Abstract
Knowledge of the extent of the symptoms of a plant disease, generally referred to as severity, is a key to both fundamental and applied aspects of plant pathology. Most commonly, severity is obtained visually and the accuracy of each estimate (closeness to the actual value) by individual raters is paramount as it directly affects inferences and decisions based on average data. Thus, the more accurate the visual estimate of severity made by a rater for each specimen, the lower the magnitude of errors in the analysis, and thus the lower the risk of incorrect decisions. Errors in estimates of severity on diseased specimens are becoming better understood and ways to minimize them have been developed during the last 25 years. The development of various tools to aid raters prior to, or during the assessments provides a basis for greater accuracy. Severity is often based on the continuous percentage scale to represent the proportion of area diseased. However, some tools, such as those based on category scales have been shown to harm accuracy of some estimates under specific conditions, potentially leading to Type II errors. On the other hand, ensuring raters are familiar with symptom identity, assessment training and the use of standard area diagrams (a series of pictorial representations of examples of actual severity), can lead to improved accuracy of each (or at least most) estimate(s) by raters, particularly those who are innately less accurate. In this review, the need for accuracy of visual estimates of disease severity on plant specimens is emphasized in relation to the sample mean and variance, and the history of accuracy in disease estimation is described, the advantages of tools to aid accuracy are presented and future research needs are outlined.