The contents of the National Audit Office of Finland performance audits, 2001–2016: An interpretive study with computational content analysis

Forskningsoutput: TidskriftsbidragArtikelPeer review


Purpose – This is a qualitative, computationally assisted examination of prominent content patterns in the 2001 to 2016 National Audit Office of Finland (NAOF) performance audit reports and foremost changes in these patterns.
Design/methodology/approach – Without ex ante researcher decisions on which content elements are important, we use computational content analysis of topic modeling for detecting content patterns and their changes in the performance audit reports.
Findings – In the content patterns, professional effectiveness auditing stands out, whereas efficiency auditing is more weakly present. Other content patterns correspond to the NAOF institutional position and its audit mandate, the institutional characteristics of Finland’s public sector, and independent NAOF decisions in orienting its work. We can discern both ascendant, ascendant and stabler content patterns.
Research limitations/implications – As the article emphasises textual analysis, examining from where changes in the content patterns derive is secondary. Implications for both more intensive and extensive future research are drawn.
Practical implications – We argue that together with other approaches our approach has potential in monitoring SAI performance auditing in the level of its contents.
Social implications – Knowing how an SAI orients its performance auditing has potential to support SAI monitoring by stakeholders – Parliament, the government, the citizens, and others.
Originality/value – Ours is a comprehensive examination of the content patterns of the performance audit reports of an SAI since the 2001 introduction of its present institutional position.
Keywords Efficiency, Effectiveness, Evaluation, Government auditing, Artificial intelligence
Paper type Research paper
TidskriftJournal of Public Budgeting, Accounting & Financial Management
Sidor (från-till)49-66
Antal sidor18
StatusPublicerad - 2020
MoE-publikationstypA1 Tidskriftsartikel-refererad


  • 517 Statsvetenskap
  • 113 Data- och informationsvetenskap

Citera det här