{"id":518,"date":"2015-01-18T17:00:05","date_gmt":"2015-01-18T17:00:05","guid":{"rendered":"http:\/\/www.agron.iastate.edu\/glsi\/?p=518"},"modified":"2025-09-11T11:56:44","modified_gmt":"2025-09-11T16:56:44","slug":"accuracy-vs-precision","status":"publish","type":"post","link":"https:\/\/www.agron.iastate.edu\/glsi\/blog\/accuracy-vs-precision\/","title":{"rendered":"Accuracy vs Precision"},"content":{"rendered":"

Scientists often measure and predict things. Therefore, we need ways to describe how much we know, how close a number is to reality, and how likely we are to get the same number again. The terms accuracy and precision are generally used to describe these things, but there can be some ambiguity. This post explains the difference between the two and explores the different aspects of precision\u2019s multiple meanings.<\/p>\n

Accuracy refers to the correctness of a measurement or prediction. The results can vary a lot, but what matters is the difference between the measurements or predictions to what is considered to be the real or accepted value. Precision is often contrasted with accuracy by emphasizing the repeatability meaning of the word. This is most applicable to measurements, but can be applied to modelling too (e.g., stochastic models). When measuring something, we want some confidence that if we were to measure the same thing again, we would get a similar answer. In this situation, precision describes the degree of this similarity.<\/p>\n

Accuracy and precision are both highly desirable characteristics for our measurements and predictions, but they are usually independent. For example, most models should produce the same results when based on the same algorithms and inputs, but this has nothing to do with the accuracy of the model\u2019s predictions. Measurements can also be precise without being accurate. When this situation occurs, the difference between the measurements and the true value is called the bias (i.e., bias of an estimator). If we can quantify the bias, then it is possible to adjust for the bias to improve the accuracy.<\/p>\n

Another meaning for precision creates some ambiguity when using the term. This second meaning of precision describes how much we know about a measurement. For example, rules of significant digits<\/a> help us to use numbers that don\u2019t express more than what we really know about a measurement. This definition of precision is mostly used to describe the exactness of measurements, but is sometimes used to describe the level of detail in spatial applications. For example, the popular term \u2018precision agriculture<\/a>\u2019 is trying to emphasize the extra level of detail that the work can be done.<\/p>\n

Confusion between precision and accuracy is fueled by the common mixing of the terms. Merriam-Webster<\/a> defines precision as \u201cthe accuracy (as in binary or decimal places) with which a number can be represented.\u201d Although it\u2019s really a misuse of the term \u2018accuracy\u2019 to define a meaning of precision, they are trying to get at the concept of exactness. However, ambiguity exists in the scientific community too. The ISO<\/a> has advocated defining accuracy to encompass both trueness and precision.<\/p>\n

In order to avoid ambiguity, using alternative terms for precision may be a wise choice. When describing the ability to get similar results from multiple measurements, using the term \u2018repeatability\u2019 would be clearer in meaning. Precision, in terms of describing the quantitative or spatial detail of something, isn\u2019t as easily replaced by a term such as \u2018exactness\u2019 because even that can conjure ideas of accuracy. As semantics evolve, maybe the meaning of precision will evolve to something more limited. In the meantime, when possible, we can use \u2018detail\u2019 and \u2018resolution\u2019 for spatial applications and \u2018significant digits\u2019 for quantitative applications.<\/p>\n

For more, check out:<\/p>\n