What Are Mean, Median, and Mode?
Most students can define mean, median, and mode just fine — the part that trips them up is knowing which one to reach for when it actually matters. All three answer the same question — where does the data live — but they answer it differently enough that choosing wrong can wreck an entire analysis, and the difference between them shows up the moment your data stops being symmetrical.
- Mean (average) — sums every value and divides by the count, which makes it sensitive to extremes. The U.S. Census Bureau reports median household income around $83,700 (2024 data) — but mean household income runs tens of thousands higher because a handful of high earners drag it upward, and that gap is pure outlier distortion.
- Median — sits dead center in sorted data, splitting the set into two equal halves. The reason infrastructure teams track median latency instead of mean is that a handful of slow requests wreck the average but leave the median untouched, which gives you a more honest picture of typical performance.
- Mode — whichever value shows up most frequently. Datasets can have zero modes, one, or several — multimodal distributions often reveal distinct subgroups hiding in the data that a single mean or median would blur together.
How to Calculate Each One
Mean
Mean = Σx / n
Add all the numbers together and divide by the count – like (3 + 7 + 7 + 12 + 15) ÷ 5 gives you 8.8.
Median
Arrange values from smallest to largest. Odd count? Grab the middle one. Even count? Average the two middle values. Like [3, 7, 7, 12, 15] – median's 7.
Mode
Count how many times each number appears. The highest count wins – that's your mode. [3, 7, 7, 12, 15] – 7 shows up twice, so it takes the crown.
Mean vs Median: Which One to Use?
The U.S. Bureau of Labor Statistics uses the median — not the mean — for its wage summaries, and the reason boils down to skew: income distributions lean so heavily to the right that a handful of high earners drag the mean tens of thousands of dollars above what a typical household actually brings in. Symmetrical data without outliers makes the mean the better choice, but the moment the distribution leans to one side, the median captures what is genuinely typical and the mean starts lying to you.
When your mean and median start pulling apart by more than 10-15 percent, that gap is your early warning system for skew — income, real estate prices, and hospital length-of-stay are the classic offenders where the mean gets dragged sideways by a long tail. The median almost always reflects the more representative center in those cases because it ignores the extreme values that the mean cannot stop chasing.
Frequently Asked Questions
What if my dataset has no mode?
When every value shares the same frequency — common with continuous measurements or completely unique data — no single mode exists, and that is not an error. Any intro stats textbook will tell you that is not an error — it simply reflects uniform frequency across all observed values.
Can a dataset have multiple modes?
Two modes makes it bimodal, three gives you trimodal, and anything beyond that is multimodal. Bimodal distributions frequently point to two distinct subpopulations mixed together — commute times peaking at 15 and 45 minutes usually means you have walkers and drivers, not one group.
How do I calculate mean, median, mode in Excel?
Use =AVERAGE() for mean, =MEDIAN() for median, and =MODE.MULT() for mode in Excel or Google Sheets. The legacy MODE() function only returns the first mode it finds and silently discards the rest, which means multimodal datasets get misreported unless you switch to MODE.MULT().
When should I use mean vs median?
The mean works best with symmetric, outlier-free data. The median handles the messy stuff — income, house prices, hospital stay durations. When the two diverge by more than 10-15 percent, that gap almost always points to skew that the mean is hiding.