Symmetric vs skewed distribution

By Jude Wallis · Published

A symmetric distribution has left and right halves that are roughly mirror images, so the mean and median are about equal. A skewed distribution has one longer tail, which usually pulls the mean toward it. Report the mean and standard deviation for symmetric data, the median and IQR for skewed data.

AP Statistics: Unit 1 (topics 1.6 Descriptions for One Quantitative Variable Distributions, 1.7 Summary Statistics for One Quantitative Variable). Describing shape, including symmetry and skew, is topic 1.6, Descriptions for One Quantitative Variable Distributions, in Unit 1 of the Fall 2026 AP Statistics course. Choosing between the mean and standard deviation and the resistant median and IQR is topic 1.7, Summary Statistics for One Quantitative Variable. This site uses the median-excluded (TI-84) quartile convention.

Symmetric vs skewed: the short answer

Shape is judged from a graph. A distribution is symmetric when the part of the graph left of the center is approximately a mirror image of the part to the right. It is skewed when one side stretches out into a longer tail than the other. The direction of the skew is the direction of the long tail: a long tail toward larger values is skewed right, and a long tail toward smaller values is skewed left.

The distinction matters because it decides which summary statistics describe the data well. In a symmetric distribution the mean and the median land in about the same place, and the mean and standard deviation are good summaries. In a skewed distribution the long tail usually drags the mean toward it, and the median and interquartile range (IQR) are the better choice.

What a symmetric distribution looks like

In a symmetric distribution, values a given distance above the center are about as common as values the same distance below it. A histogram of such data has two sides of similar length and height. A boxplot has its median line near the middle of the box and whiskers of similar length.

Because the two sides balance, the mean and the median agree. In the eight values 2, 4, 5, 6, 6, 7, 8, 10, each value below 6 is matched by one the same distance above 6, and the mean and the median are both exactly 6.

Symmetric does not mean bell-shaped. A normal distribution is symmetric, and so is a uniform distribution, where every interval of the same width is equally likely, and so is a distribution with two equal peaks. Symmetry describes the mirror property only. For real data the realistic description is "approximately symmetric," since real samples are rarely perfect mirror images.

What a skewed distribution looks like

A skewed distribution has most of its values bunched on one side and a tail trailing off on the other. Household income is a standard example of right skew: most households are in a moderate range and a small number have very large incomes. Scores on an easy test are often skewed left: most students score near the top and a few score much lower.

The long tail contains values far from the center, and the mean responds to how far away they are, while the median responds only to how many values lie on each side. So the mean is usually pulled toward the tail. Right skew tends to put the mean above the median, and left skew tends to put it below. This is a tendency and not a guarantee, which is why skew should be read from the graph first. The two directions are compared in detail in skewed left vs skewed right.

In a boxplot, skew shows up as one whisker much longer than the other, or as a median line sitting toward one end of the box.

The differences side by side

FeatureSymmetricSkewed
TailsAbout the same length on both sidesOne tail clearly longer
Mean and medianAbout equalMean usually pulled toward the long tail
HistogramTwo sides roughly mirror each otherValues bunched on one side, tail on the other
BoxplotMedian near the middle of the box, whiskers similarUnequal whiskers or an off-center median line
Preferred centerMeanMedian
Preferred spreadStandard deviationIQR
Named directionNoneRight (tail toward larger values) or left (tail toward smaller values)

The table describes typical cases. Shape is a judgment about a graph, and small samples in particular can look mildly skewed by chance alone.

When to use which summaries

When the distribution is approximately symmetric with no outliers, report the mean and the standard deviation. They use every value, and the standard deviation combines with a normal model to give proportions and percentiles.

When the distribution is clearly skewed or has outliers, report the median and the IQR. Both are resistant, so a long tail or an extreme value does not distort them. The first worked example shows the size of the effect: two data sets with the same mean and the same IQR have standard deviations of 2.45 and 5.86 because one of them has a long right tail.

Shape also matters later in the course. In Unit 4, strong skew in a small sample is one of the things to look for before relying on tt procedures for a mean, so describing shape accurately in Unit 1 is practice for a check you will repeat. The general method for describing a graph is in how to describe the shape of a distribution.

The classic mix-up and how to avoid it

The first mistake is naming skew by the location of the peak. A distribution with its peak on the left and a long tail to the right is skewed right, because the tail names the skew. Students who look at where the data are piled up get the direction backward.

The second mistake is reasoning backward from the summary statistics. A symmetric distribution has its mean equal to its median, but equal values do not prove symmetry. In 1, 2, 4, 4, 9 the mean is 205=4\frac{20}{5} = 4 and the median is 4, yet the largest value is 5 above the center and the smallest is only 3 below it. Look at a graph first and use the mean and median as supporting evidence.

The third mistake is treating symmetric as a synonym for normal. The empirical rule and normal calculations require a bell shape. A symmetric distribution that is flat or has two peaks does not follow them.

