THE EFFECT OF AUDITOR CHARACTERISTICS AND GENDER DIVERSITY ON FRAUD DETECTION

Accounting fraud can occur in any company and must be detected to reduce it. This research aims at empirically investigating the effects of auditor characteristics and gender diversity on fraud detection. The population used mining companies listed on the Indonesia Stock Exchange for 2016-2020 with 205 observation data. The data analysis technique used was the Panel Data Regression Model. The results showed that auditor independence and audit report lag had a positive effect, and audit tenure and gender diversity harmed fraud detection. Additionally, auditor rotation, auditor industry specialization, and audit firm size also had not. The auditor’s independence and adequate audit period would allow the auditor to be more effective in fraud detection.


INTRODUCTION
As business has continued to grow in complexity, the practice of crime has also developed in the form of economic fraud (Anggraini et al., 2019).One case of fraud is financial statement fraud which has resulted in enormous losses to many parties.Furthermore, fraudulent financial statements can have a significant impact such as causing harm to the company, the willingness of investors, etc.That harmed the national economy (Albrecht et al., 2014).A study from ACFE Indonesia (2019) showed that fraud cases in Indonesia, there were in total of 239 cases varying from 167 corruption cases, cases of 50 asset abuse, and 22 accounting fraud cases with the amount of the loss is Rp873.430.000.000.In Indonesia, fraud cases often occur both in the public and private sectors because of the technology that has become increasingly modern and there are techniques used by criminals in search of an opportunity to commit fraud (Aksa A.F, 2018).Financial cases that dragged the auditor showed many cases of auditor failures in detecting fraud such as corruption cases at PT Asuransi Jiwasraya and also PT Asabri.Financial statement manipulation fraud by Jiwasraya has been carried out for more than a decade (Sandria, 2021).There are also other cases, namely the mining company PT Garda Tujuh https://doi.org/10.23969/jrak.v15i2.8225Volume 15, No. 2, October 2023, Page. 259-271 Ika Sasti Ferina Dicky Pratama Buana Tbk (GTBO).The Company manipulated financial statements with an indication that the 2012 financial statements were not appropriate (Nabhani, 2013).Therefore, this fraudulent practice should never occur in the business company or during the audit process because the trust of the external parties would be realized by providing credible accountability to them.
The spotlight of this research is to inquire whether auditor traits and gender diversity affected the detection of fraud.This is based on problems that occurred in the case of auditor failure.Characteristics of the auditors reviewed in this research included audit tenure, audit rotation, auditor independence, auditor industry specialization, audit report lag and audit firm size.Gender diversity is added as a new variable as female companies' higherups are considered as factors in fraud detection, reducing the tendency for companies to engage in fraudulent activities (Wang et al., 2022).Auditor characteristics and gender diversity would be assessed on how they would affect fraud detection in a company.
This research adopted the study conducted by Khaksar (2022).The discrepancy between this research and the preceding study is that this appended new variables and a range of research periods.The new independent variable added is gender diversity.Women directors possessed the quality of being aware of risk and complying with higher ethical standards.Therefore, a board of directors having many woman directors meant that the company could detect fraud.The period used was previous research from 2012-2018 while the period of this research was from 2016 to 2020.
The objective of this research is to know the influence of auditor characteristics and board director gender diversity on fraud detection so that companies would be able to enhance fraud detection.The agency theory was sparked by Jensen andMeckling in 1976. According to Jensen &Meckling (1976), the agency relationship is defined as an event that occurred when one or more persons (principals) hired certain (agents) to offer service and entrust the authority of decision-making to others.Shareholders acts as a principal and management that manages the company acts as an agent.Management is the party given the contract by the shareholders to work for their interest.Management is given a portion of the ability on decisions for shareholders' interest, therefore management must take responsibility for all their commitment to shareholders as quoted by (Putri et al., 2017).But in reality, symmetrical information never happened.The emergence of different interests between internal and external parties caused conflicts of interest (Silaban & Suryani, 2020).External auditors are elected as mediators between management interests and shareholders.
