11 November 2022

Published in

Warning: Graphic Content

9 min readNov 11, 2022

Contextual intercourse

Whatever the subject and whatever the design, there are some things common to every Institute for Government chart: a title on navy blue, a source, and a Creative Commons logo. That may seem obvious, but you’d be surprised how often charts circulate on social media without some of that information. It is of course part of a template, but also forms a key part of our dataviz advice: charts will often be tweeted separately from the blogpost or report they came from — how would you feel if that chart were tweeted by itself without any further context? Have you included any absolutely vital information — particularly caveats which could affect how the chart is interpreted — on the chart or in its source?

YOU CAN’T HANDLE THE TRUTH: a set of bar charts from The Times on whether people thought statements about Sunak were true or false — without the answer

Even this can’t guard against everything — I’ve had people (including certain notable political bloggers and even a political editor) lop off the credit, and further explanation and a gradual development of an argument can often help a chart make much more sense.

But it’s a useful discipline and one that came to mind a few times this week. First, there was a chart from The Times about perceptions of the Prime Minister, and whether people thought certain statements were true or not. Several people on Twitter (the only ones left, etc) thought a key detail — whether the statements were true or not — should have been on the chart, as well as in the article. I agree — the chart by itself was being used to publicise the piece and that information should have been there.

Slightly more complicated is the case of a Washington Post chart comparing Joe Biden’s approval and gas prices. The article does a good job of walking the reader through a chart type which isn’t always the easiest to interpret. The tweets I saw from the Post promoting the piece avoided using the chart itself, presumably knowing that it would prove difficult out of context. Of course, that was never going to stop some people tweeting (and criticising) the chart by itself, as Jon Schwabish explores in this dataviz critique. I agree with him that taking graphs out of context can be like taking words out of context — but also that there is more the Post could have done on the chart itself to avoid some of the criticism. Getting the balance right in situations like this can be tricky, but always ask yourself that key question: how would you feel if the chart were ripped out of its context and tweeted devoid of anything not on the chart area?

That lesson applies to other images online, too — as the Scottish Parliament found this week…

As for everything else:

A big thank you to Andy and Kirsty for being such brilliant contestants for episode 3 of The Data Game (and well done to team ODI on a brilliant summit) — you may still be able to buy a ticket and watch it back. You can definitely watch episodes 1 and 2.
My next big event is the Think Data for Government conference on 29 November.
My work with mySociety’s TICTeC Labs has been a bit of an experiment this year, converting a conference into an ongoing programme of work that commissions work after highlighting various civic tech challenges. It’s very exciting to see some of that work now being published — the second piece (after this) is a toolkit for accessibility (which, rest assured, we’re currently turning into a HTML version).
*Celebration emoji*
You’ll have seen me plug Matt Jukes’ excellent digital jobs newsletter in the past. Here are some jobs he’s hiring for at the Department for International Trade.