Statistical measurement concept · BEGINNER LESSON
Measurement error
Measurement error is the difference between a measured value and the true value it is meant to represent. It can come from random chance or from systematic bias, and it can affect both individual readings and study results.
What it is
Measurement error means a value obtained by a test, survey, device, or sample does not exactly match the true value of the thing being measured. In diet studies, the true target may be usual intake, meaning long-term average intake rather than what someone ate on one day. The same idea applies to other data such as height, weight, blood pressure, exposure levels, or biomarker concentrations. If the error is ignored, conclusions can be misleading.
[2] epi.grants.cancer.gov › dietary-assessment-primer › conceptsKey Concepts about Measurement Error | EGRP/DCCPS/NCI/NIH [1] Challenges and pitfalls in analyzing, reporting, and interpreting ...
How it happens
Two broad sources are described. Random error comes from chance, such as a sample that differs from the target population or an instrument that gives slightly inconsistent readings. Systematic error is a persistent bias caused by study design, data collection, or analysis. In exposure studies, this can happen when reported exposure differs from the true exposure because of site differences, non-normal distributions, non-detectable values, or biomarker values that do not fully represent outside exposure.
[1] Challenges and pitfalls in analyzing, reporting, and interpreting ... [2] epi.grants.cancer.gov › dietary-assessment-primer › conceptsKey Concepts about Measurement Error | EGRP/DCCPS/NCI/NIH
What it does to results
With only within-person random error, estimated distributions can become too wide, so the fraction above or below a cutoff may be wrong. For a single exposure measured with this kind of random error, the estimated association with an outcome is typically attenuated, meaning pulled toward no association, and statistical power is reduced. With systematic error, the mean and shape of the estimated distribution can shift, and the association can be biased in either direction. In some settings, differential error can create spurious results.
[3] Effects of Measurement Error | EGRP/DCCPS/NCI/NIH [2] epi.grants.cancer.gov › dietary-assessment-primer › conceptsKey Concepts about Measurement Error | EGRP/DCCPS/NCI/NIH
Common forms
Classical measurement error assumes the true value varies around the observed value with noise in the measurement itself. Berkson measurement error is the opposite: the observed value is fixed, while true values vary around it, such as when people are assigned a group-level exposure. Exposure misclassification means a categorical exposure is placed into the wrong category. Researchers also distinguish differential error, where error differs by group, from non-differential error, where it does not.
[1] Challenges and pitfalls in analyzing, reporting, and interpreting ... [2] epi.grants.cancer.gov › dietary-assessment-primer › conceptsKey Concepts about Measurement Error | EGRP/DCCPS/NCI/NIH
Limitations and handling
Measurement error is common, so studies often need repeat observations, better reference instruments, or quantitative bias assessment. For exposure studies, reporting the measured range, central tendency, and variability helps readers judge error. Comparing a single sample with pooled or repeated samples can show how variable the measure is over time. When data are limited, statistical models or averaging repeat measures may reduce random error, but they cannot remove all bias.
[1] Challenges and pitfalls in analyzing, reporting, and interpreting ... [3] Effects of Measurement Error | EGRP/DCCPS/NCI/NIH
SEE THE IDEA
How measurement error affects a study
The real amount or status being studied.
What the survey, device, or sample reports.
Random noise or systematic bias.
Mean, distribution, or association estimated from the data.
True value is not directly observed; the measured value departs from it, and the type of departure changes how study results are distorted.
A study starts with a true value, observes a measured value, and then error can widen, shift, or bias the final estimate.
TRY IT YOURSELF
Test your understanding
Choose an answer, then check the explanation.
PUT IT TOGETHER
Match each concept to its meaning
Drag a label onto its description, or select a label and then select a description. This also works with a keyboard.
4 pairs to match
Sources and further reading
This general lesson was generated from public source material. It explains a concept; it does not interpret a private paper or provide medical advice.