Why show error bars?

To convey to the viewer the expected range of values that might be expected

Between the whiskers is the total confidence interval (CI) within which you are working:

• Typically this might be: 90%, 95%, or 99%
• These correspond to 10%, 5%, and 1% probability that the true value is outside this range

---

Transcript

Now, we will shift gears a little bit to one of my perennial favorites for a plot, and that is error bars.

And you might ask yourself, "Why do you bother showing error bars".  And the reason is you want to convey to our readers the expected range of the values that we expect. So between the whiskers on the error bars and these whiskers show our confidence interval (and that might be set to 90% or 95% or 99% percent) - that corresponds to in the 90% case, that 10% of the values could be expected outside of that range.  And we would say "it still basically fits our model, it's not surprising, we expect some values outside that range", but as we increase our confidence interval, then a smaller and smaller probability should be for "true values" so correctly measured values and that are values that we believe should be outside that range. So if we have a very high confidence interval, then there should be a very small number of points outside that range.