Annoying statistical statements that make no sense.

[queenofthedump]

Mothercare have put a thing on fb today saying 'On average 80% of babies are born with some sort of birthmark'.
AIBU or does this make no sense? It's really not an average.

I hate the misuse of statistics! I once worked in a call centre where the management changed the target from:
'80% of calls to be answered within 15 seconds' (fair enough)
to
'100% of calls to be answered within an average of 15 seconds' WTF does that even mean??!

The mean average is also pretty useless without knowing variation as well. And sample size. Especially sample size.

Is that arithmetic mean, or geometric mean ?

Msvestibule surely the 2nd option was better- ie. Your average call answering time is 15 seconds. Sometimes you'll answer on the first ring, sometimes it will take 20 seconds, but as long as your overall average is 15 secs you're ok. In the first option, 80% of your calls must be answered within 15 seconds.

Unless I have totally misread that!!

I hate cosmetics adverts for they'd bollocks statistics- in fact, I did my dissertation on it many moons ago and nothing's changed since then. The best is things like 'pure Matt lipstick, 45% of women agree it gives long lasting Matt finish' - what, less than half of you? More people disagree than agree????

100% of calls to be answered within an average of 15 seconds

What does that mean?

We expect your average answer time to be 15 seconds. That's all they need to say.

And what does 'within an average' even mean?

I was given a task recently:

Group A had an average points score of 4.6
Group B had an average points score of 4.7

The task asked me to compare them and analyse what the results meant for Group A and Group B. I strongly suspect the people who the results were meant for weren't strong on statistics.

Would anyone care to analyse them?

Don't get me started on the year I did my personal appraisal and was asked 'How have you given 110% this year?'

I put a snarky response about how that was completely impossible . I'm still employed by them but they did take that question out...

This is my sort of thread :)
What I hate is when a perfectly valid survey is reported and people leap on a thread to say "but they only asked 3000 people, I asked dh who's a mathematician and he says no way can you trust anything with a sample size that small".
This is absolute bollocks. A well designed sample of 3000 is a brilliant representation. But they never will believe it!

Or people who do not understand risk.
I won't let my dc do brownies, ever since a child was killed at the other end of the country on a brownie trip. Why take the risk with the most precious things in your life...
They say as they drive to the supermarket.

Can I recommend More or Less for people?

Statistics debunked

www.bbc.co.uk/programmes/b006qshd/episodes/player

They say as they drive to the supermarket

Error bars... gotta love an error bar. And measures of statistical significance.

My pet bugbear is scientists who don't understand the difference between sample statistics and population statistics and regularly use measures of statistical significance (crudely - whether your sample was big enough for your results to be likely - at a certain level - to not be due to random chance) as if they told you something about whether the difference they're looking at is important at a population level. (C.f. more studies of sex difference in cognitive performance than you can shake a stick at, none of which ever bother to tell you Pearson's d-value as a measure of how much or not the two populations overlap).

My favourite ever statistics bollocks was an article from ages ago shocked by the fact that New York, with a population of only ten million people, had a murder rate of 0.2 per mille - higher than the entire continent of Europe!

My favourite ever statistics bollocks was an article from ages ago shocked by the fact that New York, with a population of only ten million people, had a murder rate of 0.2 per mille - higher than the entire continent of Europe!

Well, it is true.

Though I am sure there is a smaller place in South America that manages an even higher murder rate (per mille) on even less space. Why, one might have a shockingly high murder rate on even one square metre!

I remember hearing Tony Blair say seriously that he wanted 75% of students to achieve above average results. Wish the interviewer had had enough knowledge to follow that up with "you mean you want the bottom 25% to do so badly they bring the average down".

My personal dislike is talking about over 100%-he gave 110% of effort type statement. Totally meaningless.

110%of the effort you have last time would make sense

Or people who do not understand risk.
I won't let my dc do brownies, ever since a child was killed at the other end of the country on a brownie trip. Why take the risk with the most precious things in your life...
They say as they drive to the supermarket.

This is spot on - so common to hear this sort of thing these days.

Agree with Orlantina we should get Tim Harford on as a guest on MN because I doubt More or Less get their journalists scouring MN for stories.

I remember seeing an ad years ago for a cellulite-busting miracle cream and it said that 90-odd% of women have cellulite.

Which made me think that in that case, having a few dimples is a totally normal part of having a female body and it's the few who don't that are the unusual ones! I don't think that was what the company was trying to achieve!

I have more than the average number of legs.

Not being 'legist' just enjoying the statistic.

Also love correlation - the more churches there are in a town, the higher the burglar rate is a fairly standard one.

Burglary!