Two models of raters in a structured oral examination: does it make a difference?

Abstract

Oral examinations have become more standardized over recent years. Traditionally a small number of raters were used for this type of examination. Past studies suggested that more raters should improve reliability. We compared the results of a multi-station structured oral examination using two different rater models, those based in a station, (station-specific raters), and those who follow a candidate throughout the entire examination, (candidate-specific raters).Two station-specific and two candidate-specific raters simultaneously evaluated internal medicine residents' performance at each station. No significant differences were found in examination scores. Reliability was higher for the candidate-specific raters. Inter-rater reliability, internal consistency and a study of station inter-correlations suggested that a halo effect may be present for candidates examined by candidate-specific raters. This study suggests that although the model of candidate-specific raters was more reliable than the model of station-specific raters for the overall examination, the presence of a halo effect may influence individual examination outcomes.