Introduction
Accuracy and precision seem to be the same at first glance; however, that is not the case. The former is the degree to which a given value or measurement is close to the true value. Precision is the degree to which the values are close to each other rather than the true value. Confusing, right? One will have better luck deciphering Yoda’s cryptic words.
So, what is accuracy? To put it in a nutshell, take a look at the target practice example. The shot is accurate if the shooter hits the bull’s eye. His aim is precise if he continues to hit the bull’s eye consecutively. Now, what does this mean? Say a dart player hits the same spot (not the bull’s eye) again and again when he throws the darts. That means his shots are highly precise, but does that mean they are accurate? No, his shots lack accuracy even if the darts are clustered around the bull’s eye.
On the other hand, his shots are highly accurate and precise if he hits the dartboard dead center each time he throws the darts. Therefore, precision can occur without accuracy. Some argue that accuracy can occur without precision, while others feel that is like looking at only half an equation. A person getting an accurate result once and consistently inaccurate result every other time they experiment cannot rely on the findings. The experiment itself may be flawed. It is clear that accuracy and precision are different whichever argument one supports.
Accuracy Vs. Precision
Accuracy refers to the extent to which the actual value (the value the measurer gets) is close to the absolute value (the value the measurer should get). Precision refers to minimal or negligible variation between the values one obtains when repeatedly measuring the same factor (high variation signals low precision).
Difference Between Accuracy And Precision In Tabular Form
| Parameters of Comparison | Accuracy | Precision |
| Meaning | Accuracy is the extent of conformity. | Precision is the extent of reproducibility. |
| Dependency | High accuracy requires high precision. | High precision does not necessarily mean high accuracy. |
| Repeatability | Repeatability has nothing to do with accuracy. A person may get lucky and shoot accurately once and not even hit the target board for the rest of their practice. Their failure to hit the target again does not negate the accuracy of the first shot. | Precision requires a process to yield the same results consistently every time it is repeated. In short, precision includes repeatability and reproducibility. |
| In Psychophysics | Accuracy is equivalent to validity (the extent to which a concept or conclusion corresponds to the real world). | Precision denotes reliability. A process that yields consistent results under unchanging conditions is highly precise. |
| Measurement Factors | A single factor is used to measure accuracy (does the value correspond to the true value?). | Multiple factors are used to measure precision (reliability, reproducibility, level of variation among the given set of measurements, and so on.). |
| Represents | The level of agreement between the measurement and absolute value. | The level of nearness between the individual values and is not concerned with the absolute value. |
What Is Accuracy?
Accuracy refers to the correctness and closeness of a measurement to the absolute value/standard value. It is common to determine the average of multiple values when determining the accuracy of a set of observations. Predictions, too, can be classified as accurate or inaccurate. For example, if a person predicts that the fish they caught weighs 6 pounds and it weighs the same (or close enough to it), the prediction is highly accurate. However, the estimate is inaccurate if the fish weighs only 3 pounds. In military terms, high accuracy indicates true aim. That is, the bullet is embedded in the desired spot.
Accuracy is described using various terms in different fields. Nevertheless, the general meaning remains the same. To most people, accuracy merely means devoid of error; this holds true in most cases except in statistics. After all, there are bound to be slight variations in the measured values. In statistics, accuracy is concerned with the difference between the measured and absolute values. The measurements are accurate as long as the error is minimal and falls within the limits specified.
Percent error aids in determining how accurate the measurement are. To calculate percent error, first, the difference between the accepted value and the measured value is divided by the accepted value. Next, the resulting value is multiplied by a hundred. Lower percent errors help convey how dependable the experiment’s findings are. Higher percent errors indicate invalid findings or conclusions.
Facts are accurate or inaccurate as well. Facts are accurate if they conform to the accepted standard. Objects fall because of gravity is an accurate fact. However, in some instances, the answers are not always so clear-cut. Therefore, facts are accurate only if they can be proven. Facts that cannot be proven (does not matter if they feel or sound right) and those that are wrong (even if accepted by most people) are classified as inaccurate. Therefore, accuracy, in this sense, refers to real-world truths.
Types Of Accuracy
The following are the three types of statistical accuracy:
Point Accuracy
Point accuracy differs from the common definition of accuracy. It refers to the accuracy of an instrument at a specific point on the scale.
Accuracy As A Percentage Of Scale Range
Percentage accuracy or accuracy as a percentage of scale range is obtained by multiplying the accuracy percentage by the pressure reading. That is, if the accuracy percentage is 0.5 percent, an error of ± 0.5 may be overlooked as it is negligible. However, an error that falls outside that limit is considered a high-value error.
Accuracy As A Percentage Of True Value
Accuracy is determined by comparing the obtained value with the true value. A ± 0.5 error (difference from the absolute value) is a negligible error. Moreover, this method is used to determine an instrument’s accuracy.
Errors In Accuracy
Calibrating an instrument helps in eliminating inaccuracy. However, in most cases, the cause of inaccuracy is much more complex. Experimenter drift, leading questions of interviewers, electronic instrument drift, and so on are some of the other causes of inaccuracy. Experimenter drift occurs when observers are too exhausted to care about collecting data using standardized procedures. A drift causes electronic instruments’ readings to be consistently higher or lower. Dealing with systematic errors is much more complex than dealing with random errors.
