What is mean absolute deviation?

Mean absolute deviation (MAD) is the average of the absolute differences between each data point and the mean of the dataset. MAD = Σ|xᵢ − x̄| / n. It measures the typical distance of data points from the mean. For example, for the dataset {2, 4, 6, 8, 10} with mean 6, the absolute deviations are {4, 2, 0, 2, 4}, and MAD = (4 + 2 + 0 + 2 + 4) / 5 = 2.40. This means data points are typically 2.40 units away from the mean.

How do you calculate the mean absolute deviation?

Step 1: Find the mean (x̄) of all values. Step 2: For each value, subtract the mean and take the absolute value: |xᵢ − x̄|. Step 3: Sum all the absolute deviations: Σ|xᵢ − x̄|. Step 4: Divide by the number of values (n): MAD = Σ|xᵢ − x̄| / n. Our calculator shows all four steps in a table, including each value's individual deviation and absolute deviation, making it easy to verify the calculation by hand.

What is the difference between MAD and standard deviation?

Both measure spread, but they treat deviations differently. Standard deviation squares each deviation (penalizing larger deviations more heavily) and then takes the square root. MAD takes the absolute value, treating all deviations equally. This makes MAD more robust to outliers — a single extreme value inflates the standard deviation much more than the MAD. For example, for {1, 2, 3, 4, 100}: SD ≈ 39.2, but MAD ≈ 19.2. Standard deviation is more common in inferential statistics; MAD is often preferred for descriptive purposes and in forecasting.

When is MAD used instead of standard deviation?

MAD is preferred when: (1) the data contains outliers that should not unduly influence the measure of spread; (2) you want an intuitive measure in the original units that is easy to explain to non-statisticians; (3) you are working with forecasting accuracy (MAD is a common error metric in supply chain and demand forecasting, where it is called Mean Absolute Error or MAE); or (4) you are working with ordinal data. Standard deviation is preferred in inferential statistics because it has better mathematical properties for hypothesis testing and confidence intervals.

What does a high MAD indicate?

A high MAD means data points are spread far from the mean on average — the dataset is highly variable. For example, exam scores {50, 60, 70, 80, 90} have a mean of 70 and MAD of 12 — students scored widely. In contrast, {68, 70, 70, 71, 71} have a mean of 70 and MAD of 0.8 — nearly uniform scores. There is no universal threshold for "high" MAD; it must be interpreted relative to the scale of the data or compared between datasets. MAD expressed as a percentage of the mean is called the Mean Absolute Percentage (MAP) and enables cross-dataset comparison.

How is MAD related to the mean?

MAD is calculated relative to the mean (x̄) — specifically, it is the average absolute difference between each value and the mean. If the mean changes (for example, by adding a constant to all values), the MAD does not change because all values and the mean shift by the same amount, leaving deviations unchanged. Multiplying all values by a constant k multiplies both the mean and each deviation by k, so MAD also multiplies by k. MAD is always ≥ 0 and equals 0 only when all values are identical.

How is MAD used in forecasting and supply chain?

In supply chain and demand forecasting, MAD is called Mean Absolute Error (MAE) and is one of the primary accuracy metrics. A warehouse forecasting 500 units per week but actually selling an average of 480 or 520 units has a MAD of about 20 units per week. This helps planners set safety stock levels: a common rule of thumb is to hold 1.25 × MAD in safety stock. A lower MAD means forecasts are more accurate and less buffer inventory is needed.

What is a typical or acceptable MAD value?

There is no universal threshold — MAD must be interpreted relative to the mean of the data. A more useful metric is the mean absolute percentage deviation: MAD / mean × 100%. For business forecasting, less than 10% is considered excellent, 10–20% is acceptable, and above 20% suggests the forecasting model needs improvement. For manufacturing quality control, acceptable MAD depends on specification tolerances and is typically a fraction of the allowable variation range.

Mean Absolute Deviation Calculator

Calculates MAD — the average distance from the mean — for any dataset, with step-by-step deviation table.

How to Use the MAD Calculator

This mean absolute deviation calculator finds MAD instantly — enter your data values separated by commas in the input field. Negative numbers and decimals are fully supported. The calculator instantly displays the mean absolute deviation, mean, and count, along with a step-by-step table showing each value's deviation from the mean (xᵢ − x̄) and absolute deviation (|xᵢ − x̄|). The table also shows the running sum and final MAD calculation.

