Dimensions of Auditor Independence : A Pilot Study

This paper reports a pilot study of Nigerian stakeholders about their perceptions of the dimensions of auditor independence (AI). The validity and reliability of the instrument was examined through a pilot survey of experts and informed users of financial statements. Data normality was also assessed using SPSS 18 software. The results affirm the instrument’s validity and reliability and the sample data showed reasonable normality. The study explored and validated an instrument of both dimensions of auditor independence which have often been examined only in exclusion by prior studies.


Introduction
Auditor independence (AI) is often described as the cornerstone of auditing and very fundamental to corporate financial reporting (Mautz & Sharaf, 1964;Previts & Merino, 1998).This is because the essence of audit entails providing reasonable and objective assurance that financial statements reflect the true state of affairs of a business thereby enhancing the credibility and reliability of the financial reports.Thus stakeholders (shareholders, potential investors, regulators, creditors, financial markets and the public) relying on the objectivity and integrity of such reports are able to make informed decisions about investments.Some studies have also shown that stakeholders' perceptions of auditor independence significantly influence the efficiency of the capital market because when shareholders and investors perceive higher risks on investments, they are more likely to demand higher returns on investments and cost of capital (Carmichael, 1999;Quick & Warming-Rasmussen, 2009).
The International Federation of Accountants (IFAC) has categorized AI into two dimensions; Independence in fact (IIF) and independence in appearance (IIA).While IIF is the state of mind which allows the auditor to carry out an audit with objectivity, integrity and professional skepticism, IIA entails avoiding circumstances that may make informed users having knowledge of all facts doubt the auditor is capable of forming an objective opinion.All professional codes of conduct require professional accountants to be independent in fact and appearance in the conduct of audit assignments.However, independence in fact is difficult to observe so regulators, investors and other financial statement users mostly examine appearances of independence by considering whether or not circumstances that may likely compromise an auditor's independence do exist as well as the safeguards in place to mitigate such threats.AI has received much regulatory and scholarly attention in the wake of the corporate collapses of the early 21 st century and the subsequent loss of confidence in auditors and financial statement credibility.Although most of the collapses resulted from corporate governance lapses, AI was also a major issue.For example, Brown (2005) reported that the auditors of Enron received substantial fees from their client that engendered a self interest threat that could have undermined their objectivity.Similarly, other studies (e.g., Banyard, 2002;Owen, 2003;Stempel, 2009) also assert that the auditors of WorldCom, HIH Insurance in Australia and Parmalat in Italy failed in exercising sufficient professional skepticism and accommodated various circumstance that engendered familiarity and self review threats that may have marred their ability to report fairly on their clients' true financial condition.In general, the corporate collapses have had an enormous negative effect on the capital market, public confidence, value of financial statements, audit function and auditor reputation.Various regulatory frameworks have focused on improving AI by strengthening the legal, regulatory and governance cultures to ensure close monitoring and compliance to standards and independence frameworks in order to avert corporate failures.One way of doing this is by regularly monitoring compliance to independence standards through a comprehensive way of measuring AI.Although there are many studies on measuring AI, combining both dimensions of AI in a single measure has been ignored.This study seeks to bridge this gap.
