Three different descriptions of center
Mean, median, and mode are measures of central tendency, but “center” has three meanings. The arithmetic mean is an equal-share value: add all numerical observations and divide by their count. The median is positional: order the values and locate the middle. The mode is based on frequency: identify the value or category that occurs most often. NIST distinguishes these location measures and notes that the mean and median can respond differently to extreme observations.
None is universally best. The choice depends on the measurement type, distribution, and question. A mean uses every magnitude and suits many roughly balanced numerical data sets. A median resists a small number of unusually high or low values and often describes a typical position in skewed numerical data. A mode identifies the most common response and is the only one of the three that naturally applies to unordered categories such as color or payment method.
How to calculate mean, median, and mode
For the mean, use mean = Σx ÷ n. For the median, sort from smallest to largest. With an odd count n, the median occupies position (n + 1) ÷ 2. With an even count, no single observation is central, so take the arithmetic mean of the values in positions n ÷ 2 and n ÷ 2 + 1. Sorting is essential; selecting the visual middle from an unsorted list can be wrong.
For the mode, count occurrences. A unique highest frequency gives one mode. If no value repeats, a finite list is commonly described as having no mode. If two or more values tie for the greatest frequency above the others, the data are bimodal or multimodal. Some statistical conventions call every value a mode when all frequencies tie, so state the convention. Mode can be a category; arithmetic mean and median require at least ordered numerical meaning.
Worked example: the three measures coincide
Input: 1, 2, 3, 3, 3, 4, 5. Formula: mean = sum ÷ count; median = the fourth ordered value for seven observations; mode = most frequent value. Calculation: sum = 21 and count = 7, so mean = 21 ÷ 7 = 3; median = 3; the value 3 appears three times. Result: mean = median = mode = 3. Interpretation: the data are centered symmetrically on the same value.
Agreement is useful evidence but not proof that two data sets have the same shape. The lists 1, 2, 3, 3, 3, 4, 5 and 0, 0, 3, 3, 3, 6, 6 both have a mean and median of 3 and a mode of 3, yet the second is more spread out. Measures of center should be accompanied by range, quartiles, a plot, or another description of distribution when variation matters.
Worked example: one outlier separates mean and median
Input: monthly amounts of 2, 3, 3, 4, and 18. Formula: mean = Σx ÷ 5; median = third ordered value; mode = most frequent value. Calculation: sum = 30, so mean = 30 ÷ 5 = 6; median = 3; mode = 3. Result: mean 6, median 3, mode 3. Interpretation: the single value 18 pulls the equal-share value upward while leaving the middle position and most frequent value at 3.
Calling 18 an outlier is a contextual judgment, not a command to delete it. It may be a valid rare event, a different subgroup, or an error requiring investigation. Report the rule used before excluding observations. In a right-skewed distribution with a long high-value tail, mean often exceeds median; in a left-skewed distribution the reverse is common. A large mean-median gap is therefore a prompt to inspect the values, collection method, and purpose.
Worked example: an even count and no mode
Input: 4, 7, 9, and 12. Formula: mean = sum ÷ count; median = average of the two central ordered values; mode = value with uniquely greatest frequency. Calculation: mean = (4 + 7 + 9 + 12) ÷ 4 = 32 ÷ 4 = 8; median = (7 + 9) ÷ 2 = 8; every value occurs once. Result: mean 8, median 8, and no mode under the common no-repeat convention. Interpretation: mean and median can agree even when neither is an observed value.
A frequent median error is choosing one of the two middle observations for an even-sized list. Both central positions are equally near the middle, so their average is used. The median may therefore be a decimal even when all inputs are whole numbers. Do not round the central pair before averaging. Empty data have neither a mean nor median; a mode also cannot be selected without observations.
Worked example: more than one mode
Input: 2, 2, 4, 5, 5, and 9. Formula: mode = category or value with the highest frequency; median for six values = average of positions three and four. Calculation: 2 occurs twice, 5 occurs twice, and all others occur once; median = (4 + 5) ÷ 2 = 4.5; mean = 27 ÷ 6 = 4.5. Result: modes are 2 and 5, while mean and median are both 4.5. Interpretation: one center number cannot express the two frequency peaks.
Reporting only “the mode is 2” would discard an equal peak. With categories, a survey containing card, card, cash, transfer, and card has mode card, but adding or ordering those labels is meaningless. This is why mode may be used for nominal categorical data while mean cannot. An ordered category scale can have a median category, but numerical differences between labels must be meaningful before a mean is defensible.
Choose the measure from the question and distribution
Use mean when totals and equal sharing are meaningful, all observations are comparable, and extreme values should influence the result. Use median when the middle-ranked experience is the target or a skewed distribution makes the mean unrepresentative, as often happens with income, house prices, or waiting times. Use mode when the most common item matters—inventory size, transport choice, or survey category—or when arithmetic operations on the values have no meaning.
Median is not automatically superior: it ignores how far observations lie from the center, while the mean preserves the total and is central to many later calculations. Mode can be unstable in small samples and may not exist uniquely. If subgroups, weights, or unequal sample sizes matter, none of the three should be calculated blindly. Inspect a distribution and sample size, then explain why the selected summary answers the practical question.
Tool capability and common reporting mistakes
The linked Average Calculator calculates arithmetic mean, count, sum, minimum, and maximum only. It does not sort data for a median, count frequencies for a mode, detect outliers, diagnose skew, or recommend a measure. You may use its mean and range outputs as part of a comparison, but calculate median and mode separately from an ordered frequency list. Do not imply that the tool produces all three measures.
Common mistakes include forgetting to sort before finding a median, averaging the two central values in an odd-sized list, declaring the first repeated value the mode without comparing all frequencies, and deleting an extreme value merely to make summaries agree. Another is averaging category codes: labels such as 1 = red and 2 = blue do not become quantitative because numbers were assigned. For a deeper arithmetic-mean procedure or a guide to interpretive failures, use the focused pages linked below. Preserve the original observations and state any tie convention so another reader can reproduce all three summaries.