For datasets with up to 50 values, the full step-by-step table is shown. For larger datasets, the summary statistics (MAD, mean, count) are displayed without the row-by-row table. For related spread measures, see our standard deviation calculator, which shows population and sample SD for the same dataset.

The MAD Formula — Step by Step

Mean absolute deviation is calculated in four steps:

MAD = Σ|xᵢ − x̄| / n

Worked Example

Dataset: 3, 7, 7, 19

Step 1 — Find the mean: x̄ = (3 + 7 + 7 + 19) / 4 = 36 / 4 = 9
Step 2 — Find each deviation:
- |3 − 9| = 6
- |7 − 9| = 2
- |7 − 9| = 2
- |19 − 9| = 10
Step 3 — Sum the absolute deviations: 6 + 2 + 2 + 10 = 20
Step 4 — Divide by n: MAD = 20 / 4 = 5.00

The MAD of 5.00 means that on average, each value in the dataset is 5 units away from the mean of 9. Note that the outlier (19) contributes an absolute deviation of 10 — but only 10, not 10² = 100 as it would in variance/standard deviation.

AdvertisementResponsive Ad

MAD vs. Standard Deviation — When to Use Each

MAD and standard deviation both measure spread, but they have different mathematical properties and use cases.

Standard Deviation

Standard deviation squares each deviation before averaging, then takes the square root. This approach:

Penalizes larger deviations more heavily (an outlier at 3σ contributes 9× more than one at 1σ)
Has better statistical properties for inferential statistics (hypothesis testing, confidence intervals)
Is the basis for the normal distribution and most parametric statistical tests
Is more sensitive to outliers

Mean Absolute Deviation

MAD takes the absolute value of each deviation, treating all deviations equally regardless of size:

Easier to interpret — directly in the original units with no squaring or square root
More robust to outliers
Preferred in forecasting and supply chain (called Mean Absolute Error or MAE)
Appropriate when deviations should be weighted equally regardless of magnitude

For a normally distributed dataset, the relationship between MAD and SD is approximately: MAD ≈ 0.7979 × SD. Or equivalently, SD ≈ 1.2533 × MAD. This means standard deviation is always larger than MAD for the same dataset.

MAD in Forecasting and Error Measurement

In operations research, supply chain management, and machine learning, MAD is often called Mean Absolute Error (MAE) and is one of the most widely used measures of forecast accuracy. Given a set of forecast errors (actual − predicted), MAD/MAE tells you the average magnitude of error, regardless of direction.

For example, if weekly demand forecasts were off by {12, −8, 5, −3, 7} units, the MAE = (12 + 8 + 5 + 3 + 7) / 5 = 7 units — meaning forecasts were typically 7 units off from actual demand. This is directly actionable for safety stock planning, unlike squared-error metrics.

MAD is preferred over Root Mean Square Error (RMSE) when all errors should be weighted equally and outliers (unusually bad forecasts) should not dominate the accuracy measure.

AdvertisementResponsive Ad

Real-World Examples of MAD

Classroom scores: If 5 students score {70, 75, 80, 85, 90} with mean 80, MAD = (10 + 5 + 0 + 5 + 10) / 5 = 6. Students are typically 6 points from the class average.
Daily temperatures: If daily highs for a week are {68, 72, 75, 71, 69, 73, 74} °F with mean 71.7°F, MAD ≈ 2.0°F — temperatures vary about 2 degrees from the typical daily high.
Manufacturing quality: Part widths of {9.98, 10.01, 9.99, 10.02, 10.00} mm with mean 10.00 mm give MAD = 0.012 mm — typical deviation from the target dimension.

For datasets where you also need the median as a central measure or want to compare spread to the mean, our mean, median, and mode calculator provides comprehensive summary statistics alongside the mean.

Sources & References

Mean Absolute Deviation — NIST/SEMATECH e-Handbook of Statistical Methods
Mean Absolute Deviation — Khan Academy — Khan Academy

Mean Absolute Deviation Calculator

Mean Absolute Deviation Calculator

How to Use the MAD Calculator

The MAD Formula — Step by Step

Worked Example

MAD vs. Standard Deviation — When to Use Each

Standard Deviation

Mean Absolute Deviation

MAD in Forecasting and Error Measurement

Real-World Examples of MAD

Sources & References

Frequently Asked Questions

Related Calculators