A positively skewed distribution is one where most values cluster toward the lower end, with a long tail stretching out to the right toward higher values. You’ll also hear it called a “right-skewed” distribution. The tail points right, and a small number of unusually large values pull the overall shape away from symmetry. If you’ve ever looked at a chart of household incomes in a country and noticed that most people earn moderate amounts while a few earn enormously more, you’ve seen positive skew in action.
How the Shape Looks
Picture a bell curve, then imagine grabbing the right side and stretching it outward. The peak of the distribution sits toward the left, where the majority of data points land. To the right, values thin out gradually but extend much further than they do on the left side. This asymmetry is what defines skewness: the distribution isn’t a mirror image around its center.
The key visual signature is that long right tail. In a perfectly symmetric distribution (like a standard bell curve), both sides taper off equally. In a positively skewed distribution, the right side tapers off more slowly, meaning there are more extreme high values than extreme low values.
Mean, Median, and Mode Shift Apart
In a symmetric distribution, the mean, median, and mode all sit at roughly the same point. Positive skewness pulls them apart in a predictable way: the mode (the most common value) is typically the smallest of the three, the median falls in the middle, and the mean is the largest. This happens because the mean is sensitive to extreme values. Those few very high numbers in the right tail drag the average upward, while the median, which only cares about the middle position, stays closer to where most of the data actually lives.
This is why median household income is often reported instead of mean household income. The mean gets inflated by the small number of extremely high earners, making it look like the “typical” household earns more than it actually does. The median gives a more representative picture of what’s normal.
Real-World Examples
Income and wealth are the classic examples. Most people earn within a moderate range, but a small percentage earn dramatically more, creating a long right tail. Research from the National Bureau of Economic Research describes income and wealth distributions as “skewed to the right, displaying thick upper tails,” meaning not only do they extend far to the right, but there are enough high earners to make that tail substantial rather than paper-thin. This pattern holds across most countries and time periods.
Other common examples include hospital length of stay (most patients go home in a few days, but some stay for weeks), home prices in a city (most homes fall in a middle range, but luxury properties push the tail far right), response times on a survey (most people answer quickly, a few take much longer), and the size of insurance claims. In general, any measurement that has a hard floor at zero but no real ceiling tends to skew positive, especially when the standard deviation is large relative to the mean.
Measuring Skewness With a Number
Rather than just eyeballing a chart, you can calculate a skewness coefficient that gives you a single number. The most widely used version is the Fisher-Pearson coefficient, which works by cubing the difference between each data point and the mean, averaging those cubed differences, then dividing by the standard deviation cubed. The cubing step is what makes the formula sensitive to direction: values far above the mean contribute large positive numbers, while values far below contribute large negative numbers.
A skewness value of zero means the distribution is symmetric. A positive value means it’s right-skewed, and the larger the number, the more pronounced the skew. A value around 1.0 indicates moderate skewness. Values above 1.0 suggest strong skewness. Most statistical software, including Python’s SciPy library, uses this exact formula as the default calculation.
Spotting Skew in a Box Plot
You don’t always have a histogram available. Box plots offer another quick way to detect positive skewness. In a right-skewed distribution, the upper whisker (extending from the box toward higher values) is noticeably longer than the lower whisker. The median line inside the box also tends to sit closer to the bottom edge of the box rather than centered. Individual outlier dots, if shown, will cluster above the upper whisker rather than below the lower one. The relative length of the whiskers is actually a more reliable visual indicator of skewness than the position of the median line within the box.
Why It Matters for Analysis
Many common statistical tests, like the t-test, assume your data follow a roughly symmetric, bell-shaped distribution. When your data are positively skewed, that assumption breaks down. Using a t-test on heavily skewed data can produce misleading results, making you think a difference between two groups is significant when it isn’t, or vice versa. The problem gets worse when the skew also creates unequal spread between groups, which is common in practice.
You have a few options when working with skewed data. The most popular approach is to apply a log transformation, which compresses the long right tail by converting each value to its logarithm. If the original data follow a log-normal distribution (which many naturally skewed datasets do), the transformed data will look approximately bell-shaped, and standard tests become valid again. Square root transformations serve a similar purpose for less extreme skew.
Alternatively, you can skip the transformation entirely and use non-parametric tests, which don’t assume any particular distribution shape. These tests work with ranks rather than raw values, making them naturally resistant to the influence of extreme observations in that long right tail. The tradeoff is that non-parametric tests are sometimes less powerful at detecting real differences, but they won’t mislead you when your data are clearly skewed.
Positive Skew vs. Negative Skew
Negative skew (left-skewed) is simply the mirror image. The tail extends to the left, most values cluster toward the higher end, and the mean gets dragged below the median. Examples include exam scores on an easy test (most students score high, a few score very low) or age at retirement (most people retire around 60 to 67, but a few retire much earlier). The same measurement and transformation tools apply in both directions. A negative skewness coefficient tells you the tail pulls left, while a positive one tells you it pulls right.