Gender socialization theory illustrated the discrepancy in moral values and ethical points of view between men and women (Dawson, 1992).This can be judged from the beginning of the life of a human being influenced by the upbringing of parents who regulated how stylish and behave, which would then affect children's nature and behaviour (Carter, 2014).Differences in ways to educate children based on gender from birth would also affect the child's ethical nature and behaviour in adulthood.Men are more confident in making decisions and are more rational because they tend to be educated to be bolder.Meanwhile, women who are educated by moral planting, affection, and fragility make her devout person in order, more concerned, responsible, and tend to avoid risk.This makes them have more ethical behaviour than men.
According to Ritonga et al. (2020), audit tenure is a period or length of engagement between employment auditor and client during the examination process of financial statements.Examiners who had a long time audit of tenure generally could easily understand the characteristics and business processes in a manner that simplified the identification of fraudulent activities (Primasari et al., 2019).However, Yuniati & Banjarnahor (2021) revealed that it did not affect long audit tenure and was not related to increasing the auditor's understanding of accounting fraud practices.Furthermore, Putri et al. (2017) explained that too long an audit tenure can cause closeness that affected the auditor's independence in reporting an error.An excessively long audit tenure would allow familiarity between the auditor and the client, which affected the auditor's independence and might cause the auditor to neglect their responsibilities on doing fraud detection.
Audit rotation is the change of independent audit of the company which is carried out regularly to cut the threat of intimacy where the auditor is too long related to the client (Suciana & Setiawan, 2018).Audit rotation aims at increasing the independence of public accounting firms both in appearance and facts and be able to give new first chances to new auditors to investigate clients with better supervision and be careful (Paputungan & Kaluge, 2018).Audit rotation did not affect fraud since external auditors seldom revealed the fraud in the company on the auditor's opinion report.Moreover, auditor rotation carried out by the company is only for the formality in following the applicable regulations (Ratmono et al., 2020).Audit rotation is conducted to maintain the auditor's ability to be objective and independent.When the objectivity and independence of auditors are upheld, auditors are more likely to find fraud in a company.
Independence is an auditor's act of being out of control or under the influence of other parties and not depending on other individuals (S.Andini et al., 2021).The attitude of the auditor's independence can hinder relationships with clients who are likely to be able to disturb the objectivity of the auditor when doing their obligations (Fransisco et al., 2019).According to Astuti & Sormin (2019), an auditor's independence is one of the very important and fundamental factors for the auditor when carrying out the audit engagement.According to Arens et al. (2015), in almost all engagements, the auditor must protect its mind and appearance.The independence that the auditor had, may affect fraud detection because being under pressure would not be happening with being independent.So that the auditor would be more objective due engagement audit period.As a result, auditors would improve their performance in detecting fraud (Sanjaya, 2017).Otherwise, Independence did not determine the responsibility of the auditor to detect fraud because it is not the most influencing factor in the fraud detection process, but it is also influenced by the knowledge and experience of the auditor (Pratiwi & Rohman, 2021).An auditor who remained impartial and carried out their responsibilities by applying independence in every audit process is more likely to recognize the existence of fraud.
According to Insani & Sulhani (2020), specialist auditors were people who have received training and have a long experience, most of which concentrated on specific client businesses and industries, as well as in-depth knowledge and understanding of company activities, special accounting and audit guidelines, which were crucial in carrying out audit quality tall.Many experiences made the auditor have a deeper understanding of businesses and other more specific information, such as how companies carried out operations, accounting policies used, as well as other things about the industry so that they are expected to provide good audit quality for the company (Ressidnarry & Sjarief, 2021).Auditor's industry specialization can provide accurate and reliable information.Therefore, specialist auditors are considered capable of disclosing misstatements of financial statements with full integrity and have more insight so that the auditor can understand the condition of the company and detect existing fraud more easily (Astrawan & Achmad, 2023).The more frequently an auditor audits a company in the same sector, the auditor generally became and gained more expertise and experience.This allowed the auditor to be more likely to detect fraud.Afifah et al. (2022) defined the audit report lag as the period required by the auditor to perform the audit procedure of a company's financial statements.The audit is a long-term activity, sometimes there may be a postponement of the announcement of income and presentation of financial statements.In addition, auditors who had more time during the audit report lag can detect fraud and mismanagement than auditors with shorter time on the audit report lag (Khaksar et al., 2022).Financial reports that are issued later than the specified time might be a guide to deception occurring in the business.Auditors with sufficient audit time in auditing financial statements tended to be more capable of detecting fraud.
