Negative percent agreement (NPA) is an important metric used in data analysis, particularly in the field of medical research. It is a measure of the level of agreement between two evaluators or observers in identifying the absence of a condition in a sample. In other words, it measures the percentage of cases where both evaluators agree that a certain condition is not present.
NPA is calculated by dividing the number of cases where both evaluators agreed on the absence of the condition by the total number of cases where the condition was absent and both evaluators made a judgement. The result is then multiplied by 100 to give a percentage.
For example, let’s say two radiologists were evaluating an X-ray image for the presence of lung cancer. In 100 cases where there was no evidence of lung cancer, both radiologists agreed that there was no cancer in 90 cases. In the remaining 10 cases, they disagreed, with one radiologist identifying cancer and the other not. In this case, the NPA would be 90%, indicating a high level of agreement between the two evaluators in identifying the absence of lung cancer.
NPA is an important metric as it complements other measures of agreement, such as positive percent agreement (PPA) and kappa statistics. PPA measures the level of agreement in identifying the presence of a condition, while kappa statistics take into account the degree of agreement beyond chance. NPA, on the other hand, measures agreement in identifying the absence of a condition, which is equally important in medical research and diagnosis.
In conclusion, negative percent agreement is a key metric in data analysis, particularly in medical research, as it measures agreement between evaluators in identifying the absence of a condition. It complements other measures of agreement, such as PPA and kappa statistics, and provides a comprehensive picture of the accuracy and reliability of data analysis in identifying both the presence and absence of a condition.