence scales”. This scale was used in the research of Shaub (1996), but it has failed in successfully measuring professional scepticism. Nelson (2009, p. 11) states that the Hurtt-scale is relatively stable over time, but there is some bias for the score to decrease when using this instrument a second time and for individual components to change significantly. Then again, Nelson (2009, p. 11) states that two other studies provide evidence that the Hurtt-scale can predict the sceptical behaviour of auditors, namely the research of Fullerton and Durtschi (2005) and Hurtt et al. (2008) (in Nelson, 2009, p. 11).

In the paper of Hurtt (2010), she states that an individual auditor’s professional scepticism is the foundation of the auditing profession. The importance of scepticism of an auditor is recognized from the earliest codes were written on professional standards till today. For example, Hurtt (2010) states that in SAS No. 1 (American Institute of Certified Public Accountants [AICPA] 1997b), they mandate an auditor’s use of professional scepticism, stating “Due professional care requires the auditor to exercise professional skepticism” (AU 230.07). The Public Company Accounting Oversight Board (PCAOB) has also stated the importance of professional scepticism both in the standards (e.g., PCAOB 2007, AS No. 5, ¶4) and in its most recent Report on the PCAOB’s Inspections of Domestic Annually Inspected Firms (PCAOB 2008).

2.2.1 Determinants of professional scepticism

As earlier stated in paragraph 2.2, there are six characteristics of professional scepticism. The first three characteristics relate to the way an auditor examines evidence. All three characteristics show a willingness to search for a fully examine sufficient evidence before deciding. These three determinants are a questioning mind, suspension of judgement and search for knowledge (Hurtt, 2010).

Hurtt (2010, p. 152) states in his article that professional scepticism needs an ongoing questioning whether the information obtained from an audit suggests that a material misstatement due to fraud has occurred. Fogelin (1994, 3) (in Hurtt, 2010) defines a philosophical

(13)

13 sceptic as someone who “calls things into question”. The questioning in mind indicates that this doubt initiates action and leads to the construction of belief.

The suspension of judgement is a characteristic of withholding judgement until there is an appropriate level of evens where the auditor can base his conclusions on (Hurtt, 2010). In AU 230.9 (in Hurtt, 2010) it is mentioned as “The auditor should not be satisfied with less than persuasive evidence.”. There has to be a will for a definite answer in the topic of discussion. This answer needs to be opposed to confusion and ambiguity. Bunge (1001, 131) (in Hurtt, 2010, p. 153) states that the sceptic should not accept naively the first things they think. They should be critical and need to detect answers before believing.

The search for knowledge is different from the characteristic “questioning mind”, because a questioning mind has some sense of doubt, while the search for knowledge is more a general curiosity or interest (Hurtt, 2010). Hurtt states again in his article that sceptics are people who search knowledge for the sake of knowledge. She states again that this scepticism could encourage an auditor’s desire to investigate. This person is prepared to search and evaluate new arguments in affiliation to any question.

As earlier stated in this paragraph, the three characteristics are associated with how an auditor evaluates evidence. The fourth characteristic, interpersonal understanding, identifies the need to also consider the human aspects of an audit when evaluating evidence. As is indicated in SAS No. 99’s (AICPA 1997b) instruction, “the auditor may identify events or conditions that indicate incentives/pressures [on individuals] to perpetrate fraud, opportunities [for individuals] to carry out the fraud, or attitudes/rationalizations [used by individuals] to justify a fraudulent action” (AU 316.31; PCAOB 2006b). Hurtt (2010, p. 154) states in his article that philosophers suggest that by understanding people a sceptic can accept and recognize the fact that different individuals may have a different perception on the same subject. If an individual’s perception is identified and understood, then an auditor will have a certain basis for asserting and improving mistaken assumptions.

The last two characteristics, self-esteem and autonomy, address the ability of the individual to act on the obtained information (Hurtt, 2010). Autonomy is a characteristic where an auditor decides the degree of confirmation necessary to accept a particular hypothesis. Mautz and Sharaf (1961, 35) (in Hurtt, 2010, p. 154) states that “the auditor must have the professional courage not only to critically examine and perhaps discard the proposals of others, but to submit his own inventions to the same kind of detached and searching evaluations.”. A sceptical auditor should be certain of his own claims and less influenced by beliefs or persuasion attempts of others.

Additionally, Hurtt (2010, p. 155) states in his article that in psychology research self-esteem is characterized as a feeling of dignity and confidence in one’s own abilities. Hurtt further

(14)

14 states that those who are low in self-esteem have a deficiency in determination to rely on their own judgements. This suggests that self-esteem is mentioned in order to challenge convincing attempts rather than simply accepting what is presented. Sceptics need to have a level of self-esteem that allows them to value their own judgement at least as greatly as those of others (Hurtt, 2010, p. 155).

