A pretty chart hides the message

James C @annelidworm sent me to this BBC chart, which he thinks is "hard on the eyes":

 

I find a few things I like, and also a few I don't.

Unlike James, I actually find the chart quite pretty. The use of a small-multiples to compare season tickets with single tickets is also nice. For someone like me, who isn't well versed in the British map, the geography lesson is appreciated - although for a local reader, this may be superfluous. The thickness of lines used to encode the data works alright.

There are a few problems with the chart:

There is a self-sufficiency problem. This is a chart in which every data element is printed on the chart, which means the graphical pieces are merely cosmetic. If the data labels were removed, the reader would be entirely lost. However, this problem can be solved by judicious use of colors.

Consider how color is used here. Blue and yellow distinguishes between season and single tickets but the small-multiples setup already does the job well enough. The tint is used in some arbitrary manner unrelated to the data, as far as I can tell.

Instead, price increases above the rate of inflation should be differentiated from price changes below the rate of inflation by using two colors. The special case of Birmingham's season ticket which increased exactly at the rate of inflation deserves its own color.

Speaking of increases relative to inflation. The analyst helpfully explains via the legend that any number above 66 percent is ahead of inflation, and any number below is behind inflation, meaning prices have actually come down. The entire dataset can be simplified by subtracting 66% from each number to show the "real" price changes.

***

Take a step back. What is the story in this dataset? The numbers on the right side are all much higher than those on the left, with Shoeburyness being a bit of the exception. It appears that the rail company is trying to push sales of season tickets. Too bad this chart doesn't bring the story to the front.

Here is one attempt using paired bar charts.

 

 

 

 

Comments

[Mick]

I'm fairly certain the tint is just there as a means to improve readability - darker tinting is applied to shorter lines to make them stand out where there is an overlap (e.g. Nottingham would be hidden by the Manchester and Leeds lines).

Also, cities are tightly-packed enough in the UK to warrant at least some labels. Without a clear scale, the point at Swindon could easily be mistaken for Bristol (which is bigger). As for Shoeburyness - I suspect you'd be pushed to find 1% of UK citizens who had even heard of it, let alone locate it on a map!

[Mat Morrison]

I live 50 miles from Shoeburyness. Had never heard of it.

But that highlights something that the BBC's chart describes that's potentially relevant to the story that's being told (I've pointedly *not* read the article to keep my perspective fresh.)

When we're talking about train journeys, distance has some relevance (even if there's no correlation.) The original chart does a very good job of communicating journey length.

There may be a certain amount of assumed knowledge, too. Certain towns are more important than others -- the inclusion of Shoeburyness is such an oddity that it highlights that for the other cities.

[conchis]

The entire dataset can be simplified by subtracting 66% from each number to show the "real" price changes

FWIW, the formula for the real price change should actually be:

(1 + nominal change)/(1 + inflation) - 1

(The nominal change minus inflation rate formula is a reasonable approximation for low inflation rates, but doesn't really work here)

[Mike Woodhouse]

I think the story is actually more sordid*, in that the very low number of season tickets sold for inter-city routes allows price increases to be kept to inflation without significantly affecting revenue.

Rail ticket price increases are subject to an inflation-based cap: I don't know for sure, but I'd guess that season-ticket prices are a factor in the formula, allowing well-above inflation increases for the tickets that are commonly purchased.

I'd guess that commuter routes would show a different pattern. My 12-mile suburb-to-City commute generally seems to increase at the maximum rate, for example.

* (or pragmatic, if you look at it from the rail companies' view)

[Meganknight]

What would have been more useful (within the context of the specific story) would be to have color-coded by rail company (since different companies run these routes).

I would love to see the sub-maps showing the commuter routes in and around London, Birmingham, Manchester, etc.

[Alex]

I think Shoeburyness is technically a commuter route, and may be subject to more regulation.

Something that would be cool - to make the distances proportionate to the fare. So Leeds is £xx further away.