An audit firm size can be defined by a variety of factors starting from the number of clients and income generated by the public accounting firm (Bagariang & Lubis, 2021).Audit firm size is usually divided into 2 categories, namely Big 4 public accounting firms and non-Big 4 public accounting firms (Fajrina & Rohkhayatim, 2021).Big 4 public accounting firm would maintain their positive reputation so they had to maintain their audit quality.So that, the Big 4 public accounting firm would prevent companies from committing fraud.Therefore, the Big 4 public accounting firms would be more careful in fraud detection than Non -Big 4 public accounting firms (Suryani et al., 2023).Wang et al. (2022) stated that women executives would portray greater ethical behaviour and a lower tendency for involvement in criminal activity and legal disputes than men.Women company leaders tend to adopt a cautious approach to financial reporting.When female company leaders are in charge of overseeing financial reporting policies, they tend to be more attuned to the potential for legal repercussions and any possible risks.Furthermore, being trustworthy and compliant with regulations is something that female leaders perceived while making financial decisions.Female leaders' ability to comply with rules would raise awareness of potential risks or fraud that can happen company.Their ethical value and compliance can contribute to stronger overall fraud prevention and detection within a company.
The identification of misappropriation is an undertaking to gather important premier indicators about corruption actions, whereas narrowing a crime space is when the culprit feels that the act has been discovered, so it is too late to avoid avoiding (Kumaat, 2011).Instructions on the ongoing fraud can be seen through its symptoms, namely, increasingly weakened internal control, accounting deviations, excessive lifestyles, analytical deviations, tips and complaints and unusual behaviour.

METHODS
This research targets the influence of auditor characteristics and gender diversity on fraud detection in mining sector companies listed on the Indonesia Stock Exchange (IDX), and the research period is from 2016 to 2020.It's been 5 years.According to ACFE, in 2019 the mining sector was the third most affected by fraud.Auditor characteristics used several variables as independent variables.Specifically, audit duration, audit rotation, auditor independence, auditor industry specialization, audit reporting delay, audit firm size, and an additional variable, namely gender diversity.In this research, the author used a quantitative approach and The first year of the KAP engagement begins with number 1 and is coupled with one for the following years (interval scale).
( W e r a s t u t i , 2013a) 3 Audit Rotation Periodic independent audit rotation If there is an audit rotation then the value (1); If it doesn't exist, then the value (0) (Dummy variable).(Ardani, 2017) 4 Auditor Independence Auditor behaviour that is independent of the influence of others, if the company change auditor every 2 years then it showed their independence Companies replacing the auditor every 2 years will be given 1, and companies using the same auditor for more than 2 years are given 0 (dummy variables).(Nirmalasari & Sapari, 2022) 5 Auditor industry specialization Auditor with a lot of experience in certain industries.
Specialist industry will be given 1, and 0 for not using specialist services (dummy variables). (

RESULTS
A targeted sampling method was used for sampling, resulting in a total of 205 study samples over 5 years.Documentation is a type of data collection method that is performed manually on the official website of the Indonesia Stock Exchange (https://www.idx.co.id/) and the official website of each company.
