Understanding Auditing as an Information Science

Sunday, February 14, 2016
Eo Jin Lee, George Washington University, Washington, DC
Background: Auditing can be viewed as an information science, pulling in relevant data from across a business entity for evaluation of its health in the past as well as the projected health in the future.  An evaluation of the flow of information within a business and the robustness of its internal processes can identify weaknesses.  However, the vast amount of information available to an auditor may become more of a hindrance than a help when information overload becomes an issue.   Beyond a given threshold, additional information can begin to have deleterious effects and any information beyond this threshold is either ignored or potentially causes unintended confusion.  Currently auditors utilize statistical sampling to minimize information overload, but techniques like anomaly detection using big data may provide a more informative audit. Methods: We survey the information technology and capabilities within the auditing field and compare to common data science techniques to look for areas that may be susceptible to information overload.  Computer Assisted Audit Tools (CAAT) like ACL perform some of the routine tasks in an audit including data collection and formatting, as well as providing a platform for audit checks.  Audit expert systems provide a series of production checks to validate the data and catch common mistakes and oversights within an organization.  Further statistical analysis is often left to the individual auditor utilizing analysis packages that are not tailored specifically to an audit. Results: We analyzed auditing techniques for handling information to determine potential sources of information overload.  Computer Assisted Audit Tools (CAAT) like ACL can eliminate a large source of information overload by properly entering information and performing routine checks that can be run on a transactional basis, allowing simple mistakes to be caught and fixed early.  Audit expert systems when well developed can help alleviate information overload but can require significant individual tailoring and expertise.  Statistical analysis is often done using sampling techniques to minimize the information load, and here is where the information moves from a specially developed auditing tool into more general statistical packages.  We found that many larger firms have moved into a model that utilizes CAAT for internal auditing without significant information overload on their auditors. Conclusions: We determine that auditing information science is capable of handling data overload through statistical sampling, and that many larger firms have moved into a model that utilizes CAAT for internal auditing without significant information overload on their auditors.  Utilization of statistical analysis and data visualization within CAAT may also improve interactions with auditors.