When the mean and median of a dataset are close together, it tells you the data is roughly symmetrical, without extreme values pulling one side of the distribution. In a perfectly symmetrical distribution, the mean and median are identical. The closer they are in your data, the more evenly balanced your values are around the center.
Why Symmetry Matters
The mean adds up every value and divides by the count. The median finds the middle value when everything is lined up in order. These two numbers respond to data very differently. The median only cares about position: which value sits in the center. The mean cares about magnitude: every single value contributes to the calculation, and a single extreme number can drag it higher or lower.
When both measures land in roughly the same spot, it means the values on the left side of your distribution roughly mirror the values on the right. Nothing is pulling the mean away from the center. In a perfectly symmetrical, bell-shaped (normal) distribution, the mean, median, and mode are all exactly equal. Real-world data almost never hits that ideal, but getting close to it tells you something useful: your data doesn’t have a lopsided tail, and the average is a reliable representation of what’s “typical” in your dataset.
What Pulls Them Apart
The easiest way to understand what “close” means is to see what happens when they’re far apart. The classic example is income data. Most people earn modest to moderate amounts, but a small number of people earn enormous sums. Those high earners pull the mean up while barely affecting the median. That’s why news reports on household income almost always use the median: it better represents what a typical household actually earns. The mean, inflated by top earners, overstates what most people experience.
This same pattern shows up whenever outliers are present. Imagine test scores of 92, 92, 94, 95, and 96. The mean and median are both close to 93 or 94. Now add a score of 40 to the group. The median barely moves, but the mean drops noticeably because it has to absorb that low score across the entire average. Khan Academy demonstrates this principle by showing that removing a single outlier from a dataset moved the mean by about 2.6 points while the median shifted by only 1 point. The mean is simply more sensitive to extreme values.
When your mean and median are close, it signals that this kind of distortion isn’t happening. Your data doesn’t have a long tail of unusually high or unusually low values skewing the picture.
How to Quantify “Close”
There’s actually a formula for measuring this. Pearson’s second skewness coefficient takes the difference between the mean and median, multiplies it by 3, and divides by the standard deviation. When the result is near zero, the distribution is approximately symmetric. Positive values indicate a right skew (the mean is higher than the median, pulled up by large values), and negative values indicate a left skew (the mean is lower, pulled down by small values).
As a rough guide, skewness values between -0.5 and 0.5 suggest the data is fairly symmetric. If your mean and median produce a skewness near zero using this formula, you can feel confident that neither measure is being distorted by the shape of the data.
It Doesn’t Always Mean a Bell Curve
One common misconception: a close mean and median doesn’t automatically mean your data follows a normal (bell-shaped) distribution. Symmetry is necessary for a normal distribution, but it’s not sufficient. A uniform distribution, where every value occurs equally often (like rolling a fair die), also has an equal mean and median. A bimodal distribution, one with two peaks on opposite sides, can be perfectly symmetric with identical mean and median values while looking nothing like a bell curve. The mean and median being close tells you about balance, not about the specific shape of your data.
Which Average Should You Report?
When the mean and median are close, it generally doesn’t matter much which one you use. Both give a reasonable picture of the center of your data. This is actually the practical takeaway most people need: if you’ve calculated both and they’re similar, your data is well-behaved enough that the mean is a trustworthy summary. You don’t need to worry that a few extreme values are distorting the picture.
The choice between mean and median becomes important only when they diverge. In skewed data, the median usually better represents what’s “typical” because it resists the pull of outliers. But when the two measures agree, that tension disappears. You can report whichever is more conventional for your field or more intuitive for your audience, knowing that both point to essentially the same center.