Nelson (2009) mentions incentives, traits and knowledge as effects of professional scepticism. For an auditor to exercise professional scepticism, he or she needs to understand the directional implications of the evidence. The auditor needs to have “knowledge of directional relations between evidence and audit risks”, where auditors possess knowledge that enables them to modify assessed risks in response to the client characteristics (e.g. management integrity, competence and turnover) (Nelson, 2009, p. 7). Auditors also need to have “knowledge of the frequencies of errors and non-errors and the patterns of evidence that suggest a heightened risk of misstatement”. Auditors need relatively accurate knowledge of error causes and effects (Nelson, 2009, p. 7).

Additionally, Glover and Prawitt (2013) states that there are developments that could affect the professional skepticism, namely “(1) professional licensing and continuing education requirements, (2) supervision, mentoring, review and inspection of work and performance evaluations has to be implemented, (3) effective planning and audit programs, (4) performance metrics that benefits auditors for their high quality performance, (5) stringent recruitment requirement, (6) effective engagement partner and leadership messaging and (7) training on core competencies and professional judgment”.

However, Glover and Prawitt (2013) discuss that there is a considerable amount of possible threats to the professional scepticism of auditors. They made a separation for four engagement treats, namely “(1) judgment traps and biases, lack of knowledge and expectations, (2) deadline pressure, inherited preferences and expectations, (3) auditor character, and personal and cultural attributes and (4) performance and compensation metrics and incentives that do appropriately encourage professional scepticism”.

Nelson (2009) mentions in his article that traits are non-knowledge characteristics that can affect the auditor’s professional scepticism. He states that the traits can be divided into problem-solving ability, ethics/moral reasoning and scepticism scales. Problem-solving ability has a basic idea where intelligence of auditors will help identifying potential misstatements. Ethics/moral reasoning identifies the extent to which auditors’ professional scepticism-related judgements and actions are affected by the incentives.

Eventually, Nelson (2009) mentions the scepticism scales, where he discusses the research attempts of multiple researchers in measuring the professional scepticism of auditors. The Hurtt-scale (and with that the six characteristics) is also mentioned in this discussion (see paragraph 2.2).

(15)

15 2.3 Usage of data analytics

2.3.1 Rise of big data and data analytics

Wang and Cuthbertson (2014) identify data analytics as a process from which someone identifies multiple observations from operational, financial and other forms of electronic data. This can be obtained internally and externally in the organization. This concept of big data and the related approaches to analysing data, often referred to as data analytics or predictive analytics, have been thoroughly discussed in the press and academic journals. University programs developed courses to address data analytics competencies (Earley, 2015).

Earley (2015) states that at the American Accounting Association (AAA) annual meeting in August 2014, a discussion, co-sponsored by PWC and the University of Illinois, was organised to discuss how the accounting studies must change to incorporate more data analysis courses. The message of the discussion was: for students to be competitive in both audit and tax, they also have to learn to become data scientists. Big data is seen as a leading part of business in the future. Any organization that falls behind in its development of data analytics capabilities is expected to perform less than its competitors and could experience bad consequences.

In his paper Earley (2015) mentions that big data in accounting literature is defined by the types of analysis that can be performed with the data, such as data analytics or predictive analytics, rather than as a type of data source. He mentions that the source of data can vary. Alles and Gray (2014, p. 5) state that:

“To auditors the data in (or contents of) big data refers to collections of multiple types of data, which could include some mix of traditional structured financial and non-financial data, logistics data, sensor data, emails, telephone calls, social media data, blogs, as well as other internal and external data.” (Alles and Gray, 2014, p. 5)

However, Earley (2015) discusses possible challenges that data analytics bring. He states that the challenges fall in three broad categories (Earley, 2015): (1) training and expertise; (2) data availability, relevance, and integrity; and (3) expectations of the regulators and financial statement users. Brown-Liburd et al. (in Earley, 2015) states regarding the first challenge, that auditors who finished their accounting programs are expected to understand how to apply accounting rules and to understand audit risks associated with particular accounts. They are not directly trained to understand whether the transaction itself makes sense or to develop expectations about sales that would then enable them to recognize when an abnormality has occurred. Or maybe even more important, how to follow up on the detected abnormality.

The second challenge that data analytics bring is the data availability, relevance and integrity (Earley, 2015). There are clients who don’t have the possibility to capture all the data

(16)

16 which is useful to the auditor. It is also possible that the data collected contain noise. It is even possible that the auditor does not have permission to access all the data and whether the client would share this. This lack of access is a potential deficiency for fraud detection.

Furthermore, since the rise of big data, it is possible that this data could be obtained internally and externally. This is the point where the auditor assesses whether the data arose from a source that is secure and whether or not the data has been meddled with. The completeness of data could also be an issue here, because a lack of quality of data may be accepted, but for auditors, it means allowing inaccuracies to ‘slip in’. Therefore, it has a negative impact with the focus on auditing on data integrity (Earley, 2015, p. 498).

