StatisticsDescriptive Statistics

Knowledge guide

Right and Left Skew: Read the Tail Before Comparing Mean and Median

Identify right and left skew from the direction of the tail, using new mirrored data sets and careful mean-versus-median checks.

Name the distribution by its tail

A right-skewed distribution has a longer or more extended tail toward large values. A left-skewed distribution has a tail toward small values. The direction describes the tail rather than the side containing the peak or most observations.

A symmetric distribution has matching shape on either side of a center. A finite sample need not be perfectly symmetric to look approximately symmetric, but a pronounced one-sided tail gives evidence of skew.

New example: mirrored waiting-time models

Consider two constructed lists, measured in arbitrary time units:

Data setOrdered valuesMeanMedian
A2, 3, 4, 4, 5, 6, 2044/7≈6.2944/7\approx6.294
B0, 14, 15, 16, 16, 17, 1896/7≈13.7196/7\approx13.7116

In A, most values cluster between 2 and 6 and a long extension reaches 20 on the right. In B, most values cluster between 14 and 18 and the extension reaches 0 on the left. B is the reflection obtained by replacing each value xx in A with 20−x20-x and reordering. These lists demonstrate shape; they are not measured observations.

Why the mean is often pulled toward the tail

The mean uses the magnitude of every value, whereas the median depends on the middle position after ordering. Replacing an extreme upper value with an even larger number changes the mean by that increase divided by the sample size. If the middle ordered observation stays unchanged, the median stays unchanged.

This explains the usual pattern: right skew often accompanies mean above median, and left skew often accompanies mean below median. This is a tendency, not an if-and-only-if classification rule. Discrete, multimodal, or otherwise irregular distributions can depart from it.

A transferable check

First retain all observations, including repeated values. Build a dot plot or frequency table with zero-frequency gaps visible when they matter. Locate the concentration of values, compare both tails, and then calculate centers as supporting information. State the shape and the evidence together. Do not classify a distribution solely from which side holds the highest bar or from mean-versus-median order.

Related question

Apply this knowledge

Use the concept guide to understand the reasoning, then return to the complete question and worked answer.

Right Skewed vs Left Skewed: Classify a Nine-Value Data Set

Sources

These references support the core concepts and interpretation boundaries explained above.