Sampling Variance: Formula, Example & Guide

Sampling variance is the variance of a point estimator across hypothetical repeated samples drawn from the same population under an identical study design. While sample variance measures the dispersion of individual data values within a single observed dataset, sampling variance measures the precision and stability of an estimator—such as a sample mean or proportion. The square root of the sampling variance is the standard error, which directly determines confidence interval widths, hypothesis test boundaries, and statistical power in empirical research.

Definition in One Paragraph

Variance measures the average squared distance of values from their arithmetic mean. In quantitative research, statistical theory distinguishes between sample variance (s²), which quantifies how individual observations disperse around the sample mean within a single dataset, and sampling variance (Var(θ̂)), which quantifies the dispersion of an estimator across repeated random samples of size n drawn from the same population. Sample variance reflects unit-level heterogeneity, whereas sampling variance measures estimation precision; as sample size increases, sampling variance contracts toward zero even when sample variance remains constant.

What Is Meant by Sampling Variation?

Sampling variation is the natural fluctuation of sample statistics across different random samples drawn from the same population. Because a sample includes only a subset of the target population, two independent probability samples will yield slightly different means, proportions, and test statistics. Quantifying sampling variation allows researchers to determine whether observed differences reflect true effects or random chance in statistical tests.

What Are Two Types of Variance?

In research methodology, two types of variance refer to two key pairings:

Sample Variance vs. Population Variance: Population variance (σ²) measures spread across all units in a population; sample variance (s²) estimates that spread from observed sample data using n – 1 degrees of freedom.
Data-Level Variance vs. Sampling Variance: Data variance (s² or σ²) measures the spread of raw observations, whereas sampling variance (Var(θ̂)) measures estimator variability across repeated sampling.

What Is the Symbol Used to Represent Variance?

Standard statistical notation differentiates variances:

σ² (Sigma squared): True population variance.
s²: Unbiased sample variance calculated from observed data.
Var(θ̂) or σ²_θ̂: Theoretical sampling variance of an estimator θ̂.
Var(x̄) or s²_x̄: Sampling variance of the sample mean.
SE: Standard error, the square root of sampling variance: SE = √(Var(θ̂)).

Formula and Mathematical Foundations

The mathematical formulation of sampling variance links data-level dispersion directly to estimator stability.

1. Sample Variance Formula (Bessel’s Correction)

For n independent observations, sample variance uses Bessel’s correction (n – 1) to ensure an unbiased estimator of population variance, so that E(s²) = σ²:

s² = [1 / (n – 1)] × Σ_i=1ⁿ (x_i – x̄)²

Where x_i represents individual observations, x̄ = (1 / n)Σx_i is the sample mean, and n – 1 represents degrees of freedom compensating for measuring deviations from x̄ rather than true μ.

2. Sampling Variance of the Sample Mean

As documented in the NIST/SEMATECH e-Handbook of Statistical Methods (Section 3.3.3.3), the sampling variance of the sample mean from a simple random sample is inversely proportional to sample size n:

Var(x̄) = σ² / n ≈ s² / n

The square root of this value yields the standard error of the mean: SE(x̄) = s / √n.

3. Sampling Variance of the Sample Variance

Under a normal distribution with population variance σ², the theoretical sampling variance of s² across repeated samples is:

Var(s²) = 2σ⁴ / (n – 1) ≈ 2(s²)² / (n – 1)

This demonstrates that variance estimates themselves carry substantial sampling variability in small samples.

4. Complex Survey Sampling Adjustments

Statistics Canada (Catalogue no. 12-587-X) specifies two adjustments for survey designs:

Finite Population Correction (FPC): When sampling without replacement from population N where n / N > 0.05: Var(x̄)_FPC = (s² / n) × [(N – n) / (N – 1)].
Design Effect (Deff): In clustered designs, intra-cluster correlation inflates variance: Var_complex(θ̂) = Var_SRS(θ̂) × Deff. Omitting Deff creates spuriously narrow confidence intervals and substantial sampling errors.

Worked Example with Real Numbers

Consider a laboratory assay calibration where a researcher records turnaround processing times (in minutes) across n = 5 standardized replicates: 12, 15, 18, 20, 25.

How Do You Do Sample Variance?

Calculate sample mean (x̄): x̄ = (12 + 15 + 18 + 20 + 25) / 5 = 90 / 5 = 18.00
Compute deviations (x_i – x̄): (12 – 18) = -6; (15 – 18) = -3; (18 – 18) = 0; (20 – 18) = +2; (25 – 18) = +7
Square the deviations: (-6)² = 36.00; (-3)² = 9.00; (0)² = 0.00; (2)² = 4.00; (7)² = 49.00
Sum squared deviations (SS): SS = 36 + 9 + 0 + 4 + 49 = 98.00
Divide by degrees of freedom (n – 1 = 4) for s²: s² = 98.00 / 4 = 24.50 min² (s = √24.50 ≈ 4.95 min)
Compute sampling variance of the mean: Var(x̄) = s² / n = 24.50 / 5 = 4.90 min²
Compute standard error of the mean: SE(x̄) = √4.90 ≈ 2.21 min
Compute sampling variance of the sample variance: Var(s²) ≈ [2 × (24.50)²] / 4 = 300.125 min⁴ (SE(s²) ≈ 17.32 min²)

Interpretation

Sampling variance reflects estimation uncertainty. While sample variance describes the spread of individual observations, sampling variance measures how reliably the sample statistic approximates the population parameter. The table below outlines practical interpretation guidelines:

Observed Pattern	Statistical Meaning	Research Impact	Methodological Action
High Sampling Variance	High estimator instability across potential repeated samples.	Wide confidence intervals; low statistical power.	Increase sample size n or stratify the sample.
Low Sampling Variance	Estimator is tightly clustered around the true parameter.	Narrow confidence intervals; high reproducibility.	Proceed with planned hypothesis testing.
Large s², Low Var(x̄)	High population heterogeneity controlled by large n.	Sample mean is precise despite variable raw observations.	Report both s and SE to reflect dispersion and precision.
Design Effect Deff > 1.5	Cluster sampling has inflated sampling variance.	SRS formulas underestimate error.	Apply survey weights and robust cluster variance estimators.