The Effect of Auditor Characteristics and ... https://doi.org/10.23969/jrak.v15i2.8225 As presented in Table 2, information can be obtained about the research variable, namely the detection of fraud in 205 units of analysis had a minimum value of 0.00 a maximum value of 1.00 and an average value (mean) of 0.341 or 34.1%.Furthermore, the standard value of deviation of the fraud detection variable is 0.475.Fraud detection's standard deviation had a higher value than the mean, which meant fraud detection had a high variety of data distribution.Audit tenure variables in 205 units of analysis with 1.00 as the minimum value 5.00 as the maximum value and an average value (mean) of 2,619.Furthermore, the standard value of the deviation of the tenure audit variable is 1,390.Audit tenure's standard deviation had a lower value than the mean, which meant audit tenure had a low variety of data distribution.The audit rotation variable in 205 units of analysis with 0.00 as a minimum value 1.00 as a maximum value and an average value (mean) of 0.478 or 47.8%.Furthermore, the standard value of the deviation of the audit rotation variable is 0.500.Audit rotation's standard deviation had a higher value than the mean, which means audit rotation has a high variety of data distribution.The auditor's independence variable in 205 units of analysis has 0.00 as a minimum value 1.00 as a maximum value and an average value (mean) of 0.834 or 83.4%.Furthermore, the standard value of the deviation of the auditor's independence variable is 0.372.The auditor's independence's standard deviation had a lower value than the mean, which meant the auditor's independence had a low variety of data distribution.The auditor industry specialization variable in 205 units of analysis had 0.00 as a minimum value 1.00 as a maximum value and an average value (mean) of 0.151 or 15.1%.Furthermore, the standard value of deviation of the auditor's industry specialization variable is 0.359.The auditor industry specialization's standard deviation had a higher value than the mean, which meant auditor industry specialization had a high variety of data distribution.The audit report lag variable in 205 units of analysis with 31.00 as the minimum value 545.00 as the maximum value and an average value (mean) of 93,663.Furthermore, the standard value of the deviation of the audit report lag variable is 50,640.Audit report lag's standard deviation has a lower value than the mean, which meant audit report lag had a low variety of data distribution.Audit firm size variable in 205 units of analysis with 0.00 as a minimum value 1.00 as maximum value and an average value (mean) of 0.424 or 42.4%.Furthermore, the standard deviation value of the audit firm size variable is 0.495.The audit firm size's standard deviation had a higher value than the mean, which meant the audit firm size had a high variety of data distribution.The gender diversity variable in 205 units of analysis had 0.00 as a minimum value 0.50 as a maximum value and an average value (mean) of 0.120 or 12.0%.Furthermore, the standard value of the deviation of the gender diversity variable is 0.186.Gender diversity's standard deviation had a higher value than the mean, which meant gender diversity had a high variety of data distribution.
Table 3 showed the results of the panel data regression for the Common Effect Model, Fixed Effect Model, and Random Effect Model.As presented in Table 3, information was obtained about the comparison of regression results from the use of the three estimated models namely the common effect model, fixed effect model, and random effect model.Fixed effects and random effects models are the most common models used for panel data, especially when the number of cross-sections is relatively large, and the number of periods is classified as small as quoted (Alhababsah & Yekini, 2021).The Hausman Test was used to examine an indication of whether the model is a Fixed Effect or Random Effect Model that was suitable for the research.Based on the hausman test shown in Table 4, the test results stated that the probability value was 0.509 where (p-value)> 0.05 so that the H0 is statistically accepted and Ha is rejected.If H0 is accepted, the appropriate estimation model used in panel data regression was the Random Effect Model.
