Variance is a statistical measure of how spread out the values in a dataset are relative to their mean. A variance of 0 means every value is identical; a larger variance means the values are more spread out. Mathematically, variance is the average of the squared differences between each data point and the mean. Because the differences are squared, variance is always non-negative, and its units are the square of the data units (e.g., if data is in meters, variance is in m²).

What is the difference between population and sample variance?

Population variance (σ²) is calculated by dividing the sum of squared deviations by n (the total number of values). Use it when your dataset is the entire population. Sample variance (s²) divides by n − 1 instead, applying Bessel's correction. Use it when your dataset is a sample drawn from a larger population — the n − 1 denominator makes the estimate unbiased. For example, measuring the heights of all 30 students in a class gives population variance; measuring 10 of those 30 students gives sample variance.

What is Bessel's correction?

Bessel's correction is the adjustment of dividing by n − 1 instead of n when calculating sample variance. Without this correction, using n systematically underestimates the true population variance because a sample tends to be slightly closer to its own mean than to the population mean. Dividing by n − 1 corrects for this bias, producing an unbiased estimate of population variance. The correction has the greatest effect for small samples — for n = 2, the correction doubles the estimate; for n = 100, the effect is just 1%.

What is the relationship between variance and standard deviation?

Standard deviation (σ or s) is simply the square root of variance. Variance (σ²) is in squared units, which makes it hard to interpret directly. Taking the square root brings the measure back into the original units of the data, making standard deviation more intuitive. For example, if test scores have a mean of 75 and a standard deviation of 10, you can immediately say most scores fall within 75 ± 10. The variance in this case would be 100, which is harder to interpret without the square root.

How do you calculate variance step by step?

For a dataset [4, 8, 6, 10, 2]: (1) Find the mean: (4 + 8 + 6 + 10 + 2) / 5 = 30 / 5 = 6. (2) Find each deviation (x − μ): −2, 2, 0, 4, −4. (3) Square each deviation: 4, 4, 0, 16, 16. (4) Sum the squared deviations: 4 + 4 + 0 + 16 + 16 = 40. (5a) Population variance: 40 / 5 = 8. (5b) Sample variance: 40 / (5 − 1) = 40 / 4 = 10. Population SD = √8 ≈ 2.83; Sample SD = √10 ≈ 3.16.

When do you use sample vs. population variance?

Use population variance (σ²) when you have data from the entire group you are studying — every member is represented. This is common in controlled experiments, small classrooms, or closed systems. Use sample variance (s²) when your data is a subset drawn from a larger group and you want to estimate the population's variance. In practice, sample variance is far more common in research, surveys, quality control, and data science because collecting data from an entire population is usually impractical.

What is the bias-variance tradeoff in machine learning?

In machine learning, "variance" refers to how much a model's predictions change with different training datasets. A high-variance model is overly sensitive to its training data and generalizes poorly to new data (overfitting). A low-variance model is more stable but may miss real patterns (underfitting, or high bias). The bias-variance tradeoff says you cannot simultaneously minimize both: as you reduce bias (by making a model more complex), variance tends to increase, and vice versa. Regularization techniques like LASSO and ridge regression are designed to control this tradeoff.

How is variance used in finance?

In portfolio theory (developed by Harry Markowitz), variance measures investment risk — specifically, how much a security's returns deviate from their average. A stock with high variance (e.g., an annual return variance of 0.09, meaning SD ≈ 30%) is riskier than a bond with low variance. The key insight is that combining assets with low covariance reduces the portfolio's overall variance below the average of the individual variances — this is the mathematical basis of diversification. The Sharpe ratio divides excess return by standard deviation (√variance) to measure risk-adjusted performance.

Variance Calculator — Population & Sample Variance

Calculates population variance (σ²), sample variance (s²), standard deviation, and mean for any dataset — with a step-by-step deviation table.

How to Use the Variance Calculator

This variance calculator computes population variance (σ²) and sample variance (s²) for any dataset — enter your numbers as a comma-separated list (e.g., 4, 8, 15, 16, 23, 42) and use the toggle to switch between Sample Variance (s²) and Population Variance (σ²). The calculator instantly returns both variance and standard deviation along with the mean, count, and sum of squared deviations. For datasets of 10 values or fewer, a step-by-step deviation table is shown so you can follow every calculation.

For a deeper look at central tendency alongside spread, use our mean, median, and mode calculator, which handles the same datasets with additional measures of center.