The third and last challenge mentioned in the article of Earley (2015) is the expectations of the regulators and financial statement users. Earley (2015) states in his article that the gap in expectations occurs when users of the financial statement (regulators or investors) believe that auditors provide 100% assurance. Although in reality, auditors only provide reasonable assurance, due to sampling on a test basis, is almost 100%. In addition, Earley (2015) mentions in his article that the focus of data analytics on non-financial information give regulators a certain kind of fear that possibly shows that auditors are paying less attention to auditing their clients and more attention on providing them non-audit services.

2.3.2 Success factors of data analytics

Verver (2015) states that data analytics enabled many audit teams’ success and return on investment. It aided organizations in such a way that it increased the productivity of the audit function and improves the quality and value of audit findings. It gives auditors the ability to examine and test entire populations of transactions that are focused in the audit area. An example given by Verver (2015) is a large car rental company, which transformed the audit process and reduced traditional audit by 10000 hours annually.

In the research of Verver (2015), he states that there are six important success factors in using data analytics. These success factors are as following: (1) strategy and leadership, (2) goals and metrics, (3) planning and project management, (4) a knowledgeable and organized team, (5) the business case for resources and (6) technology.

According to Verver (2015), it is an important starting point to define the strategic objectives for audit analytics. Many internal audit departments fail to develop and implement audit analytics, because they do not treat it as a “strategic initiative”. As well as the overall objectives are not clear and these departments lack necessary resources. Furthermore, Verver (2015) states that the involvement in an audit analytics implementation will add to the strategic importance. This could deliver significant and sustainable advantages.

(17)

17 Internal audit departments could use goals and metrics in order to establish specific objectives by prioritizing the expected benefits (Verver, 2015). Goals and metrics for these departments could be that data analytics should be used on “x percent” of audits within an y-month time frame. Another possibility may be that these departments want to reduce in audit hours for “x percent” because of the use of data analytics instead of manual methods. Establishing these goals could help align the audit team and provide a basis for managing this implementation process.

Titera (2013, p. 326) states in his article that on one side, data analytics may more look like a simple analytical procedure (e.g., gross margin analysis). However, unlike such a simple analytical procedure, data analytics typically gives the possibility to the practitioner to evaluate activity along various dimensions (e.g., time, business unit, and segment). This possibility may show significant results, especially in terms of audit effectiveness (e.g., the ability to assess risks, identify anomalies, and detect errors). As for effectiveness increases, enhanced auditor reliance may increase (Titera, 2013).

Because of the advances of technology, the audit evidence that comes from more sources, like big data, exogenous data, the ability to analytically link different processes, database-to-database confirmation, and continuous monitoring alerts Titera (2013, p. 327). Furthermore, the level of evidence gained from data analytics is a matter of professional judgment. However, factors that must be considered include the suitability of using data analytics, the relationship of the analysis to the specific audit assertions, the reliability of the data underlying the analysis, the level of disaggregation, and the scope of the analysis (Titera, 2013).

Verver (2015) mentions in his article that the implementation of audit-analytics is undermined by poor management. Verver also states that an important technology-driven initiative, effective planning and project management are critical to success. In order to achieve more advantages, audit analytics need to be integrated into the overall audit process. All auditors should know how audit analytics are used and what their role is in this particular process (Verver, 2015).

Additionally as earlier mentioned in paragraph 2.3, Earley (2015) states in his article that big data is seen as a leading part of business in the future. Any organization that falls behind in its development of data analytics capabilities is expected to perform less than its competitors and could experience bad consequences.

In having success of implementing and maintaining an audit analytics process, it has to depend heavily on the extent of knowledge and skills available within the own internal audit department and how this department is organized (Verver, 2015). Training plans needs to reflect

(18)

18 individual roles and are related to the level of knowledge. Managers and reviewers need to be trained in the overall audit analytics process.

Internal audit departments that achieve success in using analytics develop a business case to recognize investments costs and the expected benefits and measuring progress in achieving the objectives of these departments (Verver, 2015). Furthermore, a lot of data analytics software are used to support audit analytics (e.g. Microsoft Office Excel). The developments in this factor (technology) has made it possible for auditors to use this kind of specialized software to recognize costs and benefits (Verver, 2015).

2.3.3 Limitations of data analytics

According to Wang & Cuthbertson (2014), there are eight important issues identified by the practitioner. Those issues are: (1) What is the role of data analytics in risk analysis? (2) What procedures ensure quality results? (3) What are the implications and impediments to population testing? (4) Should audit data analytics utilize external data? (5) Can external auditors rely on internal auditors’ use of data analytics? (6) What factors affect interpretation of data analytics results? (7) What are the consequences of using data analytics? (8) Does the audit profession need a data analytics framework?

