How to Calculate Mean Absolute Deviation

Mean absolute deviation (MAD) is the average of the absolute differences between each data value and the mean of the set. Absolute value strips the sign so values above and below the mean both count as positive distance.

Intuition

Set A: 10, 10, 10 → mean 10 → every absolute deviation is 0 → MAD = 0.

Set B: 5, 10, 15 → mean 10 → absolute deviations 5, 0, 5 → MAD = 10 ÷ 3 ≈ 3.33.

Same mean as a tighter set can have a very different MAD. That is the point: MAD reports typical distance from the center without transforming units into squares.

Units

MAD keeps the original units — points, dollars, millimeters, seconds — which makes classroom and business explanations straightforward. Variance does not; it lives in squared units until you take a square root for standard deviation.

Because the formula is linear in absolute error, you can explain a result in plain language: “Typical distance from the class mean is about 1.4 points.” That sentence lands faster in a meeting than “sample variance is 3.7 points squared,” even though both describe spread.

For a list x₁, x₂, …, xₙ with mean x̄:

MAD = ( |x₁ − x̄| + |x₂ − x̄| + … + |xₙ − x̄| ) ÷ n

In compact form: MAD = (1/n) Σ |xᵢ − x̄|.

Population vs sample wording

Many intro stats courses always divide by n for MAD. Some tools and texts discuss sample variants or compare MAD to sample standard deviation (which divides sum of squares by n − 1). When software offers population and sample modes, match the mode your assignment or report requires and state which you used.

Related dispersion tools:

• Variance calculator — average squared deviation
• Standard deviation calculator — square root of variance (SD guide)

MAD, variance, and SD all measure spread; they are not interchangeable numbers on a rubric that expects a specific one.

Work a full example with data: 4, 8, 6, 5, 3.

Step 1 — Find the mean

Sum = 4 + 8 + 6 + 5 + 3 = 26. n = 5. Mean x̄ = 26 ÷ 5 = 5.2.

Step 2 — Subtract the mean from each value

• 4 − 5.2 = −1.2
• 8 − 5.2 = 2.8
• 6 − 5.2 = 0.8
• 5 − 5.2 = −0.2
• 3 − 5.2 = −2.2

Step 3 — Take absolute values

• |−1.2| = 1.2
• |2.8| = 2.8
• |0.8| = 0.8
• |−0.2| = 0.2
• |−2.2| = 2.2

Step 4 — Average the absolute deviations

Sum of absolute deviations = 1.2 + 2.8 + 0.8 + 0.2 + 2.2 = 7.2.

MAD = 7.2 ÷ 5 = 1.44.

Interpretation: on average, values sit 1.44 units from the mean 5.2.

For long lists, use the online MAD calculator so you do not lose a sign or a decimal mid-table.

Both measure spread around the mean. They process deviations differently.

Feature	MAD	Standard deviation (SD)
Core operation	Average of |x − mean|	Square root of average squared (x − mean)
Units	Same as data	Same as data
Outliers	Linear penalty	Heavier penalty (squares)
Common in	Intro stats, interpretable reports	Inferential stats, normal models, z-scores
Always ≥ 0?	Yes	Yes

Same data, different numbers

Using 4, 8, 6, 5, 3 again: MAD = 1.44. Sample SD is about 1.92 and population SD about 1.72 (from the squared-deviation path). SD is larger here because squaring boosts the bigger misses (2.8 and 2.2) more than absolute value does.

When MAD is clearer

• You are teaching “average distance from the mean” without squares
• Stakeholders freeze at “variance” or “σ” but understand absolute error
• You want a robust-feeling summary that does not explode as fast on one outlier as SD can

When SD is usually required

• Coursework that specifies sample SD or variance
• Normal-distribution rules of thumb (68–95–99.7)
• Z-scores: z = (x − mean) / SD — see the free z-score calculator

Pick MAD when communication matters as much as convention.

Strong use cases

• Homework that asks for MAD explicitly — do not substitute SD
• Quality dashboards for mixed audiences — “typical error about 1.4 mm from target mean”
• Comparing two classes or batches with the same units and similar centers
• Sanity-checking SD — if MAD and SD tell wildly different stories, inspect outliers and data entry

Weak use cases

• Any formula or software pipeline hard-coded for variance/SD
• Inferential tests and confidence intervals built on SD assumptions
• Comparing spreads across different units without standardization

Median absolute deviation (also sometimes abbreviated MAD in advanced stats) is a different statistic centered on the median. If your text says “mean absolute deviation,” use the mean; if it says “median absolute deviation,” follow that definition instead.

Clean data in, clear MAD out.

1. Open the free mean absolute deviation calculator.
2. Enter each value in the same units (no mixed cm and inches).
3. Confirm the mean the tool reports matches what you expect.
4. Read each absolute deviation if you are learning or checking homework.
5. Note the final MAD and any SD comparison the tool shows.
6. Choose population vs sample mode only if your assignment distinguishes them.
7. Re-run after fixing typos — a misplaced decimal inflates every deviation.

Reporting tip

Publish MAD next to the mean and n: “mean = 5.2, MAD = 1.44, n = 5.” Spread without center and sample size is hard to interpret.

Most errors are process slips, not deep theory.

• Forgetting absolute value — signed deviations sum to ~0 and destroy the measure
• Dividing by n − 1 because “that is what SD does” without instruction to do so
• Using the median as the center when the problem says mean absolute deviation
• Comparing MAD to SD as if they must match numerically
• Mixing groups (e.g., two different exams) then interpreting one MAD as if it were a single process
• Leaving units off in a written answer

This guide is educational statistics for learning and checking work. For research design or high-stakes analytics, follow your field’s required dispersion measures and software defaults deliberately.

When you present MAD in a report, add one sentence of method: “MAD is the mean of absolute deviations from the arithmetic mean (÷ n).” That single line prevents a reader from assuming you computed sample SD or median absolute deviation under the same three-letter acronym.