Assumptions in regression analysis include the absence of multicollinearity, heteroscedasticity, autocorrelation, and outliers.Because the selected model tested is the Random Effect Model, the autocorrelation test is not needed (Law, 2018).The Multicollinearity Test aims at determining whether the model of regression had correlating the independent variables.Multicollinearity will detected if the VIF value is more than 10 (VIF > 10).Based on the multicollinearity test presented in Table 5, the result showed that all independent variables had less than 10 with an average VIF of less than 10, so it confirmed that the regression model did not contain multicollinearity problems.The Heteroscedasticity Test is intended to examine whether there are similarities and variances from one observation to another.If the probability value > the signification value then in a regression model did not occur the heteroscedasticity.Based on the results of the modified Wald test for heteroscedasticity presented in Table 6, showed the probability value of 0,000 <0.05.This can be concluded that heteroscedasticity occurred in the tested regression model.To overcome this problem, robust would then be used in regression (Law, 2018).The Effect of Auditor Characteristics and ... https://doi.org/10.23969/jrak.v15i2.8225 The Outlier Test aims at testing for the presence of outliers or too much large or small data which impacted the undistributed residuals normally and caused the bias in the regression.This research used Cook's Distance Outlier Test.Based on the outlier test presented in Table 7, there were 6 outliers in the data.Outlier-indicated data would be eliminated from calculations to avoid data that tended to deviate (Law, 2018).The company that indicated the outlier and eliminated from the samples were PT.Bayan Resources, Tbk (2019-2020), PT.Exploitasi Energi Indonesia, Tbk (2019), PT.Vale Indonesia, Tbk (2016), PT.Petrosea, Tbk (2020), and PT.TBS Energi Utama, Tbk (2018).The results of the research hypothesis test were carried out by interpreting the Robust Random Effect model after the outlier because the model was a selected model that was free from multicollinearity, heteroscedasticity and outlier problems.Table 9 presented the results of the hypothesis test with the Robust Random Effect model after the outlier, i.e.: Based on Table 9, showed the results of data panel data regression that the research model obtained as follows: Fraud it = 0,206 -0,039TENURE it -0,004ROTASI it + 0,201IND it -0,108AIS it + 0,002ARL it -0,139KAP it -0,36GD it + ε it The robust Random Effect Model's result of regression after the outlier showed that the R-Square value is 0.325.Independent variables can affect the dependent variable by 32.54%, while the remaining 67.46% is explained by other factors outside the variable studied.
Based on Table 9, it is presented the probability value indicating the number of 0,000 which was smaller than the significance value (0.05) also meaning that the research model is suitable for use.This also meant that the dependent variable could be predicted by the independent variable.
This research used a 10% significance test, not 5% because producing a significant variable if using a significance of 5% only found a significant variable because most P-Value> 0.05.Therefore a significance level of 10% or 0.10 is used.
The hypothesis result can be interpreted by looking at The T-test results.The first variable called audit tenure, yielded a coefficient value of -0.0.038 and a p-value of 0.052.This result implied that the audit tenure harmed fraud detection, so the H1 is accepted.Meanwhile, the hypothesis of the second variable was rejected because audit rotation had a coefficient value of -0.003 and a p-value of 0.958.
Furthermore, the hypothesis testing results for auditor independence showed coefficient index values of 0.201 and 0.004 as p-values.This result implied that auditor independence had a positive impact on fraud detection, so the study of H3 is accepted.Whereas, the fourth variable called auditor industry, did not affect fraud detection and rejected H4.
On the other hand, the audit report lag variable had a coefficient value of 0.001 with 0.010 as the p-value.This result implied that the audit report lag had a positive effect on the detection of fraud and acceptance of the H5.Unfortunately, audit firm size did not affect the detection of fraud and rejected H6.A similar result on the last independent variable called gender diversity, the hypothesis test resulted in Table 9 indicating that the coefficient value of GD was -0.360 with a p-value value of 0.040.This result implied that gender diversity harmed the detection of fraud and also rejected the H7.

DISCUSSION
Hypothesis test results showed that audit tenure harmed the detection of fraud.The results of this research are found to be consistent with the results of the research conducted by Ritongan et al. ( 2020) and Carcello & Nagy (2004) which stated that audit tenure harmed the detection of fraud.The longer the period of relationship between auditor and client, the more likely an occurrence of fraud would happen, this is proven by the number of cases where a long audit tenure did that, for example in the British Telecom scandal and Price Waterhouse Coopers (PWC) (Ritonga et al., 2020).Furthermore, engaging with the auditor for too long would threaten the objectivity of auditor work and also tend to the potential relationship outside of work between the auditor and the auditee which can lead to a conflict of interest that would affect an auditor's independence, so the auditor would not courage to reveal the actual situation happened on the client's company.Following agency theory, audit tenure would affect the auditor's responsibility for identifying fraud (Pratiwi & Rohman, 2021).This meant that the longer the tenure period, the more difficult it is to detect fraud.
