FRAUD RISK JUDGMENT MEASUREMENT SCALE DEVELOPMENT

cite How to paper: this Julian, L., Johari, R. J., Said, J., & Wondabio, L. S. (2022). Fraud risk judgment measurement scale development [Special issue]. Journal of Governance & Regulation, 11(1), 303–311. https://doi.org/10.22495/jgrv11i1siart10 Copyright © 2022 The Authors This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). https://creativecommons.org/licenses/by/

Recently, many financial scandals and frauds have been published in mass media. It has resulted in ruining the public trust in the internal auditor profession as the third line of defense since the public perceived frauds detection and prevention as the internal auditors' responsibility (DeZoort & Harrison, 2018). The internal auditors' fraud risk judgment performance has been questioned. There are many scales to measure fraud risk judgment; however, they are mostly related to financial-statement-related frauds with external auditors as the targeted respondents and still lack those to measure fraud risk judgment of internal auditors. This paper aims to propose the scale for measuring the performance of internal auditors' fraud risk judgment. Since there are many internal auditors without accounting background, the fraud case should be developed to be more general, instead of financial-statement-related frauds. The study followed the best practice step by step in developing a scale proposed by Boateng, Melgar Neilands, Frongillo, -Quiñonez, (2018). and Young It involved 5 and developing in experts the items, validating 106 and (EFA) analysis factor exploratory the in respondents 202 respondents in the confirmatory factor analysis (CFA). All the required indicators in the steps were acceptable; therefore, we can conclude that the scale is valid and reliable. The scale was developed based on the fraud triangle theory; hopefully, it can contribute to providing alternative fraud risk judgment measurement for internal auditors.

INTRODUCTION
The daily presence in companies has led internal auditors to have better advantages in detecting and preventing frauds. Consequently, fraud detection and prevention have been perceived as the responsibility of the internal auditors (DeZoort & Harrison, 2018). The publication of financial scandals of large companies in the mass media has resulted in the effectiveness of the internal audit function being questioned. Poor performances of internal auditors in making fraud risk judgments led to considering them unable to detect and prevent fraud. It ended in reputational damages and unnecessary financial losses. Therefore, it is vital to measure the internal auditors' fraud risk judgments. This paper aims to develop the fraud risk judgment performance measurement for internal auditors.
Recently . However, currently, the internal auditing profession is no longer dominated by accountants and there are a lot of internal auditors without accounting background; therefore, the case needs to be a more general fraud case instead of a financial-statement-related fraud case. Moreover, at the moment, the role of information technology is vital to businesses, thus the case should be more related to the IT environment. Since this kind of scale is still lacking, it is necessary to develop the fraud risk judgment scale using a questionnaire survey in the IT-related environment. This paper confirmed that the proposed scale has acceptable validity and reliability through the process of exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). Hopefully, it can contribute to providing an alternative scale to measure fraud risk judgment performance of internal auditors without accounting background.
This paper consists of five sections. Section 1 is an introduction to the study, Section 2 reviews the relevant literature, Section 3 describes the detailed methodology. The next Section 4 elaborates on the result and discussion of the study. Finally, the conclusion is presented in Section 5.

LITERATURE REVIEW
In this study, fraud risk judgment refers to a predetermined course of action taken in response to an entity's vulnerability to an individual capable of combining all three aspects of the fraud triangle. Auditors will undoubtedly make judgments throughout the audit process and during each audit assignment. When auditors make judgments, they are likely to be severely inefficient in their response to fraud risk, resulting in low fraud detection rates (Trompeter, Carpenter, Desai, Jones, & Riley, 2013).
Tschakert, Needles, and Holtzblatt (2016) argued that by sharpening auditors' in evaluating red flags, internal auditors can more effectively address fraud threats and safeguard firm assets. In line with the findings, some previous studies assessed fraud risk judgments using the accuracy with which they identified a set of red flags as predictors of fraud occurrences ( This paper used the fraud triangle (Cressey, 1950) as an underpinning theory. The theory stipulates three characteristics of a fraudster: pressure, opportunity and rationalization. Wolfe and Hermanson (2004) enhanced the theory by augmenting capability as the fourth characteristic and labeled it as the fraud diamond theory. On the contrary, Dorminey, Fleming, Kranacher, and Riley (2012) viewed that the capability is only a refinement of the opportunity characteristic and should be considered a part of it. Identifying the combination of the characteristics was significantly effective in detecting and preventing fraud (Homer, 2020;Nakashima, 2017).
Fraud risk judgment is defined as an idea, opinion, or estimate about the vulnerability that an organization faces from individuals capable of combining all three elements of the fraud triangle and translating them into action to modify the initial audit plan. It tacitly uses the theory of fraud diamond which combines pressure, opportunity, rationalization, and capability as the elements of fraud. Fraud usually occurs with several red flags that precede it. The red flags can lead auditors to uncover fraud; therefore, the ability to identify red flags shows the ability to uncover frauds (

Population and sample
The population of this survey consisted of the in-house internal auditor practitioners who were registered at the Institute of Internal Auditors Indonesia and worked for public and private organizations in Indonesia. By adopting a judgment sampling method, the sample was chosen based on specific criteria to meet the objectives of the study, i.e., the respondents with a minimum of three years in service and currently still active as in-house internal auditors in West Java, Indonesia.

