Skip to content

The Audit module is developed to facilitate statistical auditing in both Bayesian and classical manifestations. The main feature is a workflow that helps guide auditors through the sampling process. Additionally, there are stand-alone analyses for planning, selecting, and evaluating a sample. The module is based on the R package jfa.

Notifications You must be signed in to change notification settings

jasp-stats/jaspAudit

Repository files navigation

The Audit Module

JASP for Audit (also known as the Audit module) is an add-on module for JASP that streamlines the process of statistical auditing. The Audit module offers a wide range of functionalities, including (but not limited to) planning, executing, evaluating, and documenting statistical audit samples. Specifically, it comprises analysis tools for determining sample sizes, selecting items using standard audit sampling methododology, and inferring the population misstatement based on the sample data or the summary statistics of the sample. Furthermore, the Audit module also includes tools for data auditing. The module provides Bayesian equivalents of most analyses, enabling users to incorporate pre-existing audit information into the statistical procedure. In all analyses, the Audit module provides comprehensive explanatory text that assists the auditor in understanding, documenting and communicating the statistical results.

Blog Posts

Articles

In order to familiarize oneself with the Audit module, it is recommended to obtain and read the manual via this download link, or to view it online through this link. A reference for the Audit module can be obtained from the accompanying academic article published in the Journal of Open-Source Software, which can be accessed through the following permanent link https://doi.org/10.21105/joss.02733.

DOI

R Package

The Audit Module is a graphical user interface that has been constructed to integrate with the jfa R package available on CRAN, which offers a comprehensive set of statistical methods for auditing. To obtain a deeper understanding of the underlying statistical principles and implementation details of the jfa package, interested parties are encouraged to consult the package website or read the accompanying textbook.

jfa

Benchmarks

The Audit Module conducts numerous daily unit tests to validate the output of the software against several well-known and established benchmarks. These benchmarks encompass, among others:

R_build_status

Analyses

The organization of the analyses within the Audit module in JASP is as follows:

--- Audit
    -- Audit Sampling
       - Sampling Workflow
       - Bayesian Sampling Workflow
       - Planning
       - Bayesian Planning
       - Selection
       - Evaluation
       - Bayesian Evaluation
       - True Value Estimation
    -- Data Auditing
       - Benford's Law
       - Repeated Values
    -- Algorithm Auditing
       - Fairness Metrics

Audit Sampling

The Audit Sampling section of the module makes it easy for an auditor to plan a statistical sample, select the sample from the population, and evaluate the misstatement in the sample compliant with international auditing standards. The analyses also provide auditors with easy documentation of their sampling procedure and statistical results. This comprehensive approach not only streamlines the audit process but also enhances transparency and efficiency.

Data Auditing

The Data Auditing section of the module provides statistical methods for auditing data, including (Bayesian) tests of digit distributions and tests for repeated values. These methods can be used to identify anomalies in a set of numbers, but do not provide conclusive results in terms of whether a data set is likely tampered with. For instance, non-conformance to Benford's law can only be seen as an indication of possible data manupulation. Unlike Benford's law, the other analysis is this section examines the entire number at once, not only the first or last digit(s).

Algorithmic Auditing

The Algorithm Auditing section of the module includes methods for auditing algorithms on the aspect of fairness and discrimination. For instance, the fairness metrics analysis aims to assess fairness in algorithmic decision-making systems by computing and testing the equality of one of several model-agnostic fairness metrics between protected classes. The metrics are computed based on a set of true labels and the predictions of an algorithm.

Changelog

Please find the module's changelog by clicking here. This changelog provides a detailed record of all the modifications and updates made to the module. It can be useful to stay up-to-date with the latest changes and bug fixes, or to check if a particular issue has been addressed.

Translations

The JASP translation project is located at the following link. The Audit module is referred to as jaspAudit and consists of two distinct components: jaspAudit-QML and jaspAudit-R, both of which require translation. The participation of any individual who is interested would be highly valued. As demonstrated below, we have made substantial progress in the translation of the Audit module.

Interface Results
image image

About

The Audit module is developed to facilitate statistical auditing in both Bayesian and classical manifestations. The main feature is a workflow that helps guide auditors through the sampling process. Additionally, there are stand-alone analyses for planning, selecting, and evaluating a sample. The module is based on the R package jfa.

Topics

Resources

Stars

Watchers

Forks

Packages

No packages published