I need some cute and amusing examples of real life bad graphs, charts or tables that are misleading (whether deliberately to sell something or just because people are stupid) or just plain wrong. Or even just stupid sentances in news articles like "53% of people die". I am helping give a talk at work about statistics and how to communicate it clearly without misleading people. My google foo is really weak - I just can't seem to hit useful keywords (I'm sure the internet must delight in making pages of things like this). If you find something that does work, please let me know how you found it, so in future I can do it myself instead of bothering you!
Ben wrote an article for a schools publication which was published a couple of months back - send him an email and I'm pretty sure he'd be able to send you the article, which has a few such examples in.
|Date:||January 22nd, 2008 02:55 pm (UTC)|| |
The back of a Head and SHoulders bottle is rather useful for this sort of thing :)
|Date:||January 22nd, 2008 02:59 pm (UTC)|| |
|Date:||January 22nd, 2008 03:00 pm (UTC)|| |
If you can get your hands on a copy of Edward Tufte's The Visual Display of Quantitative Information, it has some beautiful examples.
Yes, Tufte is great for examples of bad (and good!) presentation.
has some fake graphs some of which demonstrate how you can make your result appear different/more significant depending on your scale and start point etc.
And a point about pie charts - missing off one of the options makes the others appear more important etc....
not real graphs though...
|Date:||January 22nd, 2008 03:01 pm (UTC)|| |
|Date:||January 22nd, 2008 03:02 pm (UTC)|| |
what a horrid url.. i have a copy, but i'm unlikely to be within a lending distance any time soon :(
|Date:||January 22nd, 2008 03:03 pm (UTC)|| |
|Date:||January 22nd, 2008 03:10 pm (UTC)|| |
I'd instinctively google "misleading statistics powerpoint". Which returns some useful looking links, the first of which (mtsu32.mtsu.edu:11235/Misleading%20Statistics.ppt) seems like it would be the right thing based on the 'view as html' cache, but doesn't actually exist.
But I should be doing work...
I just downloaded it - it worked ok
|Date:||January 22nd, 2008 03:23 pm (UTC)|| |
is full of this stuff.
My favourite "spurious graph", though, is fictional: it's this one
, from the marvellous Gettysburg Powerpoint Presentation.
|Date:||January 22nd, 2008 04:30 pm (UTC)|| |
I have junkcharts
on my fl, but I don't actually read anything it says.
As others have posted the two places I'd look for really bad charts (Badscience and junkcharts - although a third would be the rise and fall of the FTSE graphs that are continually presented with no baseline), I'm going to give one example of a really good chart as a counterpoint - that of Napoleon's retreat from Moscow
For a bit of grandstanding, I'd buy a copy of the Daily Wail the day before you give the talk and grab half a dozen case studies from there.
For humourously bad use of statistics I'd suggest almost all of econometrics ever. My personal favourite is the proof that countries converge to similar GDP per capita by picking a group of countries which are all rich now and showing that their GDP per capita converged over the last 150 years.
Ah. Oh. I see a lot of econometrics in my future. Is it all tosh?
The only thing which immediately occurs having a look at is strange_maps
, which has a certain amount of statistical info. Even if you don't find what you're looking for, it's very entertaining.
|Date:||January 22nd, 2008 05:11 pm (UTC)|| |
oh for heaven's sake, I handed in my coursework on good/bad graphs 3 hours ago - couldn't you have posted this a few days ago? :P
unfortunately all my bad graphs/tables were not on the interwebs, so I'm not much use, but I'm still going to grumble about your poor timing!
|Date:||January 22nd, 2008 05:21 pm (UTC)|| |
Well, you could always email it to me if you have it in electronic form ;-) atreic at livejournal d0t c0m reaches me.