Data collection
In developing the scale, this study follows the iterative steps suggested by Boateng, Neilands, Frongillo, Melgar-Quiñonez, and Young (2018) that consisted of three phases: 1) item development phase, 2) scale development phase, and 3) scale evaluation phase.
Firstly, for Phase 1 (items development), five experts who have at least 15 years of experience as internal auditors, were interviewed. Whereas for Phase 2 (scale development), a pilot test was conducted, and 106 respondents were collected for the exploratory factor analysis. Finally, for Phase 3 (scale validation), a total of 600 self-administered e-questionnaires were distributed to internal auditor practitioners. 208 surveys were completed and returned, but only 202 (33.67 percent) were found to be legitimate for further research due to outliers. According to Hair, Hult, Ringle, and Sarstedt, (2017), the required minimum sample size for this study is just 68 (Hair et al., 2017); consequently, the 202-sample acquired was deemed suitable and appropriate.

The results of Phase 1 (items development)
The generation of items was accomplished using a combination of both inductive and deductive methods (Boateng et al., 2018). The inductive method was performed in the discussions with 5 expert panels to capture insights on the most essential fraud risk factors in their daily practices. The panel members were the practitioners who have at least 15 years of internal auditors' working experience.
At the same time, the deductive method was utilized by reviewing the literature, such as IAASB The fraud risk judgment should not be separated from the context. It can be illustrated in a mini case (see Appendix). Items without context can only be used to assess inherent risks. Therefore, the panel suggested the use of a mini case to describe the context of internal control design and effectiveness that allowed the respondents to make a judgment on hypothetical residual fraud risk.
The internal control contexts were related to information technology general control (ITGC). The results of the items generation were exhibited in Table 1. Initially, 13 items were proposed to be processed further with EFA. The proposed items were generated using the fraud diamond theory since it is deemed to be more effective for fraud risk assessment than the fraud triangle-based assessment (Boyle et al., 2015b; Santoso & Surenggono, 2018). However, based on the result of exploratory factor analysis, the fraud triangle theory should be utilized (please refer to subsection 4.2.). The following stage was the content validation to ensure that the items contained in the measurement were relevant to the fraud risk judgment (Boateng et al., 2018). In the second round of the discussion, the panel validated the item statements by assessing their relevance to the fraud risk judgment measurement in a Likert scale form. The Likert scales used were very irrelevant (1), irrelevant (2), relevant (3), and very relevant (4). The statistical means of the panel's assessment results are depicted in Table 1. Any values below 2.50 (if any) should be removed from the list since they are irrelevant to the fraud risk judgment assessment. The panel was in agreement that all items are relevant and should not be removed from the list. Likewise, the panel also considered that the items were practical and able to reflect the fraud risk judgment in the daily practices of internal auditing. The Fleiss Kappa inter-rater value was applied to assess the validity of the items in the constructs (Boateng et al., 2018). The Fleiss Kappa for 5 raters, 4 categories and 13 items was 0.65. It can be concluded that the strength of agreement of the panel was substantial (Landis & Koch, 1977).

The results of Phase 2 (scale development)
The pre-test of the scale is performed by five academicians and ten internal auditors' practitioners. In the pre-test stage, the scale would be easier to understand if the same questions were combined, instead of repeated in all items. The following stage was administered with a purpose of a data collection of pilot testing; henceforth, 106 respondents were used for the analysis. The demographic profile of them is shown in Table 2. Items reduction and factors extraction were conducted using the exploratory factor analysis. Firstly, to assess whether the data have a sufficient inter-correlation degree among the items for further processing with the EFA, Bartlett's test of sphericity must be signed at a p-value lower than 0.05 and the Kaiser-Meyer-Olkin (KMO) test for the measure of sampling adequacy (MSA) has to be higher than 0.500 (Hair, Black, Babin, & Anderson, 2018). The results of Bartlett's test were significant at a p-value = 0.000 and the KMO-MSA was 0.847. Both were in the acceptable range (as shown in Table 3). Moreover, each items' MSA ranged from 0.692 to 0.942, which were also in the acceptable range. Thus, the results indicated that the EFA was appropriate for further data analysis. Based on the extraction sum of squared loading, it was indicated that only two factors were involved as shown in Table 4. However, the fraud triangle theory said that at least three factors have to be involved; therefore, this study was forced to have three factors: opportunity, pressure and rationalization. From the rotation result shown in Table 5, the four items (CAP03, OPP01, OPP03, and RAT01) should be dropped due to low factor loadings. Hair et al. (2018) suggested that any factor loadings below 0.55 should be eliminated. Thus, the results of the EFA were the opportunity factor (consisted of four items: CAP01, CAP02, OPP02, and OPP04), the pressure factor (consisted of three items: PRE01, PRE02, and PRE03) and the rationalization factor (consisted of RAT01 and RAT02).

