Measurement error
Also known as: Observational error
The gap between what you measure and the true value you meant to capture.
What it means
Measurement error is the difference between an observed score and the true value of the quantity being measured, and it comes in two flavors that behave very differently. Random error is unsystematic noise that varies unpredictably across observations; it cancels out on average but inflates variance and, in a predictor, biases estimated relationships toward zero through a phenomenon called attenuation. Systematic error, or bias, pushes measurements consistently in one direction — a miscalibrated scale, a leading question, a socially desirable answer — and does not wash out with larger samples, so it can distort conclusions in any direction. Reliability theory formalizes this by partitioning observed variance into true-score and error components, which is why reliability sets a ceiling on how strong an observed correlation can be. It matters because no amount of clever analysis can fully recover a signal that the measurement instrument corrupted at the source, making sound measurement the foundation of credible behavioral research.
Examples
If a survey of well-being is noisy, its correlation with income will look weaker than the true relationship — a textbook case of attenuation by random measurement error.
A bathroom scale that reads three pounds heavy will never reveal your true weight, however often you step on it — systematic error does not average away with more readings.
Two interviewers score the same candidate differently on gut feel. Averaging many interviewers cancels the noise, but a single interview is one draw from a very wide distribution.
Where this comes up
- Response Bias in Surveys: A Behavioral Science PerspectiveSurvey responses aren't always accurate. Learn about the most common types of response bias, why they occur, and practi…