Although data analytics provides substantial audit benefits, such as working with large operational data, identifying patterns, etc., there are issues in ensuring the quality of the results of data analytics (Wang & Cuthbertson, 2014). Furthermore, data quality is always an issue with larger samples. This brings up questions as followed: “Are there appropriate steps to validate data from different systems/sources? Are there processes to document, to review, and to address the data quality issue? More importantly, how does the quality of the data affect the quality of the results?”. Such questions despair the certainty that the usage of data analytics should give.

There was a lot of discussion about sampling in the last few decades (Wang & Cuthbertson, 2014). With data analytics, it is possible to test 100 percent of the population. It may even be desired for a smaller company to truly test the entire population. However, Wang and Cuthbertson (2014) state that with the burst of operational data, an analysis of the entire population could be not that feasible as expected. It is important to understand the considerations or motivation for choosing a population or a sampling analysis.

Additional concerns that an auditor can get is the quality of the external data. The reliability of those sources and the assurance of them should be determined before this data can be used in the analysis (Wang & Cuthbertson, 2014). Additionally, in the Auditing Standard No. 5 it is stated that it is allowed for external auditors to depend on the internal audit function (PCAOB, 2007) (in Wang & Cuthbertson, 2014, p. 158). However, the reliance of an external

(19)

19 auditor on the internal audit function is quite limited. This raised questions as followed: “Does the reliance improve audit quality, effectiveness, and efficiency?” or “Does the use of a continuous audit and data analytics by internal auditors increase the reliance by the external audit?”.

It is important for auditors to understand how they should use the results of the analysis. It is the most time-consuming part, because of interpreting and analysing the results. It is a challenge for data analytics processes to manage the auditors’ time and effort to the components that require the most attention (Wang & Cuthbertson, 2014). However, according to Wang and Cuthbertson (2014, p. 159), the understanding of the usage of data analytics is still limited. The potential impact that the usage of data analytics has, would mean a change in the education and training of auditors in technology and problem-solving techniques.

2.3.4 Integration of data analytics in audits

Generally, many professionals use data analytics in the so-called “execution” phase of audits. However, the fact is that data analytics tools allow creating more value when used the whole time in performing the audit. This starts from planning to reporting on each individual audit unit. (Cangemi, 2014).

Cangemi (2014) states furthermore that using insights gained from data during the audit planning stage as an example, allows you to understand the process, the hotspots for risks, and the current characteristics of risk management within the organization. Cangemi uses an example that illustrates this process. During an accounts payables audit, there is a request for data done that concerns all the data from the department. The auditor performing this audit will probably send a request for data regarding; purchase to payments, the vendor master, all purchasing transactions, purchase orders, high-level payment information, and so on. Cangemi (2014) states that with this data, the auditor will know the number of transactions, vendors on file, vendors actually used, vendors used most often, and the purchasing patterns across different goods and services. When running tests on all the data, an auditor can get a better understanding of the extent of the business.

When implementing Big Data in financial statement audits, it requires auditors to have a certain knowledge about this technology and have the appropriate hardware and software resources (Cao, Chychyla, and Stewart, 2015). Because of that, many businesses outsource their Big Data solutions to other companies. Thus, the training that auditors will probably get may be out the scope of the knowledge that an auditor should possess. Therefore, it is expected by Cao et al. (2015) to hire new analytically trained professionals or use more third-party solutions. However, the use of third-party solutions could raise some concerns regarding privacy. Then again, auditors rely on banks when performing audits (Cao et al., 2015).

(20)

20 According to Cao et al. (2015), there are some issues when dealing with Big Data analytics. First, Auditors favour using ‘all’ the data in relatively large datasets, which could be rather chaotic if not saved properly. Therefore, the auditor will focus more on correlation than causation. However, Cao et al. (2015) state in their article that this thinking somewhat unfamiliar. With proper guidance and education, it is possible for them to modify themselves.

Second, the volume of the obtained data could give computational challenges. There are some analytical techniques in audit that could not be applied to Big Data. A solution for that is to use significantly fewer resources or select compartments of data that could be managed by more complex tools. This latter is considered more valuable for an audit (Cao et al. 2015).

Third, an earlier discussed challenge is the privacy concerning Big Data. Sometimes an auditor may use more information in performing the audit than that the client normally gives to the auditor. Others may benefit from this information that has gone public (Cao et al, 2015). However, this does not only happen in the auditing profession. For example, the European Union is analysing Google on antitrust and privacy concerns because of the usage of Big Data (Mayer-Schönberger and Cukier, 2013) (in Cao et al, 2015, p. 428).

Finally, it may be a possibility that the auditor would be second-guessed. This happens if ‘all’ the data is processed and there was a failure to find a fraud or error. However, this is not a new problem, as this was already the case before the usage of data analytics in audits (Cao et al., 2015).

2.4 Data analytics and professional scepticism

As stated in paragraph 2.3.3, there are multiple issues when implementing data analytics in audits. For example: “What factors affect interpretation of data analytics results?”. Understanding or analysing the results of data analytics can be the most important step in the process, but also the most time-consuming. This is especially true when auditors are distracted by false negative exceptions (Wang & Cuthbertson, 2014). It is still a challenge for data analytics processes to direct an auditors’ time and efforts to the elements that require the most attention.

