Definition
Measurement error is the difference between a recorded value and the true quantity it was intended to capture. Every empirical measurement carries some error, whether from instrument imprecision, observer variation, environmental fluctuation, or ambiguity in the underlying definition. Statisticians distinguish two broad kinds: random error, which fluctuates symmetrically around the true value and tends to average out over many observations, and systematic error (bias), which pushes every measurement in the same direction and does not vanish with more data.
The honest acknowledgement and characterisation of measurement error is what separates a credible empirical study from a misleading one.
Why it matters
How it works
Random error arises from many small, independent disturbances — slight differences in timing, instrument readings, observer judgement, or environmental conditions. Because the disturbances are roughly symmetric around the true value, the average of many measurements converges to the true value, and the spread of measurements gives a usable estimate of precision. The standard error of a sample mean shrinks with the square root of the sample size, which is why larger studies produce more precise estimates.
Systematic error is more dangerous because it does not behave so gracefully. A miscalibrated scale that reads two grams light produces a biased estimate no matter how many measurements you take. Selection bias works the same way: if a survey systematically misses certain populations, no amount of additional sampling repairs the gap. The remedies for systematic error are upstream — calibrate instruments, pilot-test questionnaires, sample with care, and define the underlying quantity clearly before measuring. Downstream the analyst is left with sensitivity analyses and honest disclosure of the limits of what the data can support.