What Is Precision?
Precision is the level of variation between the measurements of the same object/factor. A repeatable and reproducible phenomenon is said to be precise. For example, the person has a precise aim, if a person practices knife throwing and hits the same spot over and over. The spot need not be the center of the target; it merely needs to be a spot within the target area. The precision level is high in this example. However, if the person hits different spots when throwing the knives, precision is low. High precision indicates a low level of variation and vice versa.
Precision is determined by calculating the average deviation. One needs to determine the absolute deviations to find out the average deviation. The former is the difference between the average of a set of values and each value of that set. The latter is the sum of absolute deviations divided by the number of values in the set. In statistics, the larger the sample size, the greater will be the precision as opposed to accuracy (an increase in sample size leads to a decrease in accuracy).
It is desirable that an instrument is accurate and precise. However, using a precise instrument is better than using a relatively accurate instrument when forced to choose. The choice makes sense because people will know by how much a precise instrument’s measurement will be off from the true value. For example, a clock that ticks the seconds precisely even if it shows the wrong time is helpful as the observer will know exactly by how many minutes the clock is off. On the other hand, one can never determine how much an accurate instrument’s measurements or readings are off from the true value. One only knows that the readings are close to the accepted value.
Types Of Precision
Precision may be high or low and can be further classified depending on the level of accuracy with which it occurs. The following are the types of precision:
High Precision With High Accuracy
Highly precise and accurate shots indicate that the shooter hit the bull’s eye every time he shot. Remember Kyle from the film American Sniper? He takes the headshot every time and hits the mark without fail. Therefore, hitting the bull’s eye in a target board repeatedly requires high accuracy and precision.
High Precision With Low Accuracy
A beginner or an amateur would most probably end up shooting at the target board’s outer area when practicing. However, if the shooter manages to shoot at the same spot consistently, his aim is precise, though not accurate.
Low Precision With High Accuracy
Instead of hitting the target board dead center, if the shooter manages to shoot close enough at different spots (slightly to the left, right, up, or down), it indicates low precision but high accuracy.
Low Precision With Low Accuracy
If people shoot like Susan Calvin from I Robot (she shot a gun with her eyes closed), they will most probably hit different spots in the target board (if they hit it at all). Since the bullets miss the bull’s eye and land sporadically on the target area, they clearly indicate low precision and accuracy.
Errors In Precision
Precision is concerned with random errors in the field of statistics. Random errors arise due to unpredictable factors during an experiment and are ever-present. However, they can be minimized by increasing the sample size. Slight variations occur each time the weight or height of the same person is measured due to their body posture. These errors do not significantly affect the experimenter’s ability to draw conclusions when dealt with efficiently.
Main Differences Between Accuracy And Precision (In Points)
- Accuracy deals with systematic errors and measures statistical bias (the amount of inaccuracy), whereas precision deals with random errors and measures statistical variability (the extent of imprecision).
- Accuracy indicates correctness, while precision indicates exactness.
- Precision does not depend on accuracy. Accuracy has two definitions; according to the well-known definition, accuracy is independent of precision. However, according to IFO, high accuracy indicates high trueness and preciseness.
- In a confusion matrix (a specific table layout that aids in visualizing an algorithm’s performance and classifies document retrieval results), precision is defined as true positives divided by the sum of true and false positives. Accuracy is a more uncommon metric used in this matrix; it is the sum of true positives and negatives divided by the total number of documents.
- Cognitive accuracy acknowledges the unintended results of a cognitive process, whereas cognitive precision only considers the intended or desired results.
- In statistics, accuracy is referred to as bias, while precision is known as variability.
- Measurement of accuracy involves determining the level of closeness to true or absolute value. Measurement of preciseness involves determining the similarity of results when measurements are taken under constant conditions.
- Accuracy requires only one measurement, whereas precision requires several measurements to be taken.
Conclusion
Accuracy refers to the degree of correctness, and precision refers to the degree of exactness. Being accurate because one got lucky or merely increasing precision without being even remotely accurate is not a desirable outcome. That is why ISO’s definition makes much more sense.
In short, high accuracy along with high precision is the most desirable option. What is the point of one without the other? Perhaps, that is why most people are confused as to their meaning and use the terms interchangeably. Striving to be precise first and then focusing on being accurate proves to be advantageous when it comes to developing or improving one’s skill. Anyway, in the end, it is better to be like Hannibal with a scalpel – precise and accurate.
References
- https://en.wikipedia.org/wiki/Accuracy_and_precision
- https://statisticsbyjim.com/basics/percent-error/
- https://www.indeed.com/career-advice/career-development/how-to-measure-accuracy
- https://www.scribbr.com/methodology/random-vs-systematic-error/
- https://www.thoughtco.com/random-vs-systematic-error-4175358#
- https://www.thoughtco.com/difference-between-accuracy-and-precision-609328
- https://circuitglobe.com/accuracy-and-precision.html
- https://en.wikipedia.org/wiki/Wikipedia:Accuracy
- https://www.shiksha.com/online-courses/articles/difference-between-accuracy-and-precision/
- https://www.forecast.app/blog/difference-between-accuracy-precision