Same mean, different shapes

Set A is 2, 4, 5, 6, 6, 7, 8, 10. Set B is 2, 3, 3, 4, 4, 5, 7, 20. For each set find the mean, the median, the IQR, and the standard deviation, and describe the shape.

  1. Set A mean: 2+4+5+6+6+7+8+10=482 + 4 + 5 + 6 + 6 + 7 + 8 + 10 = 48, so xˉ=488=6\bar{x} = \frac{48}{8} = 6. Median: the 4th and 5th values are 6 and 6, so the median is 6.

  2. Set A shape: the distances from 6 are −4,−2,−1,0,0,1,2,4-4, -2, -1, 0, 0, 1, 2, 4. Each negative distance is matched by the same positive distance, so the set is symmetric about 6.

  3. Set B mean: 2+3+3+4+4+5+7+20=482 + 3 + 3 + 4 + 4 + 5 + 7 + 20 = 48, so xˉ=488=6\bar{x} = \frac{48}{8} = 6. Median: the 4th and 5th values are 4 and 4, so the median is 4.

  4. Set B shape: seven of the eight values are between 2 and 7, and one value, 20, is far to the right. The set is skewed right, and the mean (6) is above the median (4).

  5. IQR for A: the lower half 2, 4, 5, 6 gives Q1=4.5Q_1 = 4.5 and the upper half 6, 7, 8, 10 gives Q3=7.5Q_3 = 7.5, so IQR=3\text{IQR} = 3. IQR for B: the lower half 2, 3, 3, 4 gives Q1=3Q_1 = 3 and the upper half 4, 5, 7, 20 gives Q3=6Q_3 = 6, so IQR=3\text{IQR} = 3.

  6. Standard deviation for A: the squared deviations from 6 are 16, 4, 1, 0, 0, 1, 4, 16, which sum to 42, so s=427=6≈2.45s = \sqrt{\frac{42}{7}} = \sqrt{6} \approx 2.45.

  7. Standard deviation for B: the deviations from 6 are −4,−3,−3,−2,−2,−1,1,14-4, -3, -3, -2, -2, -1, 1, 14, with squares 16, 9, 9, 4, 4, 1, 1, 196, which sum to 240, so s=2407≈5.86s = \sqrt{\frac{240}{7}} \approx 5.86.

Set A is symmetric: mean 6, median 6, IQR 3, s≈2.45s \approx 2.45. Set B is skewed right: mean 6, median 4, IQR 3, s≈5.86s \approx 5.86. The sets share a mean and an IQR, but the long right tail in B separates the mean from the median and more than doubles the standard deviation. For B, the median and IQR describe the typical values better.

Reading shape from a five-number summary

A data set of delivery times in minutes has minimum 10, Q1=20Q_1 = 20, median 25, Q3=40Q_3 = 40, and maximum 90. Use the boxplot these numbers produce to judge the shape.

  1. Compare the two parts of the box. From Q1Q_1 to the median is 25−20=525 - 20 = 5 minutes. From the median to Q3Q_3 is 40−25=1540 - 25 = 15 minutes. The upper part is three times as long.

  2. Compare the reach on each side. From the minimum to Q1Q_1 is 20−10=1020 - 10 = 10 minutes. From Q3Q_3 to the maximum is 90−40=5090 - 40 = 50 minutes.

  3. Check for outliers: IQR=40−20=20\text{IQR} = 40 - 20 = 20, so the upper fence is 40+1.5(20)=7040 + 1.5(20) = 70 and the lower fence is 20−1.5(20)=−1020 - 1.5(20) = -10. The maximum of 90 is above 70, so at least one value is an outlier on the high side. No value is below −10-10.

  4. Each of the four parts holds about a quarter of the data. The parts above the median cover far more of the axis than the parts below it, so the upper half of the data is much more spread out than the lower half.

The distribution is skewed right. The upper part of the box (15 minutes) is longer than the lower part (5 minutes), the high side reaches 50 minutes past Q3Q_3 while the low side reaches only 10 minutes below Q1Q_1, and there is at least one high outlier. The median of 25 minutes and the IQR of 20 minutes are the appropriate summaries.

Frequently asked questions

If the mean equals the median, is the distribution symmetric?

Not necessarily. Symmetry makes the mean and median equal, but the reverse can fail. The values 1, 2, 4, 4, 9 have mean 4 and median 4 and are not symmetric. Judge symmetry from a graph and treat the mean and median as a supporting check.

Is a normal distribution symmetric or skewed?

Symmetric. A normal curve is a mirror image about its mean, so its mean and median are equal. The reverse does not hold: many symmetric distributions, such as a uniform distribution, are not normal.

How do I tell which way a distribution is skewed?

Find the longer tail. If the tail extends toward larger values, the distribution is skewed right. If it extends toward smaller values, it is skewed left. The peak is on the opposite side from the tail, so naming the skew by the peak gives the wrong direction.