By Jayson Chaw. Tue Jul 28.
No measurement is perfect. Understanding the difference between random and systematic error explains why, and what to do about each.
Every measurement in science carries some degree of uncertainty. This is not a sign of sloppy work; it is an unavoidable feature of measuring anything in the real world. What separates careful scientific practice from careless practice is understanding where that uncertainty comes from and whether it can be reduced. Errors are generally sorted into two categories: random and systematic, and they behave in very different ways.
Random error
Random error causes measurements to scatter unpredictably above and below the true value. It comes from small, uncontrollable fluctuations, such as slight variation in how quickly a stopwatch is started, tiny inconsistencies in reading a scale, or minor changes in the surrounding environment during an experiment. Because random error is unpredictable in direction, it can be reduced by repeating a measurement several times and calculating a mean. Some readings will be slightly high, some slightly low, and averaging tends to cancel much of this variation out.
Systematic error
Systematic error, by contrast, shifts every measurement consistently in the same direction. It often comes from a fault in equipment or method, such as a balance that has not been zeroed properly, a ruler with a worn end, or a thermometer that reads slightly high across its whole range. Crucially, repeating the measurement and averaging does not fix systematic error, because every repeat is affected by the same underlying fault in the same way.
Accuracy versus precision
These two errors map closely onto the difference between accuracy and precision. Precision describes how close repeated measurements are to each other, and is mainly affected by random error. Accuracy describes how close a measurement is to the true value, and is affected by systematic error. It is entirely possible for a set of results to be highly precise (tightly clustered together) while still being inaccurate, if a systematic error has shifted every reading away from the true value by the same amount.
Spotting each type in practice
Results that vary randomly around a central value, with no consistent pattern, usually point to random error
Results that are consistently too high or too low compared with a known or expected value usually point to systematic error
A wide spread of repeated readings suggests significant random error
A tight cluster of readings that still disagrees with a trusted reference value suggests systematic error
Reducing each type of error
Random er…