Assumptions and Common Mistakes

Valid sampling variance estimation relies on clear assumptions. Avoid these five common errors:

Conflating s² with Var(x̄): Reporting sample variance instead of sampling variance or standard error misrepresents estimation precision.
Dividing by n Instead of n – 1: Using n introduces negative bias because deviations from x̄ are smaller than deviations from true μ. Bessel’s correction restores unbiasedness.
Applying Formulas to Non-Probability Samples: Sampling variance formulas assume known selection probabilities. Applying them to convenience samples—such as comparing purposive vs. convenience sampling designs—invalidates inferential claims.
Omitting the Finite Population Correction: In small populations where n / N > 0.05, ignoring the FPC overestimates sampling variance and understates power.
Assuming Homoscedasticity Across Groups: Assuming equal variances when groups differ invalidates pooled standard errors, impacting residual calculations in mean squared error (MSE) models.

How to Report It (APA Style)

Under APA 7th edition guidelines (Section 6.44), statistical symbols are italicized (M, s, s², SE), numbers are rounded to two decimal places, and estimation precision is contextualized with confidence intervals.

APA Example: The laboratory assay demonstrated a mean turnaround time of 18.00 minutes with a sample variance of 24.50 min² (s = 4.95 min). The estimated sampling variance of the mean was 4.90 min², yielding a standard error of SE = 2.21 min, 95% CI [11.85, 24.15].

In tabular reports, place standard errors in a dedicated column labeled SE and state in the table footnote whether estimates reflect simple random sampling or complex design weighting.

How to Calculate It in Excel, R, or Python

Excel Formulas

For data in cells A1:A5:

Sample Variance (s²): =VAR.S(A1:A5) (Returns 24.50)
Population Variance (σ²): =VAR.P(A1:A5) (Returns 19.60)
Sampling Variance of the Mean: =VAR.S(A1:A5)/COUNT(A1:A5) (Returns 4.90)
Standard Error of the Mean: =SQRT(VAR.S(A1:A5)/COUNT(A1:A5)) (Returns 2.21)

R Script

x = c(12, 15, 18, 20, 25)
n = length(x)
sample_var = var(x) # Sample variance: 24.5
sampling_var_mean = var(x) / n # Sampling variance of mean: 4.9
se_mean = sqrt(sampling_var_mean) # Standard error: 2.2136
var_s2 = (2 * (sample_var^2)) / (n - 1) # Variance of s^2: 300.125

Python Script

import math
import numpy as np

x = np.array([12, 15, 18, 20, 25], dtype=float)
n = len(x)
sample_var = float(np.var(x, ddof=1)) # Sample variance: 24.50
sampling_var_mean = sample_var / n # Sampling variance of mean: 4.90
se_mean = math.sqrt(sampling_var_mean) # Standard error: 2.2136
var_s2 = (2.0 * (sample_var ** 2)) / (n - 1) # Variance of s^2: 300.125

Frequently Asked Questions

How do you do sample variance?

To calculate sample variance: (1) calculate the sample mean; (2) subtract the mean from each observation to obtain deviations; (3) square each deviation; (4) sum the squared deviations; and (5) divide this sum by n – 1 degrees of freedom.

What is meant by sampling variation?

Sampling variation is the expected fluctuation among sample estimates obtained from different random samples drawn from the same population due to probabilistic selection.

What is the symbol used to represent variance?

Population variance is represented by σ² (sigma squared), sample variance by s², and estimator sampling variance by Var(θ̂) or Var(x̄).

What are two types of variance?

The two primary types of variance are sample variance (s², data-level spread) and population variance (σ², population spread), or data variance versus estimator sampling variance.

What is the difference between sample variance and sampling variance?

Sample variance describes the dispersion of individual data observations within one sample. Sampling variance describes the dispersion of an estimator across hypothetical repeated samples.

Why does sampling variance decrease as sample size increases?

Because Var(x̄) = σ² / n, increasing sample size n directly reduces the denominator, allowing individual variations to cancel out and concentrating the estimate around the true mean.

Sources

NIST/SEMATECH. (2012). e-Handbook of Statistical Methods: Section 1.3.5.6 Measures of Scale. U.S. National Institute of Standards and Technology. https://www.itl.nist.gov/div898/handbook/eda/section3/eda356.htm (Read 2026-11-07).
NIST/SEMATECH. (2012). e-Handbook of Statistical Methods: Section 3.3.3.3 Selecting Sample Sizes. U.S. National Institute of Standards and Technology. https://www.itl.nist.gov/div898/handbook/ppc/section3/ppc333.htm (Read 2026-11-07).
Statistics Canada. (2003). Survey Methods and Practices (Catalogue no. 12-587-X). Minister of Industry, Government of Canada. https://www150.statcan.gc.ca/n1/en/pub/12-587-x/12-587-x2003001-eng.pdf (Read 2026-11-07).
U.S. Census Bureau. (2022). American Community Survey: Design and Methodology (Chapter 12: Variance Estimation). U.S. Department of Commerce. https://www.census.gov/programs-surveys/acs/methodology/design-and-methodology.html (Read 2026-11-07).
American Psychological Association. (2020). Publication Manual of the American Psychological Association (7th ed.). American Psychological Association. https://apastyle.apa.org/ (Read 2026-11-07).