A focus on a "normal distribution," - also known as a "bell curve," is where most people are near the average score, and only a small group is far above or below average. However, when you're dealing with extremes, small group differences can matter a lot. 
Although a small change in performance makes no difference to the whole, the fluctuations change the character of the extremes.

Not looking at extremes

Generalizations about how two groups differ in some way often draw on stereotypes while ignoring the similarities. 
Asking for the "effect size" can prevent this error. It is a measure of how much the average of one group differs from the average of another.

Assuming statistical = real-world significance

We sometimes forget to consider "third factors," or outside causes, that could be the link between two things, because both are actually effects of the third factor.
Avoid this error by always considering more factors when you see a correlation.

Forgetting to consider outside causes

Small, daily fluctuations are often just statistical noise. For instance, in the stock market or polls.
To avoid drawing faulty conclusions about the causes, request the "margin of error" relating to the numbers. If the difference is smaller than the margin of error, there is probably no real difference.

Assuming small differences are meaningful

Just because two things change at the same time, or in similar ways, does not mean they are related.
Question the observed association. Are there many occurrences, or is this merely chance? Can you predict future associations?

Trusting coincidence

When two things are related, one might be tempted to see a causal path. For instance, that mental health problems lead to unemployment. It is possible that it is reversed, such as unemployment, causing mental health issues.
When you think about the association, ask if the reverse is possible, or even if it could go both ways, creating a feedback loop.

Getting causation backward

A graph maker may choose a small range of a larger graph to highlight a little difference or association and make it look more significant.
Take care to note the graph's labels along the axes. Question unlabelled graphs.

Deceptive graphs

Statistics is a useful tool for understanding the patterns in the world around us. But our intuition often lets us down when it comes to interpreting those patterns. In this series we look at some of the common mistakes we make and how to avoid them when thinking about statistics, probability and risk.

Assuming statistical = real-world significance

Similar ideas to Assuming statistical = real-world significance

Placebos in research

Types of people