Introduction to Median Absolute Deviation
Median absolute deviation (MAD) describes how far the typical observation in a dataset lies from its median. Standard deviation is often useful for tidy data, but one unusually high or low observation can pull it sharply upward. A typing error, exceptional event, or rare but genuine outlier can therefore make ordinary spread look much larger than the variation experienced by most of the data. MAD avoids that problem by measuring distance from the median rather than the mean and by using absolute, not squared, differences.
This Median Absolute Deviation calculator turns a list of values into its median, its unscaled MAD, and a normal-distribution scale estimate obtained by multiplying MAD by 1.4826. These results are particularly useful when most values are clustered but a small number are extreme. MAD can help summarize quality-control readings, financial returns, sensor measurements, experimental results, operational data, and other numerical series where outlier resistance matters.
Enter values separated by commas, spaces, or line breaks, and the calculator performs the median and absolute-deviation steps in the browser. The example below demonstrates why MAD remains focused on the central cluster even when an outlier is present. The optional Median Lock game uses the same principle visually: identify the middle rank rather than being distracted by values at the edges.
What Is Median Absolute Deviation (MAD)?
Median absolute deviation (MAD) is the median of the absolute distances between individual values and the dataset median. Unlike standard deviation and variance, which use the mean and squared deviations, MAD is built entirely from medians and absolute differences. As a result, a small number of extreme values have far less effect on MAD than on classical measures of spread.
For reaction times, temperatures, financial returns, or similar numerical observations, a very high or low reading can pull the mean and standard deviation toward it. The median and MAD instead summarize the middle of the ordered data, so they give a clearer indication of the spread affecting the bulk of observations when extremes are present.
This calculator finds the dataset median, calculates every absolute distance from that median, and then takes the median of those distances as the MAD. It also displays a MAD-based scale estimate for approximately normal data. The calculation is performed directly from the values you enter.
Formula for Median Absolute Deviation
For a median absolute deviation calculation with n data values:
Formula: x_1, x_2, …, x_n
the median absolute deviation is found in two steps:
- Find the sample median, denoted by .
- Find each absolute distance from that median, then take the median of those distances.
Symbolically, the MAD is
MAD = median for .
The MAD calculation can also be written in MathML as:
Because MAD uses medians and absolute values rather than squared distances from the mean, a limited number of extreme observations affects it gently. MAD is reported in the original data units, such as seconds, degrees, or dollars.
Relationship to Standard Deviation for Normal Data
For a normal (Gaussian) distribution, Median Absolute Deviation has a standard scale relationship to the standard deviation :
Formula: σ ≈ 1.4826 × MAD.
The calculator uses 1.4826 to place MAD on a scale approximately comparable to standard deviation when the data are roughly normal. For substantially non-normal data, this scaled result is not a conventional standard deviation, although the unscaled MAD remains a robust description of central spread.
How This Median Absolute Deviation Calculator Works
This Median Absolute Deviation calculator parses the entered values and applies the following steps to the resulting numeric dataset:
- Parse the input. The calculator separates values at commas or whitespace, converts numeric entries to numbers, and disregards entries that do not produce a number.
- Order the data. Values are sorted from smallest to largest before each median is found.
- Compute the median. If there are values:
- If is odd, the median is the middle value.
- If is even, the median is the average of the two central values.
- Compute absolute deviations. For every data point , the calculator finds .
- Compute the MAD. The median of the absolute-deviation list is the reported median absolute deviation.
- Calculate the displayed normal-scale estimate. The result panel multiplies MAD by 1.4826 and labels the result “Approx Std Dev.”
The result line reports the sample median , the median absolute deviation, and . Every calculation stage is centered on the data median, preventing a few unusual values from determining the robust spread summary.
How to Use the Median Absolute Deviation Calculator
To calculate MAD, paste numerical observations into the input box, submit the form, and read the median, MAD, and normal-scale estimate in the result line. The median identifies the central location of the data; MAD identifies the median absolute distance from that location. The scaled value is a comparison aid for data that are approximately normal, not a replacement for inspecting the distribution.
- Enter data values separated by commas, spaces, or line breaks. For example:
2, 4, 4, 4, 5, 5, 7, 50.
- Click the button to calculate the statistics.
- Review the displayed median, MAD, and approximate standard deviation.
- Interpret MAD in your measurement units: a smaller MAD means that the middle portion of observations is closer to the median, while a larger MAD indicates more central spread.
If the input is measured in seconds, a MAD of 0.2 is 0.2 seconds. If the input is measured in dollars, MAD is in dollars. This unit-for-unit interpretation is a practical advantage of median absolute deviation.
Worked Example of MAD Calculation
This MAD example uses the dataset:
[2, 4, 4, 4, 5, 5, 7, 50]
- Sort the data. The values are already ordered: 2, 4, 4, 4, 5, 5, 7, 50.
- Find the median. There are 8 values, so use the two middle values, 4 and 5. Their average is:
Formula: x ̃ = (4 + 5) / 2 = 4.5.