Auditors owe stakeholders (shareholders, creditors, governments, business and financial community, investors and the client) a fiduciary duty to be independent, objective and honest by providing reasonable assurance that financial statements are true and fair.Since stakeholders may have divergent interests, auditors need to have integrity and be objective in rendering their attest functions.There has been much controversy over the relative importance of appearances of independence and factual independence in determining AI and its inclusion in AI framework.For instance, some scholars (Wallmann, 1996;Elliot & Jacobson, 1998) consider factual independence superior and posit that difficulties in measuring independence appearances, lack of consensus on factors and relationships affecting AI as well as whose perceptions constitute reasonable perceptions relegates the significance and inclusion of perceptions in AI frameworks.Yet others (Carmicheal, 1999;Dopuch, King, & Schwartz, 2003) argue that the difficulties of measuring an auditor's mental state of mind as implied by factual independence necessitate the need for evaluating stakeholders' assessment of AI as auditors add justified credibility to financial reports whether or not material misstatements are detected since audits represent significant assurances of the reliability of accounting information thereby enhancing its credibility (Carmicheal, 1999).However, the regulatory frameworks require auditors to possess both forms of AI in the conduct of attest functions.
The independence in fact (IIF) domain entails acting with integrity, objectivity and professional skepticism in the conduct of audit assignment (Chapple & Koh, 2007).Integrity means maintaining an honest character, client confidentiality, due care and upholding the public trust above personal gains.According to Brown, Stocks and Wilder (2007), acting with integrity entails acting with all honesty and fair dealing, observing ethical and technical standards, client confidentiality and resisting subordination of judgment or circumvention of standards.The American Institute of Certified Public Accountants (AICPA) also charges auditors to maintain objectivity by continuously assessing client relationships with public responsibility.Professional skepticism entails the auditor approaching an audit with due care, an inquisitive mind and not knowingly misrepresenting facts or subordinating professional judgment to client choices.The International Standards on Auditing (ISA) define professional skepticism as approaching an audit with an enquiring mind and critically assessing audit evidence.This may entail soliciting further information by performing more tests to obtain additional audit evidence and reduce auditor perceptions of material misstatements in the accounts (Shaub & Lawrence, 1996).
The independence in appearance domain (IIA) largely depends on users perceptions about auditors based on avoidance of circumstances that may engender threats to AI and how the threats are mitigated by safeguards to levels that may no longer impair AI.This requires the evaluation of the five major threats to AI (self interest, self review, client advocacy, intimidation and familiarity/trust threats) in line with applicable safeguards to see that threats are eliminated or reduced to acceptable levels.The IFAC provides that threats arise from circumstances such as direct or indirect material financial interest in client, provision of non-audit services, economic dependence on client, loan to or from client, contingent fees or unpaid fees, business relations with client and prior or potential employment with client, promoting client interest as shares, advocating for client by providing litigation services or acting on behalf of client in dispute resolution with third parties.Other sources of threats include threatened dismissals and litigations, client pressure to reduce extent of audit work in order to reduce fees, having close family ties with client employees, lengthy audit tenures or acceptance of material gift and hospitality from client.
According to Gay, Mills and Airasian (2006) a pilot test is a small mini trial of a proposed study that is carried out before the full scale study in order to test the instrument as well as obtain an insight into the likely conditions of the intended study.In line with this, a pilot study was conducted to attain two major objectives; to validate the proposed measure and test its reliability and to gain an insight about the proposed study.Kimberlin and Winetrstein (2008) also affirm that measuring abstract concepts requires the operationalization of such constructs in defined variables and developing or applying instruments to test the variables.Generally, the major indicators of an instruments quality are the validity and reliability measures (Kimberlin & Winetrstein, 2008).Additionally, Sekaran and Bougie (2010) explain that an instrument's reliability shows the stability and consistency with which a concept is measured i.e. there is a consistency across time and among the items measuring a particular concept.Validity on the other hand refers to the extent to which a measure or instrument accurately represents the concept intended i.e. ascertaining that an instruments actually measures what it set out to measure (Sekaran & Bougie, 2010;Hair, Black, Babin, & Anderson, 2010).In line with this, this paper reports the result of a pilot study about Nigerian stakeholders' perceptions of the dimensions of AI.

