Do you REALLY know what % increase means?

[Aceman]

You know, like when they say "People who drink more than two cups of coffee per day are 32% more likely to develop a heart condition?

(by the way - see Media Sucks - they could really help clear this up!)

It means that if you take the base rate for an event, and multiply it times 1.32, that's the new chance of something happening.

So for example:

If you have a 1 in 1000 chance of developing some bizarre illness, and you do something that increases that 32%, hat is the REAL increase in danger?

Well, let's take a look.

1/1000 = .001% chance

1.32 x .001 = .00132

or 1 and 1/3 in a thousand. Woo Hoo.

Take home message, know your base rates and do the math before you panic. And on top of that, those are "overall rates" there are almost always people with greater than 1/1000 and people with much less.

For example, that 1/1000 includes some people clearly at a 1 in 10 chance, and some that are 1 in 100,000 chance. (even going from 1 in ten to 1.3 in ten isn't that brutal...).

But people who have championed whatever it is they think YOU should be afraid of don't bother to tell you that do they?

[karpathion]

Is this some of that fury stuff you where talking about this morning?

[ParameterMan]

Thank you.

Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to Disraeli would often apply with justice and force: "There are three kinds of lies: lies, damned lies and statistics."

I have a bad habit of talking back to the TV and the things I say aren't always complementary.

[Reload in 5]

I have a bad habit of talking back to the TV and the things I say aren't always complementary.

I thought I was the only one.
I am especially bad watching football. The announcers often don't really understand the rules, at least not the most common ones, and say stupid stuff. Sometimes I watch with the sound off.
True Story
they use a "football" phrase like
"the ball hit him right in the breadbasket"
When in fact it hit his outstreched hands. That is NOT the breadbasket. Immediately after that play they went to commercial and showed a slow mo of the same play, and the announcer says "see, right in the breadbasket"

[ParameterMan]

Just read a good article along these lines.

[...]the idea of conditioning on a collider. If a case being observed is a function of two variables then this will induce an artifactual negative correlation between the variables. This is true even if in the broader population there is no correlation (or even a mild positive correlation) between the variables.

[Powdered Toast Man]

I took Statistics in university. We learned that you can manipulate stats to say anything you want them to. It's all about interpretation of the data and the manner in which it is presented. One of the first rules of stats is that correlation does not mean causation, yet it happens all the time.

This is just made up, but as an example you could take data like this:
"68% of teens who drink alcohol will also smoke tobacco."
Most people would say, "Well, obviously underage drinking is linked to underage smoking!"
However, the data is misleading. Does drinking cause smoking or is there another factor at work? Could it be that teens who engage in underage drinking are already more predisposed to risky behavior, which would mean they might have tried smoking regardless of whether they drank alcohol?

See, that's where you have to be careful with Statistics and think critically when they are reported by a third party - like the media. Stats are just data - they show trends, patterns, etc, but they don't actually explain the reasons behind things.

[ParameterMan]

Yes, and the best place to skew the stats is actually in the data-gathering process. I've thought about this quite a bit when a survey or study of people in the midwest will come out that reflects values that I know as a resident to be less prevalent than they suggest.

I know they didn't ask me, or anyone I know. I also know that the majority of people with that value system are likely to turn down participation in a survey or study, especially if they've tried it before. Why? Because the questions are always extremely leading. Especially the surveys - they're carefully engineered to divide the participants into two strong groups: intelligent progressives and stupid, violent backwater hillbillies.

I'm trying to remember to ask this question whenever I see statistics: "They have data, but where and how did they get it?"

[GuitarStv]

If anyone sees the first post in this thread as a revelation, it's very important that you do the following immediately:

- Release your keyboard and mouse.
- Stand up from your chair/desk.
- Locate the nearest fork.
- Repeatedly stab yourself into the testicular area.

This simple eugenic solution will help to prevent the degradation of the human race by tossing a little chlorine into the gene pool and getting rid of some of the pond scum.

[Rand-O-Monium]

Dude,your "fury" sucks...

Nothing "personal",but WTF?

[Funkfingers]

Ace. I was going to regurgitate my usual Frank Zappa lyric quotation - "We are dumb all over and maybe even a little ugly on the side." - but you appear to have induced plenty of regurgitation already. Keep up the good work.

[Jessie's ghost]

I've heard red two-seaters are more likely to be involved in fatal accidents than other cars. Stack the odds in your favor by trading it in on a white family sedan.