Overlapping Error Bars - Not Signifcant? Not so fast.

The overlap may not mean exactly what you think it does.

In this short clip, Aaron Reedy explains what students need to know about interpreting overlapping error bars.

Main points:

  • If 95% CIs are non-overlapping, then there is a statistical difference at the 0.05 level. But, the reverse is not necessarily true!

  • However, error bars are much better for visualizing data than they are at testing a hypothesis!

 Interested in a bit more?

This guide dives a bit deeper into the nuance of error bars

This webinar unpacks the question “Which error bar? When?”

Check out Statistics Snippets and more on the DC blog.

Subscribe to the Statistics Snippets playlist on Youtube. 

Christy ScottComment