Property boom

[DaveB]

Indeed. In Clinical Trials, the requirements are stringent, for obvious reasons.

Ever noticed the wording in TV ad's for face creme's etc. the voice over claims 80% or whatever noticed results, the small print in the corner of the screen tells you the results are based on some ridiculously small sample. A clear example of statistics being used to claim factual results without disclosing the full details.

The point is that *anyone* can take some numbers relating to a subject and make them appear to support any proposition they might wish to make. Outside of closely controlled and peer reviewed research, in whatever field, statistics are largely meaningless to anyone but those who are using them.

[Bagpuss]

Ever noticed the wording in TV ad's for face creme's etc. the voice over claims 80% or whatever noticed results, the small print in the corner of the screen tells you the results are based on some ridiculously small sample. A clear example of statistics being used to claim factual results without disclosing the full details.

The point is that *anyone* can take some numbers relating to a subject and make them appear to support any proposition they might wish to make. Outside of closely controlled and peer reviewed research, in whatever field, statistics are largely meaningless to anyone but those who are using them.

Exactly but they have to be careful what they say, usually if they say "Not proven clinically" you know it is bollocks. One of the anti ageing face creams got done for that recently, and remember the benicol advert with Voderman "not proven clinically but this is real life" or something of that order.


Re the financial stuff Janey. I imagine you are talking of forecasting? Which is all crystal ball (best) guess work.

However, I'm certainly not sat here falsifying information.

[sasguru]

Ever noticed the wording in TV ad's for face creme's etc. the voice over claims 80% or whatever noticed results, the small print in the corner of the screen tells you the results are based on some ridiculously small sample. A clear example of statistics being used to claim factual results without disclosing the full details.

The point is that *anyone* can take some numbers relating to a subject and make them appear to support any proposition they might wish to make. Outside of closely controlled and peer reviewed research, in whatever field, statistics are largely meaningless to anyone but those who are using them.

1) Face creams are not subject to the rigours of Clinical trials. There are certain substances that have been cleared for safety for use in face creams and the different brands simply use different combinations of these substances.
2) Within reason, the size of a sample is not as important as its "representativeness" of the population. Obviously testing on 5 people, say, would be meaningless.
3) Yes, anyone can take some numbers etc., but a trained statistician (or even an intelligent layman with some knowledge of stats) can easily refute their claims. The problem is that most people do not want to bother to learn or can't learn the basic stats that enables them to differentiate dross from gold.

[expat]

on average 40% (or nearly half) of all "sick" days are taken on either a Monday or a Friday

Something funny here, I'd expect far more:

2/7 of 1-day illnesses will fall on Monday or Friday; but another 2/7 are on weekends so will not become absences: therefore 2/5 (or 40%) of visible 1-day absences will be Monday or Friday. That's just from 1-day illnesses.

But 2/7 of 2-day illnesses will be F+Sat or Sun+M (and only 6/7 of 2-day illnesses will be absences since Sat+Sun will be invisible) so another 2/6 (or 33%) of 2-day illnesses will manifest themselves as 1-day absences.....

I shall leave the completion of this to the student, who will doubtless learn in the process something about statistics and assumptions.

[threaded]

on average 40% (or nearly half) of all "sick" days are taken on either a Monday or a Friday

I like that one.

[janey]

Exactly but they have to be careful what they say, usually if they say "Not proven clinically" you know it is bollocks. One of the anti ageing face creams got done for that recently, and remember the benicol advert with Voderman "not proven clinically but this is real life" or something of that order.


Re the financial stuff Janey. I imagine you are talking of forecasting? Which is all crystal ball (best) guess work.

However, I'm certainly not sat here falsifying information.

nah not talking about forecasting stuff, more to do with past trends in financial markets etc... all quite a long time ago now so all rather dusty in the old brain cells.

I would guess (and hope!) that medical research is much more rigorous since it could be life and death situations but I would still say that in an awful lot of situations statistics are meaningless. I understand that you are saying that people have to be careful when giving these figures (after all they can be sued these days) but in the example I have given wouldn't the researcher be able to turn around and say "well here's the proof.. I did xyz calculation and it proves it"?
I understand from what you say that in medical research they have to declare that they used abc calculation that actually disproved their hypothesis but I'm not aware of any such scruples in the financial research I was involved with and I would guess that is similar to a lot of industries that use statistics... eg politics!

[janey]

1) Face creams are not subject to the rigours of Clinical trials. There are certain substances that have been cleared for safety for use in face creams and the different brands simply use different combinations of these substances.

the original point about statistics was not to do with clincial trials. we were talking about statistics in general.

2) Within reason, the size of a sample is not as important as its "representativeness" of the population. Obviously testing on 5 people, say, would be meaningless.

totally agree

3) Yes, anyone can take some numbers etc., but a trained statistician (or even an intelligent layman with some knowledge of stats) can easily refute their claims. The problem is that most people do not want to bother to learn or can't learn the basic stats that enables them to differentiate dross from gold.

I have learned about statistics and I still wouldn't trust them as far as I could throw them.
why should people have to learn about how these calculations are made and what they really mean? that's what we have statisticians for. I mean would you expect us all to have medical knowledge before we go to a hostpital just so we can tell the doctor/nurse when they are doing something wrong?

[sasguru]

the original point about statistics was not to do with clincial trials. we were talking about statistics in general.


totally agree



I have learned about statistics and I still wouldn't trust them as far as I could throw them.
why should people have to learn about how these calculations are made and