The hypothesis test also showed that audit rotation did not affect the detection of fraud.The result of the research was consistent with the results of the research conducted by (Werastuti, 2013) which stated that audit rotation did not affect the detection of fraud.Whether or not the auditors carried out the audit, the auditor's scepticism was one of the factors in fraud identification.Audit rotation must be done but the audit rotation or not, the auditor will continue to reveal all things that happen within the company because the auditor still maintains its independence (Wahyu, 2020).In other words, depending on the independence of the auditor conducting the audit, fraud in financial statements may be discovered during the audit.The agency theory stated that there were differences between the interests of the agent and the principal explaining that the management as an agent would be doing the rotation of the auditor for personal interest (Silaban & Suryani, 2020).Moreover, doing the audit rotation would make a renewal, but the auditor was not necessarily capable of directly understanding the client company environment.
The auditor's independence had a positive effect on the detection of fraud.The results of the research conducted were consistent with the results of the research conducted by Laloan et al. (2021), Dasila et al. (2019) and Ni Putu Intan Eka Sari & Komang Fridagustina Adnantara (2020) which had results of positive The Effect of Auditor Characteristics and ... https://doi.org/10.23969/jrak.v15i2.8225impact between auditor independence and fraud detection.An independent auditor in the implementation of an audit had confidence, ability, and expertise and is honest and fair.In terms of reporting, an auditor conducting auditing must be free from client pressure.This meant that if an auditor had maintained independence when auditing and did not care about external pressure, the auditor had high integrity.Due to agency theory, the clients might have different interests, maybe even conflicting with the interests of the financial statement users.Therefore, the auditor must behave independently of the client's interest (Ni Putu Intan Eka Sari & Komang Fridagustina Adnantara, 2020).The greater the independence of the auditor was, the greater the ability to detect fraud would be.
Hypothesis test results showed that the auditor's industrial specialization did not affect the detection of fraud.The results of the research conducted were consistent with the results of the research conducted by Ressidnarry & Sjarief (2021) which stated that auditors of industrial specialization did not affect the Fraglement of Financial Reporting.The client company had a tendency not to use specialist auditors and preferred non -specialist auditors because they had the same good quality.In the context of mining companies, data observations from 2016-2020, showed that most mining companies tended not to elect industry specialization auditors.This is indicated by a total of 205 observation samples only 15% chose auditors in industrial specialization, while the rest chose non -industrial specialization auditors.The use of industrial specialist auditors did not have a definite impact on detecting fraud.This meant that if the auditor was industry-specialised but did not maintain the attitude of independence, it was still difficult to reveal the fraud.Both professional and non-professional auditors must follow similar rules and follow the same guidelines, so they have the same opportunity to represent the company's situation and detect fraud.(Pramaswaradana & Astika, 2017).
Audit report lag had a positive effect on the detection of fraud.The findings of this research are under the results of the research conducted by Suryanto (2016) which stated that the audit report lag had a significant effect on the detection of the Fraudulent Financial Reporting System.Audit report lag is described as the time required for an auditor to conduct an audit of a company's financial statements, giving the auditor enough time to examine the financial reports to prevent financial statement fraud.Agency theory stated that too long submission of the financial report can result in asymmetry information that showed when the agent was more aware of internal pieces of information and the company's outlook than the principal and allowed the agent to use the information provided to them to manipulate the financial statement (Mahawyahrti & Budiasih, 2017).In the context of mining companies, based on data observations from 2016-2020 it showed that most of the average financial statement delivery was within a range of 94 days or 3 months.In rules, this was still fairly normal and under normal conditions, it provided enough time for the auditor to detect fraud.
