3

I'm told there are conventions in scientific papers around graphs. I'm publish material for a general audience based on a the findings of a scientific paper (unpublished) and having a disagreement with the author of the paper about how graphs must be presented.

I'm specifically asking here about the conventions for scientific publishing. I'm very aware that conventions outside scientific papers for graphs are much more open, I take my visual data cues mostly from Edward Tufte's books.

I'm being told that displaying horizontal grid lines implies a greater accuracy in modelling data and therefore should be absent in the case of this carbon sequestration modelling since it's not the results of measurements? (I would have thought significant figures on axis, axis spacing and fundamentally the caption explaining the data source and assumptions were more relevant to that)

I'm told that titles are a no go, captions only. (I've found a Uni spec online for science papers saying titles are mandatory). I'm told titles are rare in journals.

Is there any right or wrong to these matters of convention or is it just opinion?

  • I'm pretty sure grid lines are unacceptable by APA standard, which is typical in social science journals (APA = American Psychological Association). I know this is irrelevant to your field; just giving a datapoint. – Ana Feb 28 '14 at 8:28
  • In my field, journals and conferences that publish usually specify details such as usage of captions/titles and the placement of captions/titles in relation to a figure. (Graphs being regarded as figures.) I've never heard of any standards regarding grid lines. – trutheality Feb 28 '14 at 8:54
  • 1
    displaying horizontal grid lines implies a greater accuracy in modelling data — what is this i don't even – JeffE Feb 28 '14 at 12:07
  • If displaying horizontal grid lines implies a greater accuracy in modelling data is true then you should suggest removing the y axis as well. – Penguin_Knight Feb 28 '14 at 14:12
5

On grid lines

It depends on the points you would like to make with the graph. If you're just going to show an upward or downward trend, then the grid lines are probably redundant. If you'd need refer back to a certain point of a curve, and knowing the vertical position of that point would be crucial, then grid lines can help. It's not about the graphs (or I may go so far to say even within publication culture,) it's about the points you are trying to get across. If the grid lines will get people there with less puzzling or work, then yes to grid lines. In all other occasions, then no.

I'm being told that displaying horizontal grid lines implies a greater accuracy in modelling data and therefore should be absent in the case of this carbon sequestration modelling since it's not the results of measurements

This is perhaps the oddest graphical rule I have heard in the last 12 months. For most grid lines are just an extension of the tick mark on the axes. As long as you provide the tick marks on the y-axis, anyone can draw horizontal grid lines.

It is, however, not advisable to provide tick marks or grid lines finer than what your instrument or model can discern. For example, if your measurement or prediction is in the unit of meter. Then, at most I'd just put grid lines at 0.5 m increment. I wouldn't go so far to put 0.01 or even 0.1 m increments. That would imply some precision that I never had. I believe your partner author's concern may be more related to this problem. In that case, you two need to talk and make sure at least the tick marks make sense.

On caption vs. title

I'm told that titles are a no go, captions only. (I've found a Uni spec online for science papers saying titles are mandatory). I'm told titles are rare in journals.

Yes, they are rare in my field (biomedical.) We use captions (located below the graph) most of the time. The caption usually starts with something like this:

Figure 1. The [would have been title]
[texts explaining the graph.]

If you have a title in the illustration, it only serves to duplicate information, making it redundant ink.

Though, depending on fields and journals, the rule may differ. Check with the journal's guideline and other published work in that journal for clues.

  • Thanks for these comments. This is a publication for a general audience so I'm going with my instincts and design trends at large towards simplification. Was interested about the standardisation around these ideas in academic publishing. – Alastair Leith May 7 '14 at 5:25
3

There is no right or wrong when it comes to grid lines. There may be conventions varying between disciplines. The basic question of whether to use such lines or not, is if they add something useful to the reader to better understand the data displayed. A go figure should communicate as many thought as possible to the reader without to much effort. If you want to get some ideas of thinking about graphics, try to locate the book The visual display of quantitative information by Edward Tufte. There are many constructive thoughts about displaying information there worth considering.

In the end you need to look at how others publish similar data and figure out if a "standard" has developed. It may not be the best way to display data but since many are familiar with the format it becomes an efficient communication. Otherwise you should try to display the data as clearly as possible, lines or not.

  • 1
    See Are gridlines and grey backgrounds chartjunk and should they be used only on an exception basis? for some discussion on this at the stats site. – Andy W Feb 28 '14 at 13:11
  • As I mentioned in the OP I do take many cues from Edward Tuffe on visual communication of information. Have several of his books including the one you cite. I'm confident from a comms point of view of what I'm arguing for but I was curious if this claim about implied accuracy of modeling in scientific papers had any basis in fact. – Alastair Leith May 7 '14 at 5:13
  • @Andy W I should have mentioned that by using faint gridlines I am able to remove the solid vertical-axis which makes for less chart-junk in my view. There's an obvious trend towards this presentation of column/line/scatter graphs in the design world. It's cleaner in my view and lets the data take centre stage, not the apparatus of the axis and ticks. Removing that left hand vertical element seems to remove some kind of hierarchical impediment to 'seeing' the data for me, however subtle this may be. – Alastair Leith May 7 '14 at 5:19
  • Regards audience you hit the nail on the head. That's the nub of the disagreement. For me and the publisher it is the general public, for the author it would be academics (since the paper remains unpublished at this time). – Alastair Leith May 7 '14 at 5:22

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.