Mean, median, and mode describe the center of data in different ways. The mean balances the numerical total, the median marks the middle of the ordered observations, and the mode identifies the most frequent value or category. They may agree in a symmetric data set, but they answer different questions and can separate sharply in skewed data.
One data set, three measures of center
For the values 2, 3, 3, 7, and 10, the sum is 25 and the count is 5, so the mean is 5. The ordered middle value is 3, so the median is 3. The value 3 occurs most often, so the mode is also 3.
The mean does not need to be an observed value. The median depends on order rather than the size of every gap. A mode may be numerical or categorical: “blue” can be the mode of favorite colors even though a mean color is meaningless.
An outlier pulls the mean before it moves the median
The interactive data set keeps four observations fixed. Move the fifth value and compare the two summaries. The mean changes continuously because every observation enters its total; the median changes only when the order places a different value in the middle.
Test an outlier
Move the final observation to compare the mean's sensitivity with the median's resistance.
That resistance is why median household income or median home price is often reported for right-skewed distributions. It does not make the median universally superior. If totals and additive effects matter, the arithmetic mean retains information that the median intentionally ignores.
Even samples have two middle observations
Sort the values before locating the median. With an odd count, take the single middle observation. With an even count, take the arithmetic mean of the two central observations, not the mean of the entire data set.
For 4, 7, 9, and 12, the middle pair is 7 and 9, giving median 8. The result need not appear in the sample. If the values are ordinal categories rather than numerical measurements, averaging two category codes may be inappropriate; report the two central positions or use a convention suited to that scale.
Mode may be unique, multiple, or absent
A data set has one mode when one value has the highest frequency, multiple modes when several values tie for that highest frequency, and no mode when every value occurs equally often under the chosen convention. Grouping continuous measurements into intervals can change the modal class, so state the bin boundaries.
The mode is useful for the most common shoe size, product option, or category. It does not account for how far other observations lie from it. Reporting a mode of 5 tells nothing by itself about whether remaining values are 4 and 6 or 100 and 200.
A worked outlier comparison
Start with 10, 11, 12, 13, and 14. Both mean and median are 12. Replace 14 by 100: the median remains 12, while the mean rises to 29.2. The mode is absent in both lists because no value repeats.
This comparison should guide reporting, not hide inconvenient data. Check whether the extreme value is a recording error, a valid rare event, or evidence of a mixed population. Often the clearest summary reports a center together with sample size, spread, and distribution shape instead of relying on one number.