Variance Formulas

Both formulas start the same way — compute the mean, then find the squared deviation of each point from the mean. They differ only in the denominator:

Population variance: σ² = Σ(x − μ)² / n

Sample variance: s² = Σ(x − μ)² / (n − 1)

Where μ is the mean of the dataset and n is the count of values. Standard deviation is the square root of variance: σ = √σ² for population, s = √s² for sample.

Step-by-step example

Dataset: [2, 4, 4, 4, 5, 5, 7, 9]. Mean = 40 / 8 = 5. Squared deviations: (2−5)² = 9, (4−5)² = 1, (4−5)² = 1, (4−5)² = 1, (5−5)² = 0, (5−5)² = 0, (7−5)² = 4, (9−5)² = 16. Sum = 32. Population variance = 32 / 8 = 4.00; population SD = 2.00. Sample variance = 32 / 7 ≈ 4.57; sample SD ≈ 2.14.

AdvertisementResponsive Ad

Population Variance vs. Sample Variance

The choice between population and sample variance determines which denominator to use and depends entirely on your data:

Population variance (σ²): Use when you have data from every member of the group you are studying. The denominator is n. Examples: all students in one specific class, all defective parts in a finished batch, all days in a month of temperature readings.
Sample variance (s²): Use when your data is a subset drawn from a larger group, and you want to estimate the population's variance. The denominator is n − 1 (Bessel's correction). Examples: a survey of 1,000 voters representing millions, a quality-control sample of 50 products from a production run of 10,000.

In practice, sample variance is the most common choice in research, data science, and statistics courses because collecting data from an entire population is rarely feasible.

Why Square the Deviations?

Variance squares each deviation (x − μ)² for two reasons. First, squaring makes all terms non-negative — without it, positive and negative deviations would cancel each other out, and the sum would always be zero regardless of spread. Second, squaring gives more weight to large deviations, making variance sensitive to outliers. This sensitivity is a feature, not a bug — large deviations from the mean signal high variability.

The downside is that squared units are harder to interpret. A dataset of test scores has variance in "points squared." Taking the square root (standard deviation) returns the measure to the original units, making it far more interpretable.

Standard Deviation: The Practical Companion to Variance

Variance and standard deviation always go together. Standard deviation (σ or s) is simply √variance — it lives in the same units as the data and follows the empirical rule for bell-shaped distributions:

~68% of data falls within ±1 standard deviation of the mean
~95% of data falls within ±2 standard deviations
~99.7% of data falls within ±3 standard deviations

These percentages assume a normal (Gaussian) distribution. Real-world data can deviate from this pattern, but the rule provides a useful benchmark for interpreting spread. For direct standard deviation calculation with additional options, see our standard deviation calculator.

AdvertisementResponsive Ad

Related Spread Measures

Variance is one of several ways to quantify spread. The mean absolute deviation (MAD) is a simpler alternative that averages absolute — not squared — differences from the mean, making it less sensitive to outliers. The correlation coefficient calculator uses variance internally to measure how two variables move together. For categorical data, the chi-square calculator tests whether observed frequencies differ from what you would expect by chance — variance on counts rather than measurements. The range calculator provides the simplest spread measure — the difference between the maximum and minimum values — useful as a first-pass check before computing variance. When working with experimental measurements, the percent error calculator quantifies how far an observed value deviates from a theoretical value, which complements variance analysis in lab contexts.

Real-World Applications of Variance

Variance and standard deviation are foundational tools in virtually every quantitative field:

Finance: Portfolio variance measures investment risk. A high-variance portfolio has more price swings; a low-variance portfolio is more stable. The variance of returns is the basis of modern portfolio theory.
Manufacturing and quality control: Variance in product dimensions determines whether a process is within tolerance. Six Sigma processes aim to reduce variance to the point where defects occur fewer than 3.4 times per million opportunities.
Scientific research: Variance appears in ANOVA (analysis of variance), t-tests, F-tests, and regression analysis. Understanding variance is essential for interpreting statistical significance.
Education: Test score variance shows how consistently students performed. Low variance means most students scored similarly; high variance reveals a wide range of performance levels.
Machine learning: The bias-variance tradeoff governs model accuracy. A model with high variance fits training data well but generalizes poorly to new data (overfitting).

Sources & References

Variance — Statistics How To — Statistics How To
Bessel's Correction — Khan Academy — Khan Academy
Standard Deviation and Variance — Math Is Fun — Math Is Fun

Variance Calculator