The results of the Phase 3 (scale evaluation)
The scale evaluation aims to test the reliability and validity of the scale at different times and different datasets. A new survey was administered, and 208 responses were collected, but only 202 responses were valid for further analysis using a CFA. The respondents' demographic profile is depicted in Table 6. Dimensionality tests were conducted by assessing the absolute fit, the incremental fit and parsimonious fit. The goodness of fit index (GFI) should produce a value higher than 0.90 in order to gain the absolute fit or the Chi-square/df should be lower than 5 (Hair et al., 2018). At the same time, the incremental fit was assessed using comparative fit index (CFI) and normed fit index (NFI) which should be higher than 0.90 (Hair et al., 2018). The parsimonious fit was assessed using PGFI, PNFI and PCFI all of which should be higher than 0.50 (Hooper, Coughlan, & Mullen, 2008). Table 7 depicts the result of dimensionality tests that confirmed that the model fit was not an issue. Reliability is the consistency degree obtained when the scale is repeated in identical circumstances. It can be assessed using Cronbach's alpha (CA) and composite reliability (CR). The acceptable levels were 0.700 for CA and 0.708 for CR. As shown in Table 8, the CA of the fraud risk judgment (FRJ) scale was 0.799 and the CR was 0.882, which are in the acceptable range. In the first order, the CR and the CA of the factors (opportunity, pressure and rationalization) were higher than minimum acceptable values. Therefore, the scale has no reliability issues. The convergent validity test aims to assess the extent to which a scale indeed measures the intended evaluated construct. Hair et al. (2018) suggested that the minimum acceptable level of the items can be at least 50% of the variance of the latent construct (the average variance extracted (AVE)) should be higher than 0.500) and the factor loading should be higher than 0.708 (Hair et al., 2018). As shown in Table 8, all factor loading and the AVEs are higher than the acceptable level for the first and second orders. Thus, the convergent validity is not an issue for the scale. The discriminant validity was achieved when the items dedicatedly measure a concept of construct without a potential overlap. Table 9 shows that there is no correlation between the factors since all the ratios are lower than 0.85 as suggested by Kline (2016). Table 10 shows that the cross loadings are much lower than the factor loading. These two tables indicate that there is no discriminant validity issue in the scale. Empirically, the scale has been statistically proven to have acceptable reliability. Even though, there is a shifting in the underpinning theory. Initially, the scale decomposed the fraud risk judgment into fraudsters' characteristics based on the fraud diamond theory, since the theory was an enhancement of the fraud triangle and deemed to be more effective to assess fraud risk judgment

CONCLUSION
This study was conducted to fulfill the need for measuring the performance of internal auditors in making fraud risk judgments.
Currently, the measurements of fraud risk-related judgment involve financial statements analysis on which judgments are made since previous studies mostly targeted external auditors as their respondents and few of them targeted internal auditors. Indeed, previously the internal auditor profession was dominated by accountants, but now it is no longer the case. The internal auditor profession has involved many auditors from various disciplines other than accountants. Therefore, the need for more general fraud risk judgment becomes urgent. A measurement scale that is not based on financial statement analysis is expected to be a contribution in providing an alternative measure of fraud risk judgment for internal auditors, especially those who do not have an accountant background.
This study involved the expert panel and the past related literature in developing the items. Then, the proposed items were purified by statistical EFA and lastly they were validated by the CFA. Based on the process passed, it can be concluded that the proposed measurement scale, as shown in Appendix, has adequate validity and reliability to be used to measure fraud risk judgment. There are some limitations to this study. Firstly, the measurement scale does not consider the element of integrity in measuring potential fraud. The integrity can be reflected in the form of ethical value (Said, Alam, Ramli, & Rafidi, 2017) or religiosity (Said, Alam, Karim, & Johari, 2018) which is proposed as the fourth component of the fraud theory. Another limitation is that limited information was available to make decisions or judgments due to conciseness reasons. These limitations provide opportunities for future studies. The studies which consider the integrity element assessment need to be developed. Moreover, the case which contains comprehensive information in the real circumstances of the auditors who make judgment is also needed to develop. prevention methods. Managerial