For independent measurements of the same quantity with known standard uncertainties, the standard uncertainty of the inverse-variance weighted mean is the reciprocal square root of the sum of the weights. Measurements with smaller uncertainties receive greater weights.
Weighted Standard Error
For independent measurements of the same quantity with known standard uncertainties, the standard uncertainty of the inverse-variance weighted mean is the reciprocal square root of the sum of the weights. Measurements with smaller uncertainties receive greater weights.
\(u_{\mathrm{w}}( \bar{x}_{\mathrm{w}} ) = 1/\sqrt{\sum_{i=1}^{n} w_{i}}, \text{ where } w_{i} = 1/u_{i}^{2}\)
What this formula means
Scientific principle
Principle
A weighted mean combines measurements according to their reliability. When measurement uncertainties are independent and known, inverse-variance weighting gives greater influence to more precise measurements and minimizes the variance of a linear unbiased weighted estimate.
Derivation
1. Let xᵢ be independent measurements of the same quantity, each with standard uncertainty uᵢ.
2. Assign inverse-variance weights: wᵢ = 1/uᵢ².
3. Calculate the weighted mean: x̄_w = [Σwᵢxᵢ]/[Σwᵢ].
4. For independent measurements, the variance of this weighted mean is 1/[Σwᵢ].
5. Taking the square root gives u_w(x̄_w) = 1/√(Σwᵢ).
6. Substituting wᵢ = 1/uᵢ² gives the inverse-variance weighted standard uncertainty.
Final Result
u_w(x̄_w) = 1/√(Σᵢ₌₁ⁿ 1/uᵢ²)
Weighted Mean Relation
x̄_w = [Σᵢ₌₁ⁿ wᵢxᵢ]/[Σᵢ₌₁ⁿ wᵢ]
Weight Relation
wᵢ = 1/uᵢ²
Variance Relation
u_w²(x̄_w) = 1/Σᵢ₌₁ⁿ wᵢ
Variables and their meanings
| Symbol | Meaning | Unit |
|---|---|---|
x̄_w | Inverse-variance weighted mean | Same unit as individual measurements |
u_w(x̄_w) | Standard uncertainty of the weighted mean | Same unit as individual measurements |
xᵢ | ith measurement | Unit of the measured quantity |
uᵢ | Standard uncertainty of the ith measurement | Same unit as xᵢ |
wᵢ | Inverse-variance weight of the ith measurement | Inverse square of the measurement unit |
n | Number of independent measurements | Measurements |
SI units and conversions
Weighted Mean
Same unit as the measured quantity
Weighted Standard Uncertainty
Same unit as the measured quantity
Inverse Variance Weight
Inverse square of the measured unit
Weight Sum
Inverse square of the measured unit
Conditions of use
All measurements estimate the same underlying quantity.
Measurement uncertainties are standard uncertainties.
The measurements are statistically independent.
The individual standard uncertainties are known or reliably estimated.
Inverse-variance weighting is appropriate for the measurement model.
Limitations and assumptions
The formula is not appropriate without modification when measurements are correlated.
It assumes that the reported uncertainties adequately represent measurement variability.
A systematic error shared by all measurements is not removed by averaging.
The weighted mean can be misleading when measurements estimate different underlying quantities.
If uncertainties are estimated from the same small dataset rather than independently known, additional statistical considerations may be necessary.
Step-by-step worked examples
| Problem | Solution | Answer |
|---|---|---|
Two independent measurements of the same length have standard uncertainties of 1.0 cm and 2.0 cm. Calculate the standard uncertainty of their inverse-variance weighted mean. |
| The weighted standard uncertainty is approximately 0.894 cm. |
Three independent measurements have standard uncertainties of 0.2 s, 0.3 s, and 0.4 s. Calculate the standard uncertainty of their inverse-variance weighted mean. |
| The weighted standard uncertainty is approximately 0.154 s. |
Real-world applications
Combining repeated measurements with unequal precision
Scientific meta-analysis
Calibration data processing
Experimental measurement averaging
Astronomical and physical parameter estimation
Formula Meaning
For independent measurements of the same quantity with known standard uncertainties, the standard uncertainty of the inverse-variance weighted mean is the reciprocal square root of the sum of the weights. Measurements with smaller uncertainties receive greater weights.
Sources and further reading
No source has been supplied for this item yet. Treat the claim as needing editorial verification.