3

It seems typical to include only basic statistics on subject demographics within papers (i.e. mean, max and min). It seems to me that having a knowledge of the actual distribution would be provide more insight, particularly if this could be included in a concise space-efficient manner.

To this end, I have experimented with the inclusion of histograms inline with the text following the sparkline concept. Example below: sparkline histograms

Unfortunately, I have received the following comment from a reviewer:

What is that?? This is the first time I see results presented that way!!!

Obviously they have either not understood the data presentation technique or do not like it. It does not seem beneficial to me to expend the space explaining the presentation technique within the paper. So I have the following questions:

  1. Is this is a useful addition to the paper?
  2. Is it obvious what these mean?
  3. What improvements could be made to the way it is presented? (without taking up too much paper space)
  4. Are there any good alternatives that I'm missing?
kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
welf
  • 133
  • 1
  • 8
  • I appreciate that this is somewhat opinion-based, so I would welcome comments on either how to rephrase, or where else might be appropriate to pose this question (possibly the [academia stackexchange](https://academia.stackexchange.com/questions/tagged/graphics)?) – welf Jan 07 '19 at 04:34
  • I think that's creative! Maybe useful. When I create graphs I go through *many* iterations of design and refinement to really try and maximize legibility and convey the message I would like to; keep trying. One thing I immediately tried to do was read the mean and median in your histograms: a tick mark for one and a tiny dot or diamond below the horizontal line could represent). Another issue is that I am careful about captioning my graphs: I want the caption + graph to be sufficient for an educated reader to apprehend my meaning: you have no caption! Where is the roadmap for your audience? – Alexis Jan 07 '19 at 05:16
  • @Alexis Thank you for the feedback. The addition of mean/median to the graphics is an interesting idea; however I'd be concerned that it might make things too cluttered. Similarly with the addition of captions, this will make the graphic too large to include inline, requiring a separate figure and reference within the text which will considerably increase the space required. Other types of sparkline do not typically have captions with them. – welf Jan 07 '19 at 06:08
  • I was unclear: I think a *single* caption/sentence etc. motivating and explaining sparklines is warranted if it is *not* standard in your literature: your audience deserves it. Adding median and median could be conceptually uncluttered since (a) your audience are almost certainly familiar with box and whisker plots, and (b) your give the actual number of the median right next to each graph; adding mean and median could be visually uncluttered by careful graphic design *a la* marks that do not descend farther below the horizontal line than the bracketing numbers, very light line weight, etc. – Alexis Jan 07 '19 at 15:45
  • @Alexis Thanks for the clarification, I will look at adding a brief single sentence to explain and test out the addition of mode/median marks. If you would like to expand your suggestions to a full answer then please go ahead. – welf Jan 07 '19 at 19:43

1 Answers1

2

I think this would be a great addition to a paper and think they work well in the form you have presented them. I would avoid adding more detail to them, since that would lead to visual clutter.

I agree with @Alexis that a very short description of them in the paper would help confused readers.

Hope this becomes standard in your field, and others too!

mkt
  • 11,770
  • 9
  • 51
  • 125