Furthermore, as earlier stated in paragraph 2.3.3, additional concerns that an auditor can get is the quality of the external data. The reliability of those sources and the assurance of them should be determined before this data can be used in the analysis (Wang & Cuthbertson, 2014). Additionally, it is important for auditors to understand how they should use the results of the analysis. It is the most time-consuming part, because of interpreting and analysing the results. It is a challenge for data analytics processes to manage the auditors’ time and effort to the

(21)

21 components that require the most attention (Wang & Cuthbertson, 2014). This particular challenge tests the professional judgement of the auditor.

When looking at the effect of usage of data analytics, there are multiple characteristics that can be influenced by it. Another example mentioned above is the search for knowledge. By focusing on the results that come from the usage of data analytics, it is unknown whether the auditor maintains the same level of scepticism as without the usage of data analytics. One auditor with less knowledge of how the data is generated may make mistakes in interpreting the data and this can have a negative effect on the judgement of this auditor. This could influence the quality of audit opinions. Additionally, because of having less knowledge regarding the usage of data analytics, auditors may be less certain about the correctness, timeliness and completeness of the data obtained. This could be expected to have a negative effect on the self-esteem of the auditor. This study will investigate whether the usage of data analytics affects the professional scepticism of the auditor. A theory which contributed in a similar research is the MODE model by Fazio (1990) (used in the paper of Noviyanti and Winata (2015)). This theory is based on the attitudes–behaviour relationship described by Fazio (1990) in the MODE model (Motivation and Opportunity as DEterminants of the attitude–behaviour relationship). This describes two basic classes of the attitude-to-behaviour processes: “(1) in a controlled or deliberative fashion and (2) in an automatic or spontaneous fashion” (Fazio, 1990 (in Noviyanti and Winata, 2015)).

Based on Fazio’s MODE model (Ajzen, 2005), it is stated that “when people are sufficiently motivated and have the cognitive capacity to do so, they can retrieve their attitude toward a task or object in a purposeful manner so that their attitude influences their behaviour”. Ajzen further claims that when motivation or cognitive capacity is low, attitudes can become available only if they are automatically activated. Azjen (2005) also states that, mentioned in the MODE model, strong attitudes are automatically activated. Herein, attitude leads the behaviour in a spontaneous manner, without the individual actively thinking about it and without the individual’s necessary awareness of its influence.

According to Fazio (1990) attitudes must be ‘triggered’ in order to lead behaviour, and there are two cognitive processing modes through which attitudes can be activated. An example of a triggering attitude is given in the article of Elliot, Lee, Robertson, and Innes (2014). There are two drivers. The first driver’s attitude is characterized by a strong association between the act of speeding and has a positive evaluation of that behaviour. The second driver is characterized by a weak association between the act of speeding and has the same effect. Despite having equally positive attitudes regarding speeding, the first driver would be more likely to truly speed, because this driver possesses an “an attitude of sufficient associative strength to be chronically accessible” (Elliot et al,

(22)

22 2014, p. 50). This research was designed to test whether the MODE model/theory of Fazio also counts in other cases, in this case speeding attitudes and behaviour. The conclusion was indeed that there is a correlation between the MODE model and the tests done by the researchers (Elliot et al, 2014).

Wang & Curthbertson (2014) again state in their article “that acceptance and utilization of traditional and non-traditional computer-assisted audit techniques (CAATs) or, more specifically, data analytics for an audit, is lower than expected”. There are multiple reasons for this to happen. First, there is a lack of confidence in their own abilities (which speaks to the self-esteem and self-determining of the auditor). Second, there is organizational pressure and the infrastructure where those analytics are supposed to run are not as sophisticated as expected. Last, the expectation regarding performance and the facilitated conditions are rather uncertain. This could have an effect on the questioning mind of the auditor.

(23)

23

3. Hypothesis development

When doing a case-based experiment for the research related to the effect of using data analytics on the professional scepticism, the basis that is used in the MODE Model of Fazio could also be used in this paper. As earlier stated, “when people are sufficiently motivated and have the cognitive capacity to do so, they can retrieve their attitude toward a task or object in a purposeful manner so that their attitude influences their behaviour”. It has a positive spectrum on possible variables which could influence the behaviour, and thereby the professional scepticism of the auditor.

As earlier stated in paragraph 2.4, attitudes must be ‘triggered’ in order to lead behaviour. When there is a strong association between the attitude and the behaviour towards a certain variable, then it is more likely to happen. In an auditor’s case, an attitude regarding his or her scepticism can be triggered when there is a specific case where a problem appears. A strongly motivated auditor with the right capacities will still perform the task in a good manner.

