Accuracy vs. Precision

calibration is a field all on its own.

accuracy, precision and uncertainty all have to be accounted for in calibration, all have specific meanings and mathmatical statistics to calculate them.

I would suggest you look up the precise meaning of each of the words and how they are calculated, from that you will get a good idea of the maths involved.

John.

I am not really concerned with definitions as much "procedures which "include specific" "directions and limits for accuracy and precision.

Well that's going to vary with the measuring method and type of measurement isn't it...

John.

A great answer. It is critical to determine what you need to know before you can determine the equipment needed. The manufacturers will tell you how to calibrate the equipment to optimize the accuracy and precision for their equipment.

The limits for measurement equipment (instruments) are a function of allowable tolerances for a specific process or item. Thus, if you have to measure a current, voltage, distance or anythig else for which a tolerance range has been stablished , the instrument uncertainty should be at least 1/3 of such range and the sensitivity of the instrument (the lesser change in parameter that can be detected) should be 1/10 of tolerance range,if possible.

Generally accepted rule for determination of uncertainties is the ISO "GUM" (ISO Guide for the expression of uncertainty in measurements) Most organizations involved in calibration has adopted this document as a guide. You can download the NIST guide from the website indicated or the EAL (European co-operation for Accreditation of Laboratories) document EAL-R2 from www.euromet.org/docs/calguides. In this last website you can find several guides for calibrating specific types of instruments, with the complete process, variables to take into account, examples, etc (For example EAL -10-08 for calibration of thermocouples)

Google METROLOGY. (not metereology...)

Wangito.

Yeah, we wouldn't want to talk about accuracy OR precision in meterology. Unless you consider being constantly wrong "precise."

To answer the original question without the equations, accuracy is how close you can get to the bulls eye of a target. Precision is what your score is on average. Uncertainty is the ability to predict where your next shot will land based on the statistics derived from all the previous attempts, assuming you haven't made any major changes.

"........ accuracy is how close you can get to the bulls eye of a target. Precision is what your score is on average."

I disagree, partly:

I think precision is how close your shots are together..

A precise rifle/shooter will have a tight group,

even if the group is not close to the bulls eye...

One of the first things you will have to have is a controlled atmosphere. We have a clean room that is kept between 68 and 73 degrees Fahrenheit and also has humidity control to keep the humidity between say 10 and 40 percent. I believe this is required by all standards or at least it should be. Then we have a 5 inch thick binder that contains a procedure for each type or category of measuring instrument. That procedure then defines what type of equipment can be used to calibrate a gage, how accurate the calibrating equipment can be(usually capable of measuring 10 percent or less of the smallest division on your instrument to be calibrated) and a tolerance that the gage being calibrated has to meet.

You need to assess the field you are working in and the customers you will have to find out the exact code you have to follow.

Hi,

Many labs find equipment that has the accuracy they need, then simply make sure the unit meets specs by having it calibrated regularly. Our lab sometimes uses things to a tighter tolerance. We have a documented procedure with an uncertainty analysis. The procedure must list the tolerance needed. Moving range charts or spreadsheets can be used to track the standard's compliance. For a unit that we use specs on: If the uncertainty analysis lists 90 day specs, then it needs to be calibrated every 90 days, but if it lists 1 year specs, then cal can be done once a year. Recall intervals can be changed under certain conditions specified by a standard such as ISO/IEC 17025. Hope this helps.

Regards,

S

Thanks S,

That is more along the lines, I think, of what I am looking for. Our company is going through a new focus on Quality Standards. I was wanting to know how other companies addressed the issue I presented above, since I was presented with this question by Corp. I was relating to ISO/IEC 17025.