Measurement Confidence Interval

The measured value for a meter or inventory provides an expected value. In engineering terms, this value is not necessarily the true value. The true value can actually lie anywhere within a confidence interval. For example, if a measured value is 100 and the tolerance on that measurement is set to 2%, then 2% is equivalent to two standard deviations. This is a confidence interval of 95%. So, for a normal distribution, the true value lies anywhere in the range 98 - 102, 95% of the time, but 5% of the values fall outside this range, in the tail regions of the normal bell curve, as shown in the following figure.