Is it ok to ask "aesthetic" questions", and if so, which is the correct category?

We all know that data science is (at least) 1/2 analysis, and 1/2 communicating the analysis (and this is one of the reasons R + RStudio is such a power combo in DS). Sometimes my problem is not writing the code for a certain plot, but choosing among different plots. Of course I'm not talking about relatively trivial aspects such as which colors to use in the plot, but thing like:

  • would this be more readable using coord_flip?
  • should I dodge the bars of my bar plot? Or should I use transparency?
  • I have counts of the levels of a factor variable, for two different populations with different sample size: should I show the raw data (counts), so that it's clear which population has more individuals? Or should I normalize the data and only show the proportions?

In other words, these are data visualization questions, but they're not about (or not mainly about) ggplot2 code. I may already know how to code all the alternative plots I have in mind, but I'd like someone to give me their opinion on which plot is more clear and informative.

Can I ask such questions in this community, and if so, which is the correct category? Thanks


I would say that you are very welcome to pose such questions in this meta category (unless people feel there should be a new category for such questions).

EDIT: Don't use this meta category as per next post.


Those are great questions to ask here! Providing a place for such discussions is one of the goals of this community. We have some guidelines for asking non-code questions: FAQ: Tips for Introducing Non-Programming-Problem Discussions

But please don't use the #meta category for such topics — #meta is for discussion about this site, its organization, how it works, and how we can improve it.

I think right now non-code-specific questions about data visualization theory & practice fit just fine in #general — but if there was something very #tidyverse specific about the work you are doing (for instance), the question could also go there. I think the key element is to make clear up front that you're not asking how to implement something, you are looking to open a more general conversation about data visualization practice. The FAQ linked above has some more advice about how to tag such topics.

(Edited to clarify my antecedents :sweat_smile:)


Just want to second what @jcblum is saying here. If it's about which or why one method/layout might be more effective for communication, it's probably a fit for #general, as the question itself is independent of the tool you're using to make it.

This question (as in this thread right now) fits in #meta, but the theoretical questions probably do not.