nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

How to display standard error bar values in distribution graph

Poparad

I have a distribution graph of a categorical variable with three levels. Two of the levels have very similar distribution percentages and I'm trying to see if their difference is larger than their error bars but I can't see it visually on the graph and I can't figure out how to see what the error bar values are. Also, I'm a little unclear on if the error bars are showing the standard error, the confidence interval, or something else. I've attached a screenshot of the graph in question. Thanks!

Accepted answers

Ben_BarrIngh

Hi @Poparad ,

It isn't currently possible to show the Std Error Bar values in the distribution platform, it may be worth requesting this in the JMP Wish List. As an alternative, have you thought about using the confidence intervals (red triangle > confidence interval > 0.95) to compare each group and see the true population distribution? This will incorporate the std error into the calculation.

Thanks,

Ben

All comments

Ben_BarrIngh

Hi @Poparad ,

The error bars are standard error and are activated via the red triangle > histogram options > std error bars (see here for more details). The Chisquare test you are using demonstrates that there is a significant difference in your groups despite any error that is present when the value of p <0.05.

Thanks,

Ben

Poparad

Yes, as you can see in my attached picture I have the error bars turned on. My question is how can I know what the range of the standard error is? The "middle" and "top" conditions are very close to each other and I'm trying to determine if their error bars overlap or not. From the picture, it's too close to tell so I'd like to know the exact value of the error bar range to see if their difference is within that range or not.

Ben_BarrIngh

Hi @Poparad ,

Thanks,

Ben

Poparad

Thanks! The confidence intervals will work for my purposes.