Statistics make good lies

Insurrectionist · Post by **Insurrectionist** » Wed Sep 02, 2009 4:06 am

and good liars use statistics to manipulate their point of view.

DarkHorse · Post by **DarkHorse** » Wed Sep 02, 2009 4:52 am

Fundamentally ridiculous.

You can mislead by intentionally misinterpreting statistics, but unless the statistics are gathered with intentional bias as well, that's saying facts are wrong.

Spidey · Post by **Spidey** » Wed Sep 02, 2009 7:51 am

I agree somewhat.

Sorry I didn’t see that option, so I didn’t vote.

CUDA · Post by **CUDA** » Wed Sep 02, 2009 8:19 am

as the old saying goes. \"there are Lies, Damned Lies, and Statistics\"

Foil · Post by **Foil** » Wed Sep 02, 2009 8:33 am

DarkHorse is completely right. Statistics are by definition fundamentally unbiased.

The problems arise from lack of quality data and/or poor (often intentionally misleading) statistical inference.

A book on this topic I highly recommend is A Mathematician Reads the Newspaper. Despite the title, it's not a math book; it's written for a non-mathematical audience, and it's a good read.

Pandora · Post by **Pandora** » Wed Sep 02, 2009 8:36 am

I agree with Foil and Darkhorse.
“It is easy to lie with statistics, but it is easier to lie without them!”
-- Frederick Mosteller

Gooberman · Post by **Gooberman** » Wed Sep 02, 2009 10:08 am

I think a more fitting question, is whether or not it is possible to have an accurate statistical model of any of these complex issues people like to discuss. Correlation does not equal causation, and that is the danger of all statistics.

The classic example:
People are more likely to murder on a hot day.
People are more likely to eat ice cream on a hot day.

So when ice cream sales increase, so do murder rates. Therefore, in order to save lives, we should all stop eating ice cream!

Duper · Post by **Duper** » Wed Sep 02, 2009 10:14 am

lol

Fundamentally incorrect? Not in conjunction or relating to the question posed.

What do you think Push polls are all about??

Grendel · Post by **Grendel** » Wed Sep 02, 2009 11:52 am

\"Never trust any ststistics that you didn't fake yourself.\" German proverb.

Will Robinson · Post by **Will Robinson** » Wed Sep 02, 2009 3:31 pm

Gooberman wrote:....So when ice cream sales increase, so do murder rates. Therefore, in order to save lives, we should all stop eating ice cream!

You will never make it happen because 90% of Ice Cream vendors are bigtime campaign contributors

CUDA · Post by **CUDA** » Wed Sep 02, 2009 3:52 pm

Will Robinson wrote:
Gooberman wrote:....So when ice cream sales increase, so do murder rates. Therefore, in order to save lives, we should all stop eating ice cream!
You will never make it happen because 90% of Ice Cream vendors are bigtime campaign contributors

ya but what happens when they take over the whole Ice cream Dealership industry and start canceling the franchises of all the Republican Ice cream distributors

TechPro · Post by **TechPro** » Wed Sep 02, 2009 8:24 pm

Yeah, if you outlaw the ice cream, only outlaws will have ice cream.

Behemoth · Post by **Behemoth** » Wed Sep 02, 2009 9:12 pm

I voted no, But i feel the need to point out that misinterpretation from statistics can be used to send out fraudulent messages.

Goob's post is a good example of the danger of confusing trends by backing/mixing them up with statistics.

And/or the danger of preassumptions connected by misinterpretation of the statistics.

Behemoth · Post by **Behemoth** » Wed Sep 02, 2009 9:12 pm

I voted no, But i feel the need to point out that misinterpretation from statistics can be used to send out fraudulent messages.

Goob's post is a good example of the danger of confusing trends by backing/mixing them up with statistics.

And/or the danger of preassumptions connected by misinterpretation of the statistics.

