Popular Ignorance of Statistics

bnt

Maybe I've been on Boards too long, but I wish people would understand some basic things about Statistics. In many threads here, and in the media, you can see a lack of understanding every day. People who don't get what Statistics can and can't do could at least start with the following:

- Statistics works with probabilities, not certainties.
- Statistics works with populations, not individuals.
- Statistics describe reality without judgement.

For example: the average adult male height in Ireland is 177.5 cm*. What do we learn from that?
- "are all Irish men that tall?" No, obviously.
- "Is there an "average man" who is that height?" The media seems to like this one, but it's irrelevant.
- If you are much shorter than that, is there something wrong with you? Possibly, but irrelevant, The statistics just describe how things are, not how they should be. No judgment.

Now go to the Netherlands, where the average adult male height is 183.8 cm*, among the tallest in the world at the moment.
- "are all Dutch men taller than all Irish men?" No.
- "are Dutch men taller on average?" Yes.
- "Only 6cm? Have you seen their football team?" The football team isn't a representative sample: they've been selected for fitness factors which may include height.
- "but ... but ... I know a Dutch man, and I'm taller!" Doesn't matter. The statistics describe the population, not individuals.
- "but that contradicts the statistics!" No, it doesn't. Both populations naturally include people who are much taller and shorter than the average.

This is something I see here a lot. Posters thinking that, because they've found something that doesn't match the statistics, the statistics are wrong. Nope. Sure, statistics can be wrong, and they definitely can be misused, but the existence of "exceptions" doesn't invalidate them at all. The population of Ireland includes some very tall men, and they are included in the statistics, but you're still shorter than Dutch men on average.

* source: Wikipedia

(This was originally for the "Trivial Things That Annoy You" thread, but then I realised it's not Trivial. If I put this in the Maths forum, they'd be going "well, duh". )

Birneybau

87.36584% of statistics are made up.

username123

bnt wrote: »

Maybe I've been on Boards too long, but I wish people would understand some basic things about Statistics. In many threads here..................

.........This is something I see here a lot. Posters thinking that, because they've found something that doesn't match the statistics, the statistics are wrong. Nope.

Examples of threads or posts where you've seen this?

Donkey Oaty

bnt wrote: »

If I put this in the Maths forum, they'd be going "well, duh". )

As, I suspect, will many in AH.

For Forks Sake

Birneybau wrote: »

87.36584% of statistics are made up.

74.78% of people knew that already

But if people understood that they might have to drop their self rightness and smug certainty about every thing and use their brains.

wazky

For Forks Sake wrote: »

74.78% of people knew that already

The other 47 % struggle with basic maths.

jamesbere

So I'm taller then the average Dutch person, well ain't that something

keith16

Data / stats can't tell you everything, but it can provide massive insights.

bnt

username123 wrote: »

Examples of threads or posts where you've seen this?

I'm posting in AH, not writing a paper on the topic.

gramar

jamesbere wrote: »

So I'm taller then the average Dutch person, well ain't that something

You've just broke statistics.

Duggy747

This thread needs 34% more Alf Stewart.

lazygal

I'm 100% sure that 99% of posters don't care.

I do like to annoy my mathsy husband with phrases like 'giving 110%' though.

username123

bnt wrote: »

I'm posting in AH, not writing a paper on the topic.

100% sure you're just talking ****e then.

Awkward Badger

bnt wrote: »

In many threads here, and in the media, you can see a lack of understanding every day.

In fairness in many threads and in the media you can usually see a lack of understanding or misuse of pretty much anything that can be misused or misunderstood to enable someone to make their point or sell a copy.

osarusan

bnt wrote: »

- "Only 6cm? Have you seen their football team?" The football team isn't a representative sample: they've been selected for fitness factors which may

Height is a factor in fitness?

TheBody

I always laugh at those shampoo adverts that say stuff at the bottom of the screen like "95% of 60 people agree".

maggiepip

Heres an example of something that gets sensationalized: you read that doing "x" increases your chance of cancer by 60% -shock horror - but if the normal chances of cancer were .5% then an increase of 60% would only bring your increased chance of cancer up to .8%. Sometimes I dont think people get that properly.

osarusan

maggiepip wrote: »

Heres an example of something that gets sensationalized: you read that doing "x" increases your chance of cancer by 60% -shock horror - but if the normal chances of cancer were .5% then an increase of 60% would only bring your increased chance of cancer up to .8%. Sometimes I dont think people get that properly.

I'm pretty sure the people telling you these statistics are banking on people not getting them properly - to hawk whatever product/diet/etc they are selling.

mickydoomsux

Irish people hate stats and figures because they ruin the down the pub anecdotes by proving them to be complete bull****.

bnt

osarusan wrote: »

Height is a factor in fitness?

