I don't think this needs a new scheme or indeed new anything. For
example, with bars or even marker symbols you can use one option for
fill and one for outline.
In one Tip I use various grey scales to indicate various levels on an
ordered scale but in every case insist on a dark outline colour, i.e.
mlcolor(black). See
<http://www.stata-journal.com/sjpdf.html?articlenum=gr0023>
I think the strongest case for patterned fills is when people want
different colours, but a full colour range is impossible or too
expensive and grey scale does not satisfy. That is perhaps most
convincing for choropleth (patch) maps, but even then it is difficult to
construct an unambiguous ordering with mixes of stippling, striping and
shading. For bar charts a better answer is usually to use horizontal
bars with text labels or a dot chart or indeed a table.
Nick
[email protected]
Fred Wolfe
On Fri, Feb 6, 2009 at 8:37 AM, Nick Cox <[email protected]> wrote:
> I am not clear on the connection here.
The connection ... Although I mostly agree with Nick, the point of my
email was to provide some additional evidence that at least some
journals editors want "hatched bars."
A second, perhaps hidden, point was that this journal doesn't like
grey borders. So that perhaps Stata might want to consider adding a
default non-grey bordered scheme.
As for me, I can modify the scheme myself or edit it in the graph
editor... and I was able to convince the copy editor on the telephone
that the graph was quite readable with hatching.
>
> Fred reports an exchange with an editor about the rendering of various
> greys in a diagram in a paper. The diagram shows three box plots.
>
> Fred also used the title "hatched bars", which gave Sergiy a good
> opportunity to remind us of his implementation of area shadings.
>
> Does Fred want to be able to fill the boxes of his box plots?
>
> I've always found informative text (which could be numeric) the
simplest
> way to distinguish different box plots. That applies where the plots
are
> for distinct groups, variables or combinations.
>
> Incidentally, although grey requires care, I've had no problems
> publishing Stata-produced plots as diagrams in journals in my part of
> science.
>
>
> Fred Wolfe
>
> Although this subject has been aired before, I though this letter from
> a journal copy editor to me might put the issues practically. I sent a
> paired -graph box - with three shades of grey using scheme(s2mono).
> Not one should worry, as I will fix it with the editor, but the letter
> is interesting.
>
> "Dr. Wolfe,
>
> With respect, the artwork sent for this article does not meet minimum
> standards for print reproduction <snip> ... you will note the fine
> lines and details of the text and art are blurred and turning 'grey'
> or disappearing. And complicating the problem, there's a frame of grey
> shading all around the 2 graphs (an artefact from the original slide)
> that will show up in print as a distracting mess of shadow.
>
> As a general guideline, artwork intended for publication should be
> prepared in black and white only, with no grey-colored shading or
> lines, because material in grey cannot be printed clearly with good
> contrast. Where a different 'color' is needed to differentiate data
> or columns, cross-hatching effects in black and white are useful.
>
> As well, for best quality reproduction, it is best if charts and
> graphs are sent to us in the original program in which they were
> created (such as MicroSoft Word or Excel), not as "image scans" in
> MicroSoft Document format. When an image is scanned, it inevitably
> loses some sharpness of detail. But with artwork in the original
> program we are sometimes able to make adjustments for clarity of
> print.
>
> The department of publications at one of the authors' institutions
> may be able to assist in preparation of good quality artwork that
> communicates your data effectively.
>
> Is it possible that better quality artwork could be prepared somehow?"
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
