Michael Blann/Lifesize/Getty Images
Systematic and random errors are factors to account for during measurements performed in scientific research. They are introduced into data by the state of the measuring devices the researcher has to use in order to record data.
Systematic errors are also known as known as “bias.” This describes the nature of the error. The machine is “biased” in a specific way, meaning that measurements will always be over-reported or under-reported, but the offset in the errors should always be the same. That offset may be an empirical value, such as a measurement that is always over the actual figure by the same amount every time; or proportional, which means actual figures are either magnified or reduced by the same percent every time.
Random errors display no pattern. They are not constant. They do not occur with every measurement and the extent to which they alter the figures has no unifying characteristic. Random errors are difficult to detect and difficult to extract from core data. Where research is not dealing with straight measurement, but examines behaviour, random errors may be introduced in the data by one rogue subject who is simply in a bad mood, or not feeling well that day.
The simple difference between systematic errors and random errors is that systematic errors conform to a factor, which can be extracted from the data. Random errors have no logical pattern. In both cases, the simplest method to detect and eradicate the errors is to rerun the tests. However, in the case of systematic errors, the faulty machine would have to be corrected first; otherwise it would add the same bias to every scenario.
Systematic errors should have a simple solution. Replace the measuring that displays bias, reset it, re-calibrate it, or correct whatever environmental factor is causing the error. For example, the bias may be caused by heat affecting the samples in the study. In the case of random errors, these also may be caused by environmental factors that are irregular and impact on the measuring device. Examples of such factors could be breeze, or the sun shining through a window onto the equipment at different intensities at different times of the day. Re-calibrating a measuring device that is introducing random factors into the data will not do any good, because these errors, unlike systematic errors, are not caused by faulty settings. If the measuring device is not stable, then wobble might introduce a random error. The best way to deal with those errors is to move or replace the measuring device.
- Michael Blann/Lifesize/Getty Images