Spotting Outliers

Welcome back to Reading and Describing Distributions! We are on lesson four of five, which means we are almost at the finish line for this first course. So far, we have built a solid visual toolkit: we can read the components of dot plots and histograms, classify a distribution's shape, and estimate its center and spread by eye. Every one of those skills describes where the bulk of the data lives. Now it is time to turn our attention to the values that do not fit in with the rest.

In this lesson, we will learn how to spot outliers on a graph, how to tell a genuine outlier apart from a value that simply sits at the high or low end of the main cluster, and how to think about what an outlier might mean in real-world terms.

In the previous lesson, we estimated center and spread by focusing on the region where data clusters most heavily. We deliberately set aside any isolated values sitting far from that cluster. Those isolated values now deserve their own careful look.

Imagine we track our morning commute for $20$ days and find that $19$ of those trips take between $25$ and $35$ minutes. On one day, a highway closure pushes the commute to $80$ minutes. That $80$-minute trip does not just sit at the top of the list — it lives far removed from the tight cluster of the other $19$ values, with a huge empty stretch of the number line in between. That sense of separation is exactly what makes a value an outlier.

Not every value at the edge of a distribution earns this label, though. If the longest commute were $38$ minutes instead of $80$, it would simply be the high end of a gradual range — unusual, maybe, but not truly separated. The difference between "a bit high" and "way out there" is what this lesson is all about.

An outlier is a data value that falls noticeably far from the rest of the distribution. On a dot plot, it shows up as one or more dots sitting off by themselves, separated by a visible gap from the main cluster.

Look at the dot plot above, which shows daily Uber ride costs over a month. Most dots are packed between $8 and $15, forming a clear main cluster. Far to the right, a single dot sits near $45, separated by a wide empty stretch where no data appears at all. That gap is the visual signal: the $45 ride is an outlier.

When scanning a graph for outliers, keep two things in mind:

• Look for gaps, not just extreme position. A value is suspicious when empty space separates it from the nearest neighbor in the main cluster.
• Check both ends of the distribution. Outliers can appear on the low side, the high side, or even both.