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; } } }

Adding the user-defined error interval option to the graph builder

dainius

Is it possible to plot the mean and then add a summary statistics from Points-Summary Statistic as a separate error interval, which you define Yourself?

Because calculating mean and SD in different columns in JMP is very easy, but then adding them together and plotting them as a dot and error bars does not function. And this is frankly very annoying because this feature is present in many other statistical softwares .

Hereby is my suggestion for further JMP development - make a 'USER DEFINED STATISTICS' as a possibility to add as an error bar (Error Interval).

Then one could calculate and define SD separately in a separate column from a biological triplicate - and so plot them together on top of each other in all possible conditions and variations, voila.

I could see this feature is lacking for some time already (from discussion trends), and all alternatives to solve this issue around are very clumsy and code-requiring.

Thank you

Accepted answers

All comments

Dan_Obermiller

Have you seen the Interval drop zone in Graph Builder? I think this is what you are looking for:

dainius

Aha, now I see what it is doing finally. So you are saying this is a completely user-defined interval?
If so, how can you make it thicker and more visible in on the plot?
Because it appears super finely drawn and I cannot see it nicely.
What should I change in the general preferences to make it always happen - i.e. to always draw a thick line and visible end bars?

Thank you

dainius

I have added the intervals as suggested, but frankly I do not understand what it is.

Now it is plotted as: Mean as a dot +/-SD as bars

And I have zoomed in the graph.

1. Why are they different widths? What else is plotted there as a thick line - a median or a mean of my SD? How do I get rid of it then?

2. Then, how does each bar relate to the mean - how are they connected, which mean belongs to which bars? Can one just make a line going through the center of the points as in all other stat softwares?

I cannot easily understand. I think this could be improved...?

question2.png

Dan_Obermiller

I can't really tell how you created your graph, but it looks like you do not have a grouping variable defined, so you are getting intervals for each data point. See this example:

I had a data table with 3 columns: Group, Y, and Interval.

I drug the Y column to the Y-axis. I then drug Group to the X-axis and Interval to the Interval role to create the graph. Note that the Interval column is only needed because you wanted a Custom interval. If you just want something standard, you can drag the Y variable into the Interval role and specify the statistic you want to create the interval.

I believe this is the type of interval graph that most people expect. Perhaps if you share a bit more of your problem setup (or a fictional example of your data), people will be able to help more.

dainius

Dear Dan,

yes - I would like to get the error bars for each point, correct.

I have created a sample zoo-dataset, where I would like to plot the animal mean weight +/-SD in different temperatures.

And I have SD calculated for each animal in each temperature. That is, I have weighted my elephant three times using the same scales and then calculated the mean and the SD for each animal at each temperature.

Let's agree that SD of a duck kept at 37 C has nothing to do with the SD of the elephant kept in 54 C, because 1. the used scales can be different, 2. days the measurement has been done can also be different and 3. one veterinarian which weighs animals has a better eye to spot things more precisely compared to the other one. So I cannot always agree to use a combined SD from all the means and all temperatures and then put the same one on all experiments in all temperatures and days etc. I know it is easier, but this is incorrect in my particular case. Therefore, I plot SD for each animal independently.

Sorry, the internet was not working yesterday so I could not upload the dataset.

Example dataset.jmp

graph.png