- Compute absolute deviations from the median. Subtract 4.5 from each value and take the absolute value:
- |2 − 4.5| = 2.5
- |4 − 4.5| = 0.5
- |4 − 4.5| = 0.5
- |4 − 4.5| = 0.5
- |5 − 4.5| = 0.5
- |5 − 4.5| = 0.5
- |7 − 4.5| = 2.5
- |50 − 4.5| = 45.5
The absolute-deviation list is:
[2.5, 0.5, 0.5, 0.5, 0.5, 0.5, 2.5, 45.5]
- Find the median of the absolute deviations. Ordered from smallest to largest, they are:
0.5, 0.5, 0.5, 0.5, 0.5, 2.5, 2.5, 45.5
There are again 8 values. The 4th and 5th deviations are both 0.5, so the MAD is:
Formula: MAD = 0.5 + 0.5 /2 = 0.5.
For the calculator’s normal-scale estimate, multiply MAD by 1.4826:
Formula: σ^≈ 1.4826 × 0.5 = 0.7413.
The value 50 produces a very large individual deviation, 45.5, yet it does not change the middle two entries of the ordered deviation list. MAD therefore remains 0.5 and describes the tight central cluster around the median of 4.5 rather than allowing one extreme observation to dominate the spread measure.
Comparing MAD With Standard Deviation
This comparison shows how median-based spread and mean-based spread respond differently to the MAD example.
Robust spread versus classical spread in the worked example
| Statistic |
Value in Example |
Key Property |
| Median () |
4.5 |
Central ordered value, not pulled by the outlier. |
| Median Absolute Deviation (MAD) |
0.5 |
Robust spread around the median; one extreme value does not determine it. |
| Approx. Std. Dev. from MAD |
0.7413 |
Scaled MAD (1.4826 × MAD) for approximately normal data. |
| Classical Standard Deviation |
≈ 15.56 |
Sensitive to extreme values and much larger here because of 50. |
MAD and standard deviation can be useful together. Standard deviation is appropriate when mean-based variation and all observations, including extremes, are central to the question. MAD is often more informative when measurement errors, heavy tails, or exceptional points could obscure the typical spread. A substantial difference between the two measures is a reason to inspect the observations and their distribution more closely.
Interpreting Your MAD Result
A Median Absolute Deviation result should be interpreted in the units and context of the data being analyzed. A MAD of 3 units means that the median absolute distance from the data median is 3 units. It does not say every observation is within 3 units, nor does it establish that the distribution is symmetric.
- Magnitude relative to the median. When MAD is small compared with the median, such as a median of 100 units and MAD of 2 units, the central observations are tightly concentrated.
- Units of measurement. MAD has the same units as the input. A MAD of 0.2 seconds for reaction times or 0.2 °C for temperatures has a direct practical meaning.
- Comparison across datasets. MAD values can compare groups or periods when the datasets use the same units and represent comparable quantities.
- Using the 1.4826 factor. For approximately normal data, multiplying MAD by 1.4826 produces a robust scale estimate comparable to standard deviation.
Median and MAD are also ingredients in robust z-score procedures, although this page does not calculate robust z-scores. The calculator supplies the robust center and robust spread needed for an initial review of unusual observations.
Practical Tips for Using MAD Well
Median absolute deviation is most useful when the goal is to characterize the ordinary bulk of a dataset without giving a few extreme values disproportionate influence. Early in analysis, MAD can provide a stable baseline while you decide whether unusually high or low values are errors, rare events, or meaningful parts of the process.
MAD is not a universal replacement for every measure of spread. If the question concerns tail risk, peak volatility, rare failures, or the magnitude of extreme losses, a median-based statistic may intentionally underemphasize the observations of greatest interest. Pair MAD with percentiles, tail-focused measures, or domain-specific metrics when those extremes matter.
For a useful diagnostic, compare MAD with a sorted list, box plot, or histogram. A small MAD alongside very distant values indicates a tight center with notable extremes. A small MAD and a much larger standard deviation often signals influential outliers. These comparisons help turn the calculated MAD into evidence about the structure of the dataset.
Assumptions and Limitations of Median Absolute Deviation
Median absolute deviation is a robust spread statistic, but its interpretation and this calculator’s input behavior have important limits.
- Numeric input only. The calculator is intended for numerical observations. Text labels and other nonnumeric data do not provide meaningful distances or medians.
- Dataset size. MAD can be computed from one or more numeric values. A one-value dataset has MAD zero; very small datasets may not provide a stable picture of variability.
- Data scale and type. MAD is appropriate where ordering and differences have a meaningful numerical interpretation. It is not suitable for purely categorical labels.
- Skewed or multimodal distributions. MAD still reports median distance from the median for skewed or multimodal data, but one number cannot reveal every important feature of a complex distribution.
- Normal-approximation factor. The 1.4826 multiplier is tied to a normal-distribution assumption. For data far from normal, the displayed approximate standard deviation should not be treated as the conventional standard deviation.
- Outliers versus real extremes. Reducing the influence of extremes is useful for robust central summaries, but it can be a drawback when those extremes are the main subject of the analysis.
- Rounding and numerical precision. The displayed values are rounded to four decimal places. Floating-point representation can cause small differences from hand calculations or other software.
Used with these limits in mind, MAD provides a clear answer to a specific question: how far the middle of the data typically sits from its median, without allowing a few extreme values to set the scale.