In fact, I suspect if I asked you to define precision most readers of this blog would do so by using another term that we’ve considered in this series, accuracy. That’s because laymen tend to think “precision” is a matter of being extremely accurate. It’s not. In fact, the two terms mean quite different things as we’re about to see.

Defining precision

To make the point, let’s stick with accuracy for a moment because illustrating how it differs from precision will help us to better define and understand the latter.

You may remember the above diagram from our recent blog about true value (if you didn’t read it, you can do so here). Let’s use it again. The center of the bull’s eye represents True Value. The red dots are measurements that have been taken. As you can see, accuracy shows proximity to true value. Put another way, the closer the measurement is to the bullseye, the more accurate it is said to be.

Here’s the rub: this has nothing to do with precision. Precision, instead, refers to how close measurements are to each other (with no regard to True Value at all). The point here is clear: to be accurate has nothing to do with being precise, at least in the world of measurement. When we look back the diagram, we can grasp that it’s possible for a set of measurements to be any combination of accuracy and precision; accurate but not precise, precise but not accurate, and so on.

Precision, repeatability, and reproducibility

Now that we understand what precision is, are two more important terms to bring into the conversation: repeatability and reproducibility. That’s because when the term precision is used, it needs to be accompanied by a statement about the conditions under which the measurement was taken. The most common of these statements are repeatability conditions and reproducibility conditions.

Not unlike being mistaken for accuracy, precision is also often used synonymously with repeatability. Technically, they’re not the same thing but once again, the other term can help us understand what we’re talking about when it comes to precision. So, what’s the difference between precision and repeatability?

If we make the same measurement twice, we need to ask will the two results differ from each other and if so, by how much? Generally, high repeatability means that the result is not expected to change much and when this is the case, it will be very precise. In other words, achieving repeatable conditions is a forerunner of precise measurements.

To be more specific, repeatability refers to the set of conditions under which we measure precision itself. The more consistent those conditions during the process of measurement, the higher the degree of repeatability will be and the greater the likelihood measurements taken will be precise. The more conditions change, in contrast, the less the likelihood of repeatability. Thus if, say, we take a measurement in Norway and another one in India, even using the same measurement system, environmental and other conditions will differ, and repeatability will inevitably be reduced. Repeatability seeks to eliminate every possible disturbance when measurements are being taken, minimizing possible variations that could affect results.

Inevitably, the story is a bit more complicated, so let’s look a little deeper. If you’re measuring something, you have to decide where within the measurement process you are looking for repeatability (see above). This means that the measuring process has to be structured for consistency which requires defining the process itself. The latter is generally done by setting a series of benchmarks that reduce variability and leave nothing to chance that could cause imprecise outcomes, while ensuring the deviation between measurements is as low as possible. You need to set up the right conditions if you want to achieve the most precise results.

The most common of these conditions are:

• The same instrument is used.
• The same object is used for all measurements. (We’re interested in the process, not the actual object.)
• The environmental conditions should be as equal as possible. (Temperature, light, pressure etc.) To achieve this, the normal requirement is that the measurements should be done within a limited amount of time.
• Operator influence should be minimized. This means that the same operator should perform all the measurements.
• If possible, the measured object should not be moved (removed from the instrument) or handled between measurements.

If we were to apply the above requirements to a measurement process, we would expect to obtain the same result for every measurement we make. Put another way, the process will be repeatable and the results precise. Any variation we get (and if we look closely enough there is always a variation) can be understood as describing the precision under repeatability conditions of the measurement process. We will also see the repeatability of the measurement instrument itself, which is another factor to consider.

Reproducibility

The other term we mentioned earlier was reproducibility so let’s also say a few words about reproducibility conditions. Reproducibility is related to repeatability but is used to describe a more general situation. It has less rigid requirements than repeatability as a result. Let’s look at the example used above for context.

What if we measure different objects (of the same type), in different locations with systems of different ages and handled by different operators?  In other words, we break all the rules we agreed to in the section above. Intuitively we would expect a wider variation in measurement results than in the repeatability test. Precision would suffer. If that’s what you thought would happen, you’d be right. The variation we see under these less consistent measurement conditions is what we call precision under reproducibility. Characteristics of repeatability include:

• Consistency: although less precise than repeatable measurements, reproducible measurements should still fall within an acceptable range of variation.
• Reduction of bias: reproducibility should reduce bias errors in the measurement process.
• Instruments: Reproducibility considers that different instruments may be used while still seeking consistent results.
• Transparency: Clear and detailed descriptions of measurement procedures, materials, and instruments used are central to a reproducible process.

Summary

Knowing about, understanding, and using the correct terms is always important particularly in the world of measurement where accuracy, precision, and other terms must be properly used. At Conoptica, we manufacture equipment for measuring very small tools and parts with high accuracy and repeatability, so for us this is obvious. We’re hoping that after reading our blogs, you’ll better understand why.