Statistics: How can you *possibly* know that?

[HYHYBT]

Something that drives me up the wall, and this seems like a good place to ask.

Often, you will see claims like "x% of rapes are never reported" or "y% of people with Disease Z don't know they have it." That sort of thing. How can you *possibly* know that for certain? If the rapes are unreported, then where are you getting the numbers from? If people don't know they have a disease, how do *you* know they have it? (So far as I can see, numbers of diagnoses don't cut it for that purpose. Not, at least, without knowing *how long* the person has had it and whether those who do and do not have it are equally likely to be checked.)

[Lachrymose]

23% of statistics are made up on the spot.

[bara]

87% of people polled admitted lying when polled.
13% of those polled are believed to be lying.

[Greenday]

Often, you will see claims like "x% of rapes are never reported" or "y% of people with Disease Z don't know they have it." That sort of thing. How can you *possibly* know that for certain?

Polls, surveys, that sort of stuff.

[Giggle Goose]

It's just the best educated guess that the "experts" can make. There's all sorts of quantitative measures you can use in statistics that are beyond the scope of my patience/intelligence to learn.

[BlaqueKatt]

Often, you will see claims like "x% of rapes are never reported" or "y% of people with Disease Z don't know they have it." That sort of thing. How can you *possibly* know that for certain? If the rapes are unreported, then where are you getting the numbers from?

data extrapolation-
for the rape statistic:

The U.S. Department of Justice’s National Crime Victimization Survey (NCVS) estimates the number of unreported serious crimes nationwide. the NCVS includes crimes reported and not reported to law enforcement; Also, the BJS derives NCVS estimates from interviewing a sample and are, therefore, subject to a margin of error.

if I have a 4 groups of 10,000 people and an average of 1000 per group hate the color green, I can safely estimate from that sample that out of 100,000 people 10,000 will hate the color green. Its averages based on samples of the population, or simply-statictics....

[Gravekeeper]

If the rapes are unreported, then where are you getting the numbers from?

Unreported means not reported to the police. Answered on an poll that promises anonymity would be a different matter for example. From there it's a matter of averages.

If people don't know they have a disease, how do *you* know they have it? (So far as I can see, numbers of diagnoses don't cut it for that purpose.

I can see two methods here:

1) How many were diagnosed with it after a lengthy period of time ( Having the symptoms for months without thinking it was serious enough to go to a doctor ) How many were hospitalized or died from the disease without knowing they had it. Thus being diagnosed after the fact.

2) Having a contagious disease and being unaware of it until they infect someone else who does get diagnosed leading the trail back to them.

[gremcint]

Statistically speaking everyone has one testicle and one ovary.

[jackfaire]

I make up statistics when I am trying to talk someone into something.


I am very good at talking people into things.

[Sleepwalker]

When you dunnae want to take the time to educate yourself on something, by all means recite tired jokes about it.

[Peppergirl]

When you dunnae want to take the time to educate yourself on something, by all means recite tired jokes about it.

Maybe MY sense of humor is off tonight, but I don't get it.

The 'dont want to take time to educate' yourself, I get. But not in relation to the 'tired jokes' aspect.

[jackfaire]

When you dunnae want to take the time to educate yourself on something, by all means recite tired jokes about it.

If your referring to me I wasn't making a joke. I really do make up statistics when trying to talk people into things.

Most people don't take the time to understand something like statistics so it's harder for the average person to tell when your full of it.

Not to mention repetition helps. The more someone has heard "according to the Statistics Group X is true" It doesn't matter how many times they are wrong because it's rarely pointed out to the average person that they were and so the average person gives that group credibility because obviously if they are always being quoted they must have a good track record.

[Sleepwalker]

Maybe MY sense of humor is off tonight, but I don't get it.

The 'dont want to take time to educate' yourself, I get. But not in relation to the 'tired jokes' aspect.

gremcint's crack about ovaries. I think the thing that bugs me most about statistics illiteracy is that studies, at least legit ones, have their methodology fucking SPELT OUT in plain text!

[HYHYBT]

Not to mention repetition helps. The more someone has heard "according to the Statistics Group X is true" It doesn't matter how many times they are wrong because it's rarely pointed out to the average person that they were and so the average person gives that group credibility because obviously if they are always being quoted they must have a good track record.

True enough. For example, FRC (Family Research Council, of which even at least two of the three words in their name are deceptive) puts out statistics all the time. There source... receives very little sunlight. And yet I keep running into what turns out to be their stuff.