Then again as earlier stated, Earley (2015) discussed possible challenges which data analytics bring. Brown-Liburd et al. (in Earley, 2015) states regarding the first challenge, that auditors who finished their accounting studies have been expected to understand how to apply accounting rules and to understand audit risks associated with particular accounts.

As earlier stated in paragraph 2.4, “acceptance and utilization of traditional and non-traditional computer-assisted audit techniques (CAATs) or, more specifically, data analytics for an audit, is lower than expected”. There are multiple factors discussed. First, there is a lack of confidence in their own abilities (which speaks to the self-esteem and self-determining of the auditor). Second, there is organizational pressure and the infrastructure where those analytics are supposed to run are not as sophisticated as expected. Last, the expectation regarding performance and the facilitated conditions are rather uncertain. This could have an effect on the questioning mind of the auditor. Additionally, since the rise of big data, it is possible that this data could be obtained internally and externally. This is the point where the auditor assesses whether the data arose from a source that is secure and whether or not the data has been meddled with. The completeness of data could also be an issue here, because a lack of quality of data may be accepted, but for auditors, it means allowing inaccuracies to ‘slip in’. This means that it has a negative impact with the focus on auditing on data integrity (Earley, 2015, p. 498). This has an effect on the questioning mind of an auditor. It is not certain that the auditor could not be affected by the challenges that the usage of data analytics brings.

Furthermore, the volume of the obtained data could give computational challenges. There are some analytical techniques in audit which could not be applied to Big Data (Cao et al. 2015). This, however, could affect the search for knowledge of the auditor in a positive way. A possible

(24)

24 reason for this is the fact that auditors are more sceptic about the information that is obtained. Auditors have to think about other possibilities (like specialized tools) to make such calculations possible.

Finally, because of the advances of technology, the audit evidence that comes from more sources, like big data, exogenous data, the ability to analytically link different processes, database-to-database confirmation, and continuous monitoring alerts Titera (2013, p. 327). Furthermore, the level of evidence gained from data analytics is a matter of professional judgment.

The article of Earley (2015) and Wang & Curthbertson (2014) show that there are various challenges data analytics have in the field of audit. These challenges can affect the professional scepticism of the auditor. Meaning that in developing a hypothesis, there are two possible directions in carrying out this research.

Therefore, based on this discussion, the alternative hypothesis of this paper will be non-directional. Meaning that the usage of data analytics is associated with the professional scepticism of auditors. It will not be specified whether the effect is positive or negative. The effect of this phenomenon will be specified after conducting this research. Specifically, this paper will form the following hypothesis:

H1: “The usage of data analytics is associated with the professional scepticism of auditors.”

H0: “The usage of data analytics is NOT associated with the professional scepticism of auditors.”.

(25)

25

4. Research method and design

4.1 Research design

The empirical approach of this study will be focusing on quantitative research. This thesis is conducted with a case-based experiment. The participants of this experiment had to read a case and answer, among other things, a couple of questions regarding the case. The participants had to act as a senior auditor of a Dutch company called Printz. This company focuses on selling printers. In this case, because of the recent developments in the printing industry (sustainable ink), Printz is at risk of having outmoded inventory. Furthermore, the results of Printz were doubtful. The auditor had to determine how much he or she questions the audit opinion of the provisions of Printz regarding the audit procedure which was used in the case. There were more questions asked in the first questionnaire, but those questions were not used in this research. The goal of these questions was shift the focus of the participants. If they knew the goal of this research, then this could affect their answers. The second questionnaire was a 10-item questionnaire containing questions with a six-point scale differing from “strongly disagree” to “strongly agree”.

Futhermore, in the second questionnaire of the experiment contained 10 of the 30 questions of the Hurtt-scale. These questions were changed in regard of the experiment, as many questions of the Hurtt-scale are individual-oriented. The other questions of the Hurtt-scale were therefore not sufficient enough to use in this experiment. These questions would give information about the scepticism as an individual and this would not be relevant for this experiment and manipulation. In addition, questions 5 and 6 are reverse scored. In the article of Hurtt (2010) it is stated that those are a couple of the questions that needs to be reverse-scored to determine the professional scepticism of the participants. The maximum score in this case would be 60, where if the Hurtt-scale is conducted with all the questions, the maximum score is 180.

The participants read in the case about the events of Printz in 2017. Those were comparisons of 2016 and 2017 regarding the events. These events contained, among other things, about profit, sales expectations, inventory position, impairments and provisions. The information and the occurring events are equal in both studies. The manipulation in both cases are: (1) The usage of data analytical procedures or (2) traditional analytical procedures.

In this thesis the independent samples t-test is used to compare two groups with an independent variable. The dependent variable is the professional scepticism of the auditors. With the help of Qualtrics, it was possible to collect the data by sharing a link to the participants and Qualtrics made it possible to randomly assign one of the two cases to the participant. The experiment and the setting can be found in appendix two.

