Intraobserver agreement (or variance) is often used to evaluate analytical methodologies when some aspect of the analysis is subjective or interpretive in nature.
Intraobserver agreement is a measurement expressed in terms of a percentage; the total number of single observers whose analysis of a material was the same on two different occasions is divided by the number of participants in the test.
The participant (observer) must be given a sufficient number of unmarked samples in each test period so it cannot be easily ascertained that one of the samples is a duplicate.
Intraobserver agreement is primarily used to evaluate the reliability of a medical diagnostic test. If a large number of doctors participating in the study provide different diagnoses from the same test results viewed on two separate occasions, it is evidence the test is not reliable as a sole diagnostic indicator.