rhythmaning: (Armed Forces)
rhythmaning ([personal profile] rhythmaning) wrote2009-06-02 03:34 pm
Entry tags:

Abuse of Graphics

I get annoyed at people who either deliberately or through a lack of understanding misquote or otherwise abuse numbers or statistics. I have been known to shout at the radio when either journalists or politicians misstate or obfuscate, using numbers in an attempt to prove their argument.

[livejournal.com profile] frankie_ecap looked at the leaflet that popped through my door courtesy of the LibDems, and pointed to this graphic:

libdems001



In case it isn't clear, it compares the number of votes of the three main parties in the last general election - Labour (15,138), LibDems (14,664), Conservative (9,559), with 2,314 other votes.

It is trying to re-inforce the view that the LibDems are only 474 votes behind Labour, and a vote for the Tories is wasted.

The numbers are right, but the graphic distorts the picture dramatically. It shows the Tory votes to be somewhat less than half of the LibDems, and barely ahead of other votes. In fact, the Tory vote was 65% of the LibDems, and over four times the size of the votes for others.

I put the numbers into Excel. Here's what the chart should actually look like:

Untitled-1



A rather different picture. The the website of my local LibDem candidate shows a more accurate picture to the one on the leaflet, too:



You probably know by now that I am a LibDem voter. I have been assured that the candidate is a top bloke. But this misrepresentation of the numbers - a blatant distortion - may well have lost him my support.
matgb: Artwork of 19th century upper class anarchist, text: MatGB (Default)

[personal profile] matgb 2009-06-02 04:02 pm (UTC)(link)
Hah!

They've actually put the full numbers onto the chart. The campaign for honest barcharts makes a step forward, even if only slightly.

I'm actually surprised they showed the 'others' at all, I'd have cut it off at, say, 5000 votes, which would have the same effect but be more honest (as long as the 5K is marked.

There's a huge amount of debate on the efficacy of barcharts, how to use them, etc. I'm in favour of them being used honestly as FPTP is designed to create two leading candidates and a bunch of also rans. Replace the voting system, the bar charts disappear (doubtless to be replaced by some other dubious graphic or other).

I dislike them being used dishonestly, but by the standard of the ones I've seen over time, that one's actually remarkably honest. Really.

Scary, isn't it?

Heh, the Tories during the Ealing Southall campaign (I think it was them) put out a chart showing how many Cllrs each party had, it was the best they could come up with showing them as having a chance...

[identity profile] rhythmaning.livejournal.com 2009-06-02 10:13 pm (UTC)(link)
I too would have abbreviated the chart at 5,000 - or 9,000: it would have the same effect (though it would also make the difference between Labour and LibDems look bigger, too).

I have no problem with using data - but it has to be used honestly.

[identity profile] rosathome.livejournal.com 2009-06-02 05:15 pm (UTC)(link)
Even the one on the website is fairly rubbish because of the variation in widths, which gives the impression that this is a histogram rather than bar chart, again with the effect of reducing the size of the vote that went to the Conservative candidate.

[identity profile] rhythmaning.livejournal.com 2009-06-02 10:16 pm (UTC)(link)
And I just thought they were trying to fit the numbers in!

[identity profile] nimoloth.livejournal.com 2009-06-02 05:56 pm (UTC)(link)
We got exactly the same thing from them here - a very distorted graph showing the Conservatives and SNP much smaller than Labour and Lib Dem, disproportionately so compared to the actual numbers. It was about the first thing I noticed about the flyer. It annoyed me, but I'll still vote for them - the lesser of four evils!

[identity profile] missedith01.livejournal.com 2009-06-02 07:40 pm (UTC)(link)
Shropshire reporting. Bar chart of Lib Dem candidate in European elections beyond reproach: percentage vote Lab/Lib Dem/Con 21:28:42, length of columns in mm 31:42:63.

She's picked the only recent election that had Labour in 3rd place and so her point is completely bogus, but her bar chart is accurate.

[identity profile] kittenexploring.livejournal.com 2009-06-03 07:16 am (UTC)(link)
Now that's proper misrepresentation. If you're going to fudge your conclusion then use carefully selected real data to back it up!

[identity profile] burkesworks.livejournal.com 2009-06-03 07:21 pm (UTC)(link)
Now that Rennard has gone home to Eastbourne and Vauxhall to spend more time with his money, I am hoping that I never have to see another bar chart on any of our campaign literature ever again.

That said, the top honours fall to the Labourites when it comes to the master-class in real misrepresentation.

Have added you, partly because you're another LibDem, partly because you know some of my London friends (were you at the Old Mitre a couple of weeks ago?) and partly because you are using a Thelonious Monk title as your handle.

[identity profile] rhythmaning.livejournal.com 2009-06-03 07:42 pm (UTC)(link)
Jeez, that borders on the criminal - kind of thing one might expect from a lunatic fringe party, but not Labour!

[livejournal.com profile] clanwilliam and I were talking about you a while back - well, we were talking about jazz. When she mentioned your user name, I thought she meant Birk's Works (http://www.google.co.uk/search?hl=en&q=birks+works&btnG=Google+Search&meta=&aq=f&oq=). Any relation?!

[identity profile] burkesworks.livejournal.com 2009-06-03 11:03 pm (UTC)(link)
No relation to the Diz, but my user name/e-mail address/former domain name was indeed inspired by said track.