Hi JMP Community,

I had a colleague ask me today, "What's the best way to graphically represent PSD?"

This is a tough one, and there are different ways one can go about it (several are shown below). One of the challenges with PSDs is that many of the ways that measure PSD don't measure it as a continuous "spectrum", but rather as values forming discrete levels according to the average size of the particle (usually on a mesh). The other problem is the constraint that all values must add to 100% -- i.e. all the mass of the test sample that is sieved (or sent through your particle analyzer) must add up to the total mass you used, so each level in the PSD represents a percentage of that total mass.

Typically, we might use a bar chart (first image) to show the PSD that was tested. Other options might be the line graph (second image) or filled line graph (third image), which depending on preferences might be better to show differences in sample PSDs. Each of these can be cumbersome when there are many such PSDs to consider -- here I'm only considering three. One very strange one that came to mind was a parallel plot (last example), but this is very unappealing in my opinion -- too cluttered and doesn't help to also show the overall shape of the PSD, which is also often of interest.

All of this is to prompt the question in the title of the post: what is your preferred method to represent discrete values across different levels? Do you work with similar kinds of data and need to graphically show the results? What have you found is the most "pleasing" for the audience to look at and understand the message?

There is of course no "right" answer to these questions, but my colleague's question prompted me to consider what are some better options, and if so, can the answer be crowd-sourced from the JMP community.

Thanks for your feedback!,

DS

PS - I've attached a mock-up data table of what some different PSDs might look like.