Opportunities for the use of classification methods to support sampling procedures in audits

Abstract

The target of my article was to show how to support the State Audit Office of Hungary’s (SAO) work in several areas by using classification methods in order to improve the quality of SAO’s reports.
I presented the directives and guidelines of ECA, GAO, INTOSAI and NAO related to sampling procedures and stratified sampling. I described the stratified sampling in the next section. After this I provided a summary of literature in connection with classification methods focusing on the potential application area of cluster analysis and discriminant analysis in SAO’s audits.
I showed the process of database construction in connection with child welfare services, using cluster analysis. After amending the databases with new metric and binary variables I presented the results of cluster analysis executed in two databases with various hierarchical, nonhierarchical and TwoStep clustering methods for the year of 2015 using IBM® SPSS®® Statistics statistical analysis software (version 23). In the end of the section I pointed out that in my opinion this result could support the SAO in course of stratified sampling. In my point of view the greatest challenge in the procedure of cluster analysis was to find the most adequate relevant metric variables connecting to audit and cluster analyzing goals and to collect data. Concerning the results of cluster analysis I presented some applying possibilities of discriminant analysis.
In the end of the article I summarized my recommendations. I strongly believe that applying classifical methods – primarily cluster  analysis with discriminant analysis – can assist SAO’s audit in several stages and analysis too, – including risk analysis - which creates added value and thereby intensifies SAO's contribution to the „good governance” according to SAO’s strategic objectives.