Pandora · Post by **Pandora** » Thu Sep 03, 2009 4:06 am

Gooberman wrote:Correlation does not equal causation, and that is the danger of all statistics.

ah, but the fault in your example lies not in the statistics, but in a poor scientific strategy.
Your example is involves simple observation of events happening in everyday life. To prove causation you need experimental designs, that is, you have to manipulate your factor of interest.

Suppose you manipulate the availability of ice cream from day to day on a random basis, and you find that murder rates double on days that ice cream is available. You still have a correlation between ice cream and murder rates. But now you can be reasonable sure that there is something in the ice cream that is causing this.

snoopy · Post by **snoopy** » Thu Sep 03, 2009 10:38 am

Insurrectionist wrote:Statistics make good lies and good liars use statistics to manipulate their point of view.

Basically, the quote is lacking in that the term "Statistics" is not qualified.

"Statistics" as in the number process that says that 1 in 4 is equivalent to 25%, and the number process by which an average, std deviation, etc. are calculated cannot be lies when properly followed.

"Statistics" as in the way that a person gets from "I wonder how many people are happy with how the president is doing" to "the president has a 50% approval rating" involves a lot more steps than the mathematical process, and many of them are subject to bias and/or sample issues.

Thus, the large majority of the statistics that we deal with on a day to day basis aren't really all the accurate, because we deal with samples that can't be 100% random, tests that are framed to encourage certain results (many times unknowingly), and result interpretations that are looking for certain pre-conceived conclusions.

I'd word it this way: Good liars know how to frame subjective information as if it is an objective truth. Statistics are a good way to make your subjective take on something appear to have more objectivity than it really does, by nature of our mis-conceived idea that statistical gathering & interpretation is a purely objective science.

dissent · Post by **dissent** » Thu Sep 03, 2009 8:30 pm

from here

Greg Mankiw wrote:Suppose we were to graph average SAT scores by the number of bathrooms a student has in his or her family home. That curve would also likely slope upward. (After all, people with more money buy larger homes with more bathrooms.) But it would be a mistake to conclude that installing an extra toilet raises yours kids' SAT scores.

Foil · Post by **Foil** » Fri Sep 04, 2009 10:39 am

Gooberman wrote:So when ice cream sales increase, so do murder rates. Therefore, in order to save lives, we should all stop eating ice cream!

An equivalent mistake would be to assume that the murder :: ice cream correlation means the opposite causation, that murder makes people eat ice cream.

Baskin Robbins is supported by murderers, OMG!

Behemoth · Post by **Behemoth** » Fri Sep 04, 2009 5:02 pm

Wal-mart sponsored terrorism, heh.

Carrying that mentality could spit out quite a few different causality scenarios Foil.

One could say that Universal effect (A) on another universal effect receiving stimulus (B) creates the same given action each time (C)
A x B = C

The problem would be that life isn't predictable enough to sum up in such small measurements.

One could also say that the specifics for a given action are a bi-product of the nature of the universal effect on the recieving end:
(Example: It's a hot day so Tommy is in a bad mood, Someone hits Tommy's Mother with a car and drives off.. Tommy grabs a gun and follows the hit-run driver to a destination and shoots the driver.)
.. The situation was provoked into a more intense perception of what happened by his bad mood, which could've been alieviated by eating ice cream.

Even that situation would be too simple for it to produce such a dramatic sum at the end, But this stuff happens everyday.

Spidey · Post by **Spidey** » Fri Sep 04, 2009 5:09 pm

Ice Cream, always used to put my wife in a good mood.

Insurrectionist · Post by **Insurrectionist** » Fri Sep 04, 2009 5:55 pm

So Statistically speaking if your doctors name is Kervorkian you have a statistically higher chance of dying from his care.

CUDA · Post by **CUDA** » Fri Sep 04, 2009 6:12 pm

statistically speaking.

but only if you had Icecream for desert

Statistics make good lies

Do you agree with this statement?

Statistics make good lies

Re:

Re:

Re:

Re: Statistics make good lies

Re: