Discussion


Note: Images may be inserted into your messages by uploading a file attachment (see "Manage Attachments"). Even though it doesn't appear when previewed, the image will appear at the end of your message once it is posted.
Register Latest Topics
 
 
 


Reply
  Author   Comment   Page 2 of 2      Prev   1   2
wd

Registered:
Posts: 167
Reply with quote  #16 
The change in slope isn't really about starting at zero - it's about aspect ratio or vertical length / horizontal length.  The graph on the left can be made to behave like a "zero start chart" simply by eliminating the white space at the bottom and shortening the vertical length without changing the distance between the scale divisions. 

Given that there is still quite a bit of art connected with dataviz, a well done chart will "look right" to the builder if he/she understands the data and its context.  To quote Edward Tufte, "above all, do no harm" - this includes the proper use of the aspect ratio such that the message conveyed is clear and undistorted.  If the addition of a zero start adds too much white space, I don't use it - but I am careful with the use of the aspect ratio.

Aspect ratio is discussed by W.S. Cleveland in The Elements of Graphing Data and by Stephen Few in Show Me The Numbers. 

__________________
Bill Droogendyk
vizwiz

Registered:
Posts: 3
Reply with quote  #17 
Thanks Bill!  Let me try to restate back to you what you're saying so that I can be sure to understand it correctly.  In the example below, has the aspect ratio changed?  

What I mean by that is, looking at North America, the steepness of the slope looks like it changed.  And this is the core of the problem.  Doesn't this mislead the reader?

From what you're saying, it's ok to not start at zero, as long as the aspect ratio stays the same.  In this example the slope from point to point has changed, but has the aspect ratio changed?  

jlbriggs

Registered:
Posts: 194
Reply with quote  #18 
"What I mean by that is, looking at North America, the steepness of the slope looks like it changed.  And this is the core of the problem.  Doesn't this mislead the reader?"

IMO, as long as the axes are properly labeled, no - not at all. 
Most line charts do not exist for the purpose of displaying data in relation to a 0 baseline.
Bars and area charts do, because you are meant to understand the data in terms of total length or total area.

The other part of why it does not mislead readers is because you can set a chart up with whatever height and width you want, whether you use a 0 baseline or not, and achieve whatever degree of slope you want.

There is no "standard" slope to go from 0 to 75 %...so how can it be misleading to have one slope vs another?

It's all about context and labeling.  I see nothing wrong with eliminating the 0 baseline for a bump chart like this at all.
What I would do is label the axis on the right as well, to allow for easier comparisons of start/end values, or perhaps only have axis labels at the start and end of each line, so each axis label is there for a specific purpose.


vizwiz

Registered:
Posts: 3
Reply with quote  #19 
Makes sense, thanks!
wd

Registered:
Posts: 167
Reply with quote  #20 
Yes, the aspect ratio has changed.  The vertical distance for 50 - 80% is about 40mm on the left graph and about 110mm on the right graph.  The horizontal length has stayed the same, thus v/h has changed.

I do think though, that this problem of possibly exaggerating slope is a bigger problem when the graph has only one line.  Then one can do amazing things with scaling.

When comparing multiple lines, as your example does, scale fades into the background somewhat.  We're comparing positions and direction on a common scale - #1 in Cleveland's hierarchy! 

__________________
Bill Droogendyk
jlbriggs

Registered:
Posts: 194
Reply with quote  #21 
"I do think though, that this problem of possibly exaggerating slope is a bigger problem when the graph has only one line.  Then one can do amazing things with scaling."

I'm not sure I really see a difference between multiple lines and a single line.  You've got more points of comparison with multiple lines, but the fact remains, no matter how many lines or what your axis ranges are, you can manipulate the slope as much as you want just by the size and shape of the chart itself.
Jeff

Registered:
Posts: 53
Reply with quote  #22 
I would question the value of such a graph if all you had was one line. As noted by several people on this thread, the actual slope of the line (aside from whether it's going up or down) isn't very meaningful; the chart above works because it invites comparisons of different slopes and because the point of the chart is that Europe and Middle East/Africa are bucking a general downward trend.

Jeff
jlbriggs

Registered:
Posts: 194
Reply with quote  #23 
Sure.

Just saying that whether the axis scales from 0 or not is not the reason such a chart with only one line could be problematic.
Previous Topic | Next Topic
Print
Reply

Quick Navigation: