How to lie with statistics

[Mecha Wolf]

I've come to realise how easily manipulated people in general are, so I'm putting together this as a public service (oh I thought people were gullible before, but I was truly ignorant to how the world works). No need to thank me. Really, I mean it.
There are different categories that I've arbitrarily named, so as to make it easier for you to spot these lies when seen in the wild and untamed wilderness of the untamed and wild media. The first thing to make sure is that the figure is derived from an actual scientific study, but even if it is you should look out.

<b>The correlation lie</b>
If you hear someone saying that about 4% of people who get liver cancer drink artificially flavoured chicken soup, make sure that 4% isn't also the amount of people who drink artificially flavoured chicken soup in general. If you see any statistic about something that doesn't mention the same statistic for the general population, it's likely to be skewed.

<b>The Michael Moore lie</b>
Stating how many people died because of gun violence in different countries proves absolutely nothing unless you mention the population. I can safely say that there are fewer people who eat food in Greenland than there are in Canada. It means nothing. I know that's a popular slam against Moore though, so hopefully people are catching on (and hopefully they'll see what's accurate criticism and what's just more bullshit from the right wing pundits, but that's another story).

<b>The omission of percentage lie (pretty much the same as the previous lie, but more frequently used and accepted)</b>
If you come across a sentence saying that you're 3 times more likely to get AIDS if you own a leather thong, make sure to check how much different the actual percentage is. If that doesn't make sense to you, remember that 0.000003 is three times as much as 0.000001 -- a difference that's quite easy to dispute.

<b>The Fox News/CNN/New York Times website said it so it must be true lie</b>
If you see a figure from a non-scientific study, it's always bullshit. The result depends on the size of the group involved, location, etc. Add to that that only their fanbase is represented, so it's fairly clear that it's biased from the start (Would you accept a study done by High Times on how many people are for legalising marijuana? No? Really?).
Also of interest is that averages are irrelevant in small numbers, you need a very large group of people from different backgrounds to accurately represent anything.

I believe most people base a majority of their beliefs on lies and bullshit statistics, but I have no figure to support that.

If you find something inaccurate or misleading in this post, post a correction below and tell me why I'm wrong and why you're right.

Hail Satan! :)

[trög]

What I don't get is why Michael Moore can't stop being such a douchebag these days. He used to be cool, and then he made Fahrenheit 9/11. You don't need to mess with statistics to prove that Bush has done a bad job as president, but he did it anyway.

[Shiv]

Manipulating statistics is hilarious in my opinion. In school I had a teacher who said outright, 'you're less likely to commit suicide if you believe in God'... whenever we challenged her she just shrugged and told us she wasn't 'suggesting' anything and that that's just what the statistics said. Luckily enough of us had minds of our own and realised early on that everything she said was full of crap.

I'm not sure what to call it but I think I have another fairly common statistical lie. Like statistics showing that autism is on the rise when really it's just that autism is becoming more and more likely to be diagnosed. It's more about mixing up causalities and correlations. Using correct statistics to come to inaccurate conclusions... you get the idea.

[Mecha Wolf]

I'm not sure what to call it but I think I have another fairly common statistical lie. Like statistics showing that autism is on the rise when really it's just that autism is becoming more and more likely to be diagnosed. It's more about mixing up causalities and correlations. Using correct statistics to come to inaccurate conclusions... you get the idea.
Yeah, that's very true. It's also used by end of the world "prophets" claiming earthquakes are becoming more and more frequent.


What I don't get is why Michael Moore can't stop being such a douchebag these days. He used to be cool, and then he made Fahrenheit 9/11. You don't need to mess with statistics to prove that Bush has done a bad job as president, but he did it anyway.
I still like Michael Moore's movies, I just think he's become more of a partisan hack after 9/11, somewhat like Bill Maher, Christopher Hitchens, Dennis Miller etc. etc.
Netcraft confirms it, libertarianism in the US is dying. Thankfully there's still P&T, so there's still some hope for you. Oh, and that constitution thing rocks, I wish Sweden had one of those.

[Davey Rootbeer]

WELCOME TO PUBLIC RELATIONS 101.

we made a whole INDUSTRY based on choosing to reveal only certain parts of statistical data which caters to our cause, therefore not explicitly outright lying, but not giving the public the full story, while still being able to pass it off as the truth.


I can't throw a pebble without hitting a case study here. Need i say more?

[when you are the moon]

http://www.fallacyfiles.org/

[Liz]

I get confused over the one on food products, "1/2 less calories than product B!"
Does that mean you take product B, divide by two, and subtract that number (aka cut product B in half caloriewise) or does it mean take product A, divide by two, and add that number to make product B?
I think the two differences are.... different.

I also hate it when they say something like, "Statistics show that people who excercize regularly are less likely to be depressed then those that do not excercize regularly". Other than other weaknesses, the one I would like to point out is that this statement suggests that excercizing relieves depression (and that not excercizing causes it), when in reality, it is just as likely that in any given case a depressed person does not excercize because they are, well, depressed. So a statement can quote a statistic and be biased to indicate one of the things caused the other, when it might be just as possible the reverse.

[trög]

I still like Michael Moore's movies, I just think he's become more of a partisan hack after 9/11, somewhat like Bill Maher, Christopher Hitchens, Dennis Miller etc. etc. Roger and Me is still awesome, I just don't like him very much right now.

But yeah. I agree with that other stuff you said.

[implode]

in any case, any statistic that lies for the sake of something universally good (like exercising) doesn't really need to be shat upon for being misleading. it's not something i personally would do, but if it works, hey, whatever. treadmill your problems away to your hearts content.

[MST3Kakalina]

there are three kinds of lies:

lies, damned lies, and statistics.


and my brother just bought a book called "How to Lie With Statistics," funnily enough.


in conclusion, people are stupid gullible fucks.

[Forever Finite]

in co-conclusion, people are manipulative little triptychs.

[MST3Kakalina]

how about:

in conclusion, people suck?

[Forever Finite]

or how about, "hey rachel you just used the name of a type of art peice as an insult. what's up with that?"

[MST3Kakalina]

hey rachel you just used the name of a type of art peice as an insult. what's up with that?

[Shiv]

Statistics show that people who excercize regularly are less likely to be depressed then those that do not excercize regularly". That's a pretty crappy example. Excercising not only makes you feel good cause your healthy, it also releases happy hormones. Still, that example would fall under the misinterpritation of correlation and causation. Like 'it's more likely to be sunny when people are wearing shorts', or ' you're more likely to be eaten by a shark on the week end'. How many clever peoples can see the flaws in those arguments -_-'?

[when you are the moon]

Well, there's also the thought that, "hey, maybe a segment of the people who don't exercise regularly are suffering from a debilitating illness which is more directly responsible for their depression than not exercising?"

It's all about the sample population.

[Forever Finite]

hey rachel you just used the name of a type of art peice as an insult. what's up with that?
good girl! you're such a good dog!
koba, you are my favorite silky terrier in the whole wide world. yes you are!