Methodology
Because this is a pilot study of an ongoing research, only a small sample of stakeholder respondents are presented and analyzed.According to Malhotra (2008) a pre-test sample is usually small and could range from 15-30 respondents but could be higher in tests having several stages.For this study, seventy two instruments were distributed to auditors and other financial statement users.Sixty five questionnaires were retrieved of which five instruments were discarded because they were not completed properly leaving a total of sixty usable questionnaires.Kimberlin and Winetrstein (2008) assert that Cronbach's alpha coefficient is the most widely used test for estimating inter-item internal consistency which shows the average inter-correlations of items measuring a concept.Following this, the study uses the Cronbach alpha to assess the internal consistency of individual variables captured in the instrument.Data collected was analyzed using SPSS 18 for windows.The pilot study was conducted within a month from the first week of November to first week December, 2013.

Measurements and Instrumentation
The instrument used consisted of multiple choice questions which were close ended measured over a five point likert scale ranging from 1-strongly disagree to 5-strongly agree.The choice of scale is supported by various studies (e.g.Krosnick & Fabrigar, 1997;Beattie, Brandt, & Fearnley, 1999) as being more reliable and not forcing the respondents to respond in a particular direction thereby increasing measurement error.This choice is also in line with prior interdependence studies examining perceptions of AI (e.g., Alleyne, Devonish, & Alleyne, 2006;Al-Ajmi & Saudagaran, 2011).The two dimensions examined are Independence in Fact (IIF) which is evaluated by three variables; Perceived objectivity, Perceived integrity and perceived Professional skepticism and Independence in appearance (IIA) which is evaluated by six variables; self interest, self review, familiarity, advocacy, intimidation and safeguard implementation.The instrument is structured in three major parts.The first part is the introduction and request to participate in the survey.The second part presented the nine variables as measured by various questions/items (perceived objectivity-five items, perceived integrity-five items, perceived professional skepticism-thirty items, self interest-ten items, self review-five items, familiarity-ten items, advocacy-six items, intimidation-five items and safeguards implementation-thirteen items.The last part enquired about respondents' demographic information.

Content and Face Validity
According to Creswell (2012) content validity refers to the extent to which an instrument is adequately representative of the concept it seeks to measure.In other words, ensuring content validity means making sure that questions asked about a concept adequately cover the essence of the concept being measured.Researchers (e.g., Hair, Money, Samouel & Page, 2007;Sekaran & Bougie, 2010) have explained that this can be achieved by expert review of an instrument.In line with this, two professors of accounting and two professional accountants were given a copy of the instrument to determine its content validity.Also five PhD accounting students who can proxy for informed financial statement users and were familiar with auditing practices in Nigeria were given a copy of the instrument each to assess its clarity and understandability.The outcome of this procedure resulted in the rewording and rephrasing of some questions and minor alterations.Following the expert recommendations, the instrument was amended before being administered for the pilot study.

Reliability Tests
Reliability tests generally assess the extent to which items measuring the same concept mesh together consistently in measuring that concept.Studies have shown that the Cronbach alpha test is most commonly used to assess the internal consistency of items measuring particular constructs.Following this, the study also used the Cronbach alpha test to examine the instrument's reliability.Data from the pilot was keyed into SPSS version 18 for windows and each construct was evaluated by the items measuring it.According to Hair et al (2010) and Hair et al. (2007) the generally acceptable benchmark for Cronbach alpha is 0.70 and above although 0.60 is acceptable in exploratory studies.All constructs were found to possess acceptable internal consistency with Cronbach alpha ranging from 0.609 to 0.795 after some items were deleted for some constructs.For example, no item was deleted from perceived objectivity, one item from perceived integrity, seventeen items from perceived professional skepticism, one item from self interest threat, four items from familiarity threat, none from intimidation and self review threats, one item from advocacy threat and five items from safeguard implementation.From a total of 89 items, 26 were deleted leaving 63 items remaining.
Table 1 shows the results of the reliability test using Cronbach alpha co-efficient after deletion and the number of items dropped.Since all alpha values are from 0.6 and above, it can therefore be presumed that all the constructs are reasonably reliable and items of each construct are consistent among themselves in measuring the construct.The descriptive statistics in Table 2 shows that there were more male respondents compared to female respondents.Furthermore, respondent educational qualification shows that over 61 percent of the respondents have acquired a higher degree and about78 percent have over ten years of experience using financial statements while 74 percent have completed at least seven or more accounting courses making them suitable to respond knowledgeably to the issues raised.Majority of the respondents constitute professional accountants (18.3%), bank loan officers (31.7%) and financial reporting council officers (26.7%) who are all primary stakeholders of audit quality.

Data Distribution
One of the foremost requirements of inferential statistics is that data should be normal (Tabachnick, & Fidell, 2007).Fulfilling the normality assumption implies that data should follow the normal bell shape curve and be symmetrical.Using SPSS 18, data normality can be evaluated through kurtosis and skewness.According to Tabachnick and Fidell (2007) skewness values should not be greater than 2 while kurtosis values should not be greater than 7 for data to be considered normal.Based on this benchmark, the pilot data can be considered as reasonably normal, as all items have skewness values below 2 and kurtosis values well below the 7 benchmark as shown in table 3.

Conclusion
The paper presented a pilot study of an ongoing research which examined the validity and reliability of the research instrument prior to the actual field work in order to gain more insight into the proposed study and the viability of the instrument to be used.As this is just a pre-test, further investigations will need to be conducted to achieve the research objectives and reach conclusions.However, in establishing content and face validity, minor amendments and rewording of the instrument items were made based on expert review opinions.In addition, reliability tests using the Cronbach alpha co-efficient revealed all items met the minimum criteria of 0.60 and above after some items were deleted.The kurtosis and skewness values were also within limits indicating that pilot data collected was fairly normal.