The hypothesis test showed that audit firm size did not affect the detection of fraud.The findings of this research followed the results of the research conducted by A. Randa & S. Dwita (2020) which stated that audit firm size did not affect the auditor's ability to detect fraud.Big 4 public accounting firms were inseparable from the failure of the auditor in detecting fraud as evidenced by the existence of Multiple financial scandals entangling public accounting firms, such as Deloitte Indonesia's inability to find fraud-proof at SNP Finance.Under agency theory, big public accounting firms had deep levels of work and more clients, so they would be less dependent on specific clients and have greater incentives to audit quality (Paputungan & Kaluge, 2018).Auditors affiliated with Big 4 public accounting firms and Non -Big 4 public accounting firms used the same audit procedure, so the opportunity to produce quality audits quality and the ability to find fraud is also the same (Zulfikar & Waharini, 2019).This meant that the auditor who joined the Non-Big 4 public accounting firms would also report if there was an error or fraud to maintain an independent attitude.
Gender diversity harmed the detection of fraud.The findings of this research were under the results of the research conducted by Indiraswari (2021) which stated that gender diversity harmed fraud on financial statements.The existence of women's directors in a company can reduce cases of fraud in financial statements.Moreover, a woman director had more responsibilities and predictably detected and reduced fraud on the financial statements.This can occur because women had more cautious properties, prevented risk, and had a higher ethical approach to reducing cheating behaviour in the company (Setyaningrum et al., 2019).Following gender socialization theory, woman leaders can identify and reduce financial statement fraud.In the context of mining companies, data observations from 2016-2020 showed that most companies preferred to lead or be on the board of trustees.The percentage of gender diversity was only 12% which showed the dominance of men and at least women who occupied the position of the board of directors.This showed that the lower gender diversity in a company, it tended to be more capable of detecting fraud.

CONCLUSIONS
This research examined the effect of auditor characteristics and gender diversity on fraud detection.Auditors' characteristics such as independence and audit report lag can positively affect fraud detection.Furthermore, audit tenure harmed the fraud detection.Whereas audit rotation, the specialization of the auditor industry and audit firm size did not affect the detection of fraud.Besides that, gender diversity harmed the detection of fraud.Audit tenure and gender diversity harmed fraud detection that the longer the auditor audited a company and had a woman leader, it would affect the process of identifying fraud.Audit rotation, industry specialization, and firm size did not affect fraud detection, as more auditors with other characteristics such as independence can be a better factor in fraud detection.An auditor having independence and an adequate audit period would allow the auditor to be more effective in identifying fraud in companies.This research had some limitations.One is that while the research focused only on the mining sector, other sectors can also be investigated.This research only showed that 32.54% of the independent variables had an impact on fraud detection, and the remaining 67.46% were influenced by factors other than investigated variables.This research also used M-Score Beneish measuring devices as a proxy for fraud detection variables but had weaknesses because the model was a probabilistic model so it did not predict the fraud of financial statements.
In addition, it is recommended that other sectors be explored and used as samples such as the financial and banking industry sectors for future purposes.Research in the future can also explore other variables such as audit opinion, changes in directors, external pressure, share ownership, quality of external auditors, and others.This research used the Beneish M-Score as a proxy variable for detecting fraud.However, this measurement tool had its limitations because it's a probabilistic model, and as such, it cannot predict the occurrence of financial statement fraud with 100% certainty.Further research on the use of other measures such as the Piotroski F-score for fraud detection variables is expected to assess the accuracy of the measurements.This research is limited by only doing data documentation as a form of data collecting.Further research might also consider alternative research methods such as conducting interviews or retrieving data directly from the Financial Services Authority, especially information that is related to getting maximum outcomes.

Table 1 .
Operational Definition and Variable Measurement

Table 8 .
Random Effect Model Outlier and Robust Random Effect Model Outlier