1

I am writing my Master's degree Thesis in Economics and I'm dealing with quite a lot of plots (my topic is more related to statistics, in particular I'm working with B-spline free knot placement) and I am wondering what's better, if to provide very fancy plots (for example using ggplot2) or instead to keep things simple and use the basic plots (from R). What do you prefer when reading an article? Thanks in advance.

  • 1
    What have you learnt so far when showing data? – Solar Mike Mar 3 at 18:05
  • 2
    This is why advisors exist. They need to approve your work, ultimately. – Buffy Mar 3 at 18:30
  • 1
    Check out the diagram and link I provide here (give it a vote if it helps...) : academia.stackexchange.com/a/124249/72855 – Solar Mike Mar 3 at 20:46
  • 2
    You can create very fancy plots with base R. – Roland Mar 4 at 7:23
  • 2
    I think, that you need to clarify, does "fancy" mean: "cool", but senseless effects (see Excel) or some less-conventional, but more insightful forms of visualization (in the gist of violin plot vs. boxplot). Obviously, the answers are "no" and "yes", depending on what you'd say. – Oleg Lobachev Mar 4 at 22:43
3

Plots should be clean. For instance, MSFT adds some cruft to charts (lines), that is better off culled. Also avoid 3D pie charts or 3 D column charts (makes you look like an undergrad). After that, it probably doesn't matter if you use R, some custom plotting tool or Excel or the like.

Read "Say it With Charts" book for some good advice on simple chart communication. Tells you what style shows what relationship best.

Make your figure captions shine. They should be on the longer side and give clear identification of different symbols and lines. Figure captions are some of the most read text in a paper.

4

I prefer plots which communicate information and tell the story so clearly, that I do not stop to think which software was used to create them. There are several good ways of achieving it and many ugly ways of failing it. The art of preparing a good plot (and infographics in general) is a complex matter and can not be boiled down to a one-bit answer like "do/don't use ggplot".

To learn this art, read more papers, highlight plots you like and analyze what you like about them and how it was achieved. Try reproducing a nice plot in your favourite software. Consult your supervisor and colleagues for more hints.

  • I think the recognition of a plotting tool is often almost subconscious. I have seen a lot of plots (papers and posters alike) where it was apparent to me, whether it's been done using Mathematica, Matlab or Matplotlib. But I agree with you on the oint that communication of information has to come first and foremost. – Nox Mar 5 at 8:33
  • @Nox The recognition is instant if default settings were used. – Dmitry Savostyanov Mar 5 at 15:42
1

Clean and simple is best. However, ggplot2 is excellent at producing clean and simple plots.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.