Currently for line graphs I am often plotting the median with IQR error bars or the mean with Confidence Intervals. Currently if I select Q3 for upper error bar of my median line Spotfire actually plots Q3 + the median. I then have to create a custom expression that subtracts the median from Q1 to get the error bar that matches what is typically graphed. The same is true for mean and Confidence intervals, I have to create a custom expression to subtract the mean from U95 to get the error bar that I want. When selecting Q3 and Q1 this should match what you see in a box plot and it does not (makes error bars too large). This seems inconsistent with other graphing programs. For scientific publications I have not seen anybody use error bars that include the median or mean.
I plot hundreds of graphs as do many other in my biotech company and this is a tedious process as it is now.
Hi, I also agree with Julie it is cumbersome to need to write a formula to plot the inter-quartile ranges or confidence intervals as the error bars instead of being able to directly select them in the menu. Thanks for your consideration!
Thank you for the response Niklas. You are correct in what I meant. The same is true for the lower error bar in that it includes the median in addition to the quartile.
Julie, thanks for contributing. While I think I understand the intention, I do have some trouble understanding the details. You wrote "Currently if I select Q1 for upper error bar of my median line Spotfire actually plots Q3 + the median." Did you mean: "Currently if I select Q3 for upper error bar of my median line Spotfire actually plots Q3 + the median." ?