That's up to the coach to decide. If he's selecting for the ability to flatten incoming strikers the way Jaap Stam used to do, then sure. My point is that the team is not a representative sample of Dutch men.

bnt

Similar to what maggiepip pointed out, I've seen stats used to justify male circumcision. Apparently, having a foreskin slightly increases the risk of getting cancer of the penis ... the statistical risk is increased from infinitesimal to minuscule. :rolleyes:

RDM_83 again

My personal pet peeve is the way averages are always used to describe the "normal", its fine for stuff like height* but take number of sexual partners as an example i have seen (in an absolutely terrible in other ways womans magazine article). Wages is another time its misused.

you have 9 people with numbers varying between 0-10 but if you have one person with a 100 partners you have a really skewed view if its an average used.

*because its a continuous scale with no one at 0 and few very high or low value outliers

PizzamanIRL

You're wrong. I'm not ignorant toward statistics.

A Neurotic

I was reading some interesting stuff about how wobbly healthcare professionals' grasp of statistics can be yesterday:

http://www.bbc.com/news/magazine-28166019

http://www.vox.com/2014/6/24/5837832/is-america-better-than-europe-at-treating-cancer

shruikan2553

You can use statistics to prove anything you want. There's a graph of the decline of murder rates and decline of internet explorer usage. Correlation does not mean causation.

Franticfrank

Statistics are great! And they aren't always so boring...has anyone stopped by Statista? It's pretty cool.

bnt

PizzamanIRL wrote: »

You're wrong. I'm not ignorant toward statistics.

You know, I actually got that one ...

topper75

Awkward Badger wrote: »

In fairness in many threads and in the media you can usually see a lack of understanding or misuse of pretty much anything that can be misused or misunderstood to enable someone to make their point or sell a copy.

Yup - that is the only data/figures that matters to Joe Journo.

My favourites are damned lies. They are more exciting than ordinary lies but not so bad that they warp reality into a knot like statistics.

the culture of deference

try the birthday paradox

http://www.mathsisfun.com/data/probability-shared-birthday.html

http://en.wikipedia.org/wiki/Birthday_problem

jimgoose

I've been saying it for years - statistics, like shotguns, credit-cards and strong liquor, should be kept away from children, imbeciles, and people who spend weeks arguing back-and-forth with each other on epic five-hundred-page threads. :pac:

ytpe2r5bxkn0c1

Lies, damn lies and statistics.

solomafioso

bnt wrote: »

* source: Wikipedia

Now I know you're taking the p!ss! :pac:

Pherekydes

Google the Sally Clark case for the damage that ignorance of statistical methods can do.

Also, when ice cream sales go up so do drownings. Therefore, eating ice cream can cause you to drown!

OldGoat

- Statistics works with probabilities, not certainties.
- Statistics works with populations, not individuals.
- Statistics describe reality without judgement.

I am 100% certain (certainties) that
in my case (individual)
you are mistaken (judgement)

The Diabolical Monocle

People posting stats don't always seem able to differentiate average and median however.

Procasinator

The Diabolical Monocle wrote: »

People posting stats don't always seem able to differentiate average and median however.

You probably want to say the mean and the median. While usually when some uses average they are implying the mean, it could also refer to values like the median or the mode.

Mr.David

Very true.

Same thing applies to discussion regarding 'average income'.

Suppose you take a sample of 5 people, 4 of whom earn €30k/annum and the 5th earns €250k/annum. The average income of the group is €74k/annum despite the fact that 80% of the group earn much less than that.

People don't grasp the concept of skew. In this instance, the distribution of earnings has a lower bound of €0/annum but no upper bound and so upside outliers have a large distorting effect.

Aglomerado

Mr.David wrote: »

Very true.

Same thing applies to discussion regarding 'average income'.

Suppose you take a sample of 5 people, 4 of whom earn €30k/annum and the 5th earns €250k/annum. The average income of the group is €74k/annum despite the fact that 80% of the group earn much less than that.

People don't grasp the concept of skew. In this instance, the distribution of earnings has a lower bound of €0/annum but no upper bound and so upside outliers have a large distorting effect.

I wish people would take this into account when reading reports of public sector earnings. Sure, the 'mean' wage is x amount, but the 'median' is closer to reality.

Peter B

shruikan2553 wrote: »

You can use statistics to prove anything you want. There's a graph of the decline of murder rates and decline of internet explorer usage. Correlation does not mean causation.

Statistics are not proving anything here

Penny Tration

All I've gleaned from the OP is confirmation that I'm a hell of a lot shorter than average.

looking_around

lazygal wrote: »

I'm 100% sure that 99% of posters don't care.

I do like to annoy my mathsy husband with phrases like 'giving 110%' though.