(26)

26 4.2 Sample description

The total participants of this experiment are 59 whereof 53 have completed the survey. Of the 53 participants, there were four who did not pass the manipulation check, meaning that they did not remember which procedure is used in the case. The results of those 49 participants were used in this research. Therefore, the hypothesis is tested with 49 auditors. The participants were invited by an email containing a link. In this email, they were invited to participate in the voluntarily research.

The following participants are involved in this experiment: (1) trainee, (2) staff, (3) senior staff, (4) managers, (5) senior managers, (6) directors and (7) partners. The functions of the participants mentioned before are directly related to the audit, as the first three groups are the practitioners of the audit and the last 4 groups review the performed audit by the staff. Therefore, the 49 participants consist of 35 males (71,4%) and 14 females (28,6%), which have a staff-level as mentioned before. These people are used for testing the hypothesis. Table 1 contains an overview of the staff-level of the participants (number and percentage).

Table 1 - Participants

Staff level N Percent

Trainee 2 4,10 Staff ₂₃ _46,90 Senior staff 16 32,70 Manager 3 6,10 Senior manager 3 6,10 Director 1 2,05 Partner 1 2,05 Total 49 100,0 4.3 Experimental task

In this experiment, two cases were made. The introductions and events of these cases are the same. The participant’s task was the same for both cases. As mentioned earlier, they had to fulfil the role of senior auditor of an accounting firm in the Netherlands and do an audit of Printz. This company has 10 locations in the Netherlands and has approximately 1000 employees.

This case started with the fact that 2016 was a bad year for Printz. Because of recent developments, Printz thought that 2017 would be a much better year. However, sadly enough, this was not fully the case. The results were less than expected by the upper management. Sales were less compared to the previous year and the accounts receivable increased. The cash flow has

(27)

27 decreased in comparison with the previous years. The provision for bad debts were constant in comparison with the previous years. The prior accounting numbers for 2017 suggest that the profit has increased a bit relative to 2016. The firm had less impairments on company assets in 2017.

The goal of this information given in both cases was to set off the scepticism of the auditor. The information mentioned before were the same in both cases. Additionally, the participants also got information regarding the used audit procedure in the case (being traditional analytical or data analytical). This information was different in both cases, meaning this information was manipulated. After reading the case, the auditor had to indicate how much he questioned the audit opinion about the provisions of Printz with regard to the audit procedures on a scale from 1 to 10.

4.4 Manipulation check

After finishing the case, the auditor has to answer the second questionnaire. This questionnaire contained a question with a manipulation check. In this manipulation check, the participant was asked which analytical procedure was used in the audit. The possible answers were the two manipulations (traditional analytical or data analytical) and not knowing the answer. As mentioned before, 4 participants did not pass this manipulation check. Finally, after answering these questions, the participant was asked to fill in other data, regarding gender, age, staff level and work experience in years. These questions can give specific information about different factors which also could influence the professional scepticism. Table 2 contains a summary of the variables used in this thesis, with additional descriptive statistics (mean, range and standard deviation).

(28)

28 Table 2 - Variables used in test

Name variable Theoretical values Mean Range St. Dev.

Dependent variable

Case question scepticism Scale: 1 to 10 6,61 9,00 2,15 Hurtt measurement scepticism Score: 10 – 60 41,96 22,00 4,80 Independent variable Analytical procedure in audit 1 = Classical analytical procedures - - -

2 = Data analytical procedures - - -

Manipulation check

Used analytical

procedure in case 1 = Classical analytical procedures - - - 2 = Data analytical procedures - - -

3 = I don't know - - - Control variables Age - 28,49 32,00 6,62 Gender 0 = Female _0,73 _1,00 _- 1 = Male Work experience - 5,71 29,50 6,40

Staff level 1 = Trainee

2,80 6,00 1,19 2 = Staff 3 = Senior Staff 4 = Manager 5 = Senior Manager 6 = Director 7 = Partner

(29)

29

5. Empirical results

5.1 Descriptive statistics

For conducting this research, the variables are operationalized. There are two variables in this research, namely the dependent and the independent variable. According to Field (2013), the dependent variable is a variable that the researcher measures, after making changes to the independent variable, which is expected from that the independent variable is affected. In this case the dependent variable is the professional scepticism. Field (2013) again states that the independent variable is a variable manipulated by the researcher, assuming that it affects the dependent variable. The independent variable is in this case the usage of data analytics.

In table 3 the descriptive statistics are summarized for the case question divided in the two groups. This table reports the numbers of participants and the average scepticism of the two groups according to the Hurtt questions. This shows that a difference is visible in the mean of the two groups, namely a score of 42,74 for the group with data analytical procedures and 40,77 for the group with traditional analytical procedures. This difference in scepticism is visible when using data analytical procedures, in comparison with the usage of traditional analytical procedures. This difference equals the assumption set earlier as H1: “The usage of data analytics is associated with the professional scepticism of auditors”. Then again, this does not make it clear whether this difference will be significant or not. Furthermore, the case question (question 3) from the first questionnaire is not visible in table 3. What stands out in this table is that, ignoring the significance at this moment, that there is a positive effect on professional scepticism of the auditor when using data analytical procedures.

