Main

Chadwick S, Holsgrove G. Fac Dent J 2011; 2: 184–190

The reliability of an examination is demonstrated by the Cronbach's alpha coefficient. If the value attains the recently raised benchmark value of 0.90, then 90% of what is measured is true difference between the candidates, with only 10% comprising errors associated with the assessment process. However, Cronbach's alpha cannot be used when there are less than one hundred candidates. In addition, if the examination assesses high performers such as a postgraduates, the range of marks (subject variance) is narrowed. For small cohorts therefore, apart from reliability and validity, other factors are recruited that, taken in the round, are referred to as utility. Such include the impact of the training experience, benefits accruing by those being treated by learners who have met standard, and opportunity costs.