That drives me nuts.

played a game that half way through said something like "120% finished". I stop playing the game XD.

dxhound2005

bnt wrote: »

Maybe I've been on Boards too long, but I wish people would understand some basic things about Statistics. In many threads here, and in the media, you can see a lack of understanding every day. People who don't get what Statistics can and can't do could at least start with the following:

- Statistics works with probabilities, not certainties.
- Statistics works with populations, not individuals.
- Statistics describe reality without judgement.

For example: the average adult male height in Ireland is 177.5 cm*. What do we learn from that?
- "are all Irish men that tall?" No, obviously.
- "Is there an "average man" who is that height?" The media seems to like this one, but it's irrelevant.
- If you are much shorter than that, is there something wrong with you? Possibly, but irrelevant, The statistics just describe how things are, not how they should be. No judgment.

Now go to the Netherlands, where the average adult male height is 183.8 cm*, among the tallest in the world at the moment.
- "are all Dutch men taller than all Irish men?" No.
- "are Dutch men taller on average?" Yes.
- "Only 6cm? Have you seen their football team?" The football team isn't a representative sample: they've been selected for fitness factors which may include height.
- "but ... but ... I know a Dutch man, and I'm taller!" Doesn't matter. The statistics describe the population, not individuals.
- "but that contradicts the statistics!" No, it doesn't. Both populations naturally include people who are much taller and shorter than the average.

This is something I see here a lot. Posters thinking that, because they've found something that doesn't match the statistics, the statistics are wrong. Nope. Sure, statistics can be wrong, and they definitely can be misused, but the existence of "exceptions" doesn't invalidate them at all. The population of Ireland includes some very tall men, and they are included in the statistics, but you're still shorter than Dutch men on average.

* source: Wikipedia

(This was originally for the "Trivial Things That Annoy You" thread, but then I realised it's not Trivial. If I put this in the Maths forum, they'd be going "well, duh". )

I don't agree. Some things are capable of being measured exactly e.g how many people over a certain age died in a particular year. We have a Central Statistics Office and I would imagine they think they are publishing reliable information.

http://www.cso.ie/en/index.html

Akrasia

TheBody wrote: »

I always laugh at those shampoo adverts that say stuff at the bottom of the screen like "95% of 60 people agree".

I too like to play 'hunt the disclaimer' on cosmetics and 'supplements' ads in tv and print although i feel like I'm gonna need a bigger higher definition tv to catch them as they're often printed in white text over an actress wearing white clothes while standing in the snow on a cloudy day.

regarding the OP, the human brain is absolutely rubbish at processing probabilities.

If I buy a lottery ticket, I have a 1 in 2 million chance of winning. Hmm, don't like those odds, so I'll buy two tickets, I've just doubled my chances to 1 in a million...
er wait, nope, now I only have a 2 in 2 million chance of winning, i've increased my chances of winning from .00000005% to 0.000001% Still doubled my chances though

If i ask you
Pick the person from this list who is most likely to describe a Mathematician
1 - Male - 40years old
2 - Male 40 years old - has a cat
3 - Male - 40 years old - wears glasses and/or has a crazy beard

A large percentage of the population will pick options 2 or 3 even though statistically, it has to be option 1 because options 2 and 3 are all subsets of option 1 and are by definition less probable

Ficheall

Akrasia wrote: »

If i ask you
Pick the person from this list who is most likely to describe a Mathematician
1 - Male - 40years old
2 - Male 40 years old - has a cat
3 - Male - 40 years old - wears glasses and/or has a crazy beard

A large percentage of the population will pick options 2 or 3 even though statistically, it has to be option 1 because options 2 and 3 are all subsets of option 1 and are by definition less probable

Perhaps if you phrased the question a little more clearly, it might help...

Arthur Beesley

I'm somewhat annoyed that you started this thread, OP. You are so smart, I wish I could be more like you. :rolleyes: No doubt Mensa will come knocking to head hunt you.

Arthur Beesley

bnt wrote: »

That's up to the coach to decide. If he's selecting for the ability to flatten incoming strikers the way Jaap Stam used to do, then sure. My point is that the team is not a representative sample of Dutch men.

you mean the Dutch don't all play football at a professional level? I'm glad you explained that to us.

Arthur Beesley

Aglomerado wrote: »

I wish people would take this into account when reading reports of public sector earnings. Sure, the 'mean' wage is x amount, but the 'median' is closer to reality.

Still overpaid though, since they don't mention the gold plated pensions etc.

Ficheall

I'm concerned by how long it's taking you to read this thread, Arthur...

Arthur Beesley

looking_around wrote: »

That drives me nuts.

played a game that half way through said something like "120% finished". I stop playing the game XD.

Well you were already finished, so why would you or indeed how could continue playing?

That'd be like continuing searching when you find something in the last place you looked...

Arthur Beesley

Ficheall wrote: »

I'm concerned by how long it's taking you to read this thread, Arthur...

Thanks for the concern, but I have more going on that just this thread. Plus I had to look up the meaning of XD.