Mind the gap
Dealing with skew

The matter of bad choice

Right on the heels of the disastrous bubble chart comes another, courtesy of the NYT Magazine.  Bubble charts are okay for the conceptual ("this is really big, and that is really tiny").  This chart wants readers to compare the sizes of the bubbles, which highlights the worst part of such graphs.

Poor scaling is the huge issue with bubble charts.  They are the prototype of what I call not "self-sufficient" charts.  Without printing all the data, the chart is unscaled, and thus useless (see below middle).  When all the data is printed (as in the original, below left), it is no better than a data table.


In the above right chart, we simulated the situation of a bar or column chart, i.e. we provide a scale.  For this chart, the convenient "tick marks" are at 10, 20, 34, 41.  Unfortunately, this scaled version also fails to amuse.

Note further that the data should have been presented in two sections: the party affiliation analysis and the gender analysis.  Also, it is customary to place "Independents" between "Republicans" and "Democrats" because they are middle-of-the-road.

Redo_pewpoll A profile chart is an attractive way to show this data.  Here, we quickly learn a couple of things obscured in the bubble chart.

On the issue of abortion, Independents are much closer to Democrats than Republicans.  Also, there is barely any difference between the genders, the only difference being the strength of support among those who want to legalize.

Reference: "A matter of Choice", New York Times Magazine, Oct 19 2008.

PS. Based on RichmondTom's suggestion, here are the cumulative profile charts.


Bernard L. suggested a "tornado" chart:

A matter of choice


Feed You can follow this conversation by subscribing to the comment feed for this post.


Your profile chart is a great improvement, but how about a cumulative profile chart, so that I could quickly see the total support for the middle positions?

Mr. Harrison

I'm an independent, but no way am I middle of the road.


RichmondTom: I've added the cumulative chart. In this case, I don't think it adds much; one issue with the cumulative is that one has to sort the categories in one specific order.

Bernard Lebelle

What about a % based tornado chart (using line bars) ? I sent a copy of a rework by mail. I considered that sorting options would come in the following order (illegal in all/ in most -> legal in most/ in all cases) with a vertical virtual baseline to seperate the 2 subsets(as we are crossing from illegal to legal which makes a clear distinction)
I also incorporated Kaiser suggestions of separating gender and politics.
PS: what's the best way to incorporate the reworks in the topic ?


I liked the cumulative chart suggested by RichmondTom because it revealed what the profile chart might fail to, that the reason more men are for legal in most cases is that more women are for legal in *all* cases. (otherwise it almost looks like men are more in favor of legality than women, which I would not expect)

The tornado chart goes even further, revaling what even the cumulative chart can't (unless it's turned upside down, which then causes problems at the other end), which is that more women than men are in favor of "*illegal* in all cases", as well as legal in all cases.

Bernard, is that the design you mailed? Good work.

Bernard Lebelle

Derek, yes it is.


The cumulative profile charts warp and falsify this data.

Just try reading a few points from the cumulative chart: 97% of democrats, republicans and independents feel that abortion should be illegal in all cases? No, that's a lie.

You can't use a scatterplot or a profile chart to show that a set of categories total to 100% (or would total to 100% if the non-answers were included). These are discrete data points, with no intermediate values, and they can't be added to each other in a scatterplot.

Beyond that, why are you using lines to imply that there are intermediate values? The connecting lines in these profile charts have no meaning whatsoever. The slopes of the line segments have no meaning whatsoever. The connecting lines don't pass your own self-sufficiency test, and only succeed in drawing more attention to themselves than to the actual data points.

Of these redesigns, the tornado chart is the only one that's even remotely plausible. The profile charts are a waste of time.


J, you're only right in that the figure for Democrats holding the most extreme or some lesser view should actually be 100%, not 97%--there are some rounding errors there. But it's completely legitimate to accumulate the data like that. I did it to income bands on my blog recently. However, since it's a given that the total will add up to 100%, it's usually best to drop the final value when accumulating.

Fortunately, Bernard's tornado chart is best of all, accumulating only the groups on either side of a yes/no divide.

Dave T

I prefer the traditional layout of a stacked bar chart rather than the tornado variation.

Here is what I mean:

Mr. Harrison

I'll add my vote for the tornado chart in this case. All the things I get from the profile chart I also get from the tornado chart, but without having to think as much.

In this example, I find the connecting lines on the profile chart to be more distraction than they're worth, but as the number of states being charted grows from four, the profile chart might start to be more useful. With more complex data, being able to compare shapes can give you a good picture.

I tend to agree with J on the cumulative profile chart. The accumulation means little in this case, and your perception is going to depend on the ordering of the values along the x axis. You can extract the data with careful study, but at first (and even second) glance it suggests that more Democrats than Republicans believe abortion should be illegal in most cases. A quick look at the other charts tells you a dramatically different story.

The comments to this entry are closed.