5.2 Results for hypothesis test

The hypothesis as stated in chapter 3 states: “the usage of data analytics is associated with the professional scepticism of auditors”. This hypothesis will be supported when there is a significant difference between the means of the two groups (Data analytical procedures and Traditional analytical procedures). The mean of the respondents with data analytical procedures is higher than the respondents with traditional analytical procedures. Then again, as earlier

Table 3 -Descriptive statistics Hurtt questions

Dependent variable (1) Data analytical procedures (2) Traditional analytical procedures Total Professional scepticism

Hurtt questions N = 23 Mean = 42,77 N = 26 Mean = 41,04 N = 49 Mean = 41,96

(30)

30 mentioned, this difference must be statistically significant (P < 0,05). This association is tested with the independent-samples t-test.

According to Field (2013), an independent-samples t-test must be done to determine whether there is an association between the usage of data analytics and the professional scepticism of the auditor. Before doing an independent-samples t-test, the data has six assumptions it has to meet (Field, 2013). Field (2013) states that the first assumption is that the dependent variable needs to be measured at the interval- or ratio-level. The second assumption is that independent variables should have two or more categorical groups. The third assumption is that there has to be independence of observations. The fourth assumption is that there has to be no significant outliers. The fifth assumption is that the dependent variable has to be approximately normally distributed for each category of the independent variable and the sixth and final assumption is that there has to be Homogeneity of Variances (Field, 2013).

For the fifth assumption, the Shapiro-Wilk test is conducted. This is shown in table 5. Field (2013) states in his book that the test of Normality shows whether the sample is normally distributed. The alternative hypothesis is supported when P < 0,05, meaning that the distribution of the selected variable is not approximately normally distributed. Because this is a sample of 49 people and this is seen as a small sample, it is better to look at the Shapiro-Wilk test (“Normaliteit”, 2014). As shown in table 5, the significance of the Shapiro-Wilk test is in both cases higher than 0,05 (P > 0,05). This shows that the alternative hypothesis is not significant and therefore the null-hypothesis is supported, meaning that the tested variables are approximately normally distributed.

Table 4 - Tests of Normality

Kolmogorov-Smirnova _Shapiro-Wilk Statistic df Sig. Statistic df Sig. Total score Hurtt Questions Classical 0,104 23 0,200* _0,965 ₂₃ _0,575

Data 0,172 26 0,046 0,960 26 0,386

*. This is a lower bound of the true significance. a. Lilliefors Significance Correction

For the sixth assumption, the Homogeneity of Variances test needs to be conducted. If the assumption is met, then this research could be done with the independent-samples t-test. The Homogeneity of Variances and independent samples t-test are shown in table 7. According to this table, the assumption for Homogeneity of Variances, also known as Levene’s test, is supported. The significance is 0,709 according to Levene’s test. This is higher than 0,05 (P > 0,05), meaning that equal variances are assumed. This means that for this research the first row is used to determine the significance.

(31)

31 Table 5 - Group Statistics

N Mean Deviation Std.

Std. Error Mean

Total score Hurtt Questions Classical 23 41,04 4,487 0,936

Data 26 42,77 4,998 0,980

Table 6 - Independent Samples Test - Hurtt questions Levene's

Test for Equality of

Variances t-test for Equality of Means

F Sig. t

df Sig. (2-tailed) Difference Mean Difference Std. Error 95% Confidence Interval of the Difference

Lower Upper Equal variances assumed 0,141 0,709 -1,265 47 0,212 -1,726 1,364 -4,470 -0,019 Equal variances not assumed -1,274 46,986 0,209 -1,726 1,355 -4,452 -1,000

Furthermore, the results for the independent-samples t-test are shown in table 6. This shows that the P-value equals 0,212 (P < 0,05). This means that the P-value is higher than 0,05, meaning that H1 is not supported in this study. Because of the significant result of the test H1 is not supported, meaning that “the usage of data analytics in performing the audit is not associated with the professional scepticism of auditors”.

5.3 Results for hypothesis test with case question

Song et al. (2013) states in his article that it is possible to combine variables that have different measures. In his article he showed an example of questions with a 5 scale and 3 Likert-scale. This was combined by Song et al. (2013) using composite variables. A composite variable (C) is a variable created by combining two or more variables that are highly related to one another conceptually or statistically (Song et al., 2013). The same method is used for combining the case question and the Hurtt questions. First the raw scores are transformed into z-scores. Those are scores with a mean of 0.000000 and a standard deviation of 1.000000. By having the same mean and standard deviations, the variables are standardized. This is tested by transforming

Effect of the usage of data analytics in audits and the professional scepticism of the auditor