Making HRA a more consistent science

This paper reports a study to investigate the inconsistency potential of three Human Reliability Assessment (HRA) techniques. Human Error Assessment and Reduction Technique (HEART); Technique for Human Error Rate Prediction (THERP); and Justification of Human Error Data Information (JHEDI). The study included some preliminary work on consistency of usage of the techniques and partly on more formal analyses of the techniques themselves. For each inconsistency identified, a counter-measure was proposed: some of these were short-term measures, others required more long-term treatment. The results of the preliminary analysis fed forward into a trial of the most important short-term measures for consistency improvements in order to determine how HRA technique consistency and assessor practice could be improved. The candidate improvement measures were then fed forward to the final part of the project that developed both 'long' and 'short'-term consistency improvement measures for HEART, THERP and JHEDI into a guidance package for assessors. Specific and general guidance measures were distilled into a checklist that HRA practitioners can now use for ensuring consistency in their HRA approach and minimising individual assessor bias in their assessments. Additionally various 'qualifications' and 'training' measures, that are believed will lead to I-IRA being a more consistent discipline were identified.