What are examples of data scatter due to measurement imprecision?
All methods, no matter how carefully performed, have some imprecision associated with them. Precision is generally characterized by the standard deviation of the method. Any one measurement is not a conclusive demonstration of anything. All that can be said is that if one took many measurements, 95% would fall within + or – 2.8 standard deviations of the mean value. More concretely, if the standard deviation were 1, and the “true” HAPS level were 15, there is a 95% chance an accurately performed measurement would fall somewhere between 12.2 and 17.8. Round robin testing of method 311 suggests it may have a standard deviation of 1 or more. How can one reduce the error from imprecision in the method? The only way to effectively reduce this error is through multiple measurements of the sample. One should never draw conclusions from a single unreplicated measurement. In addition, in light of the issue of laboratory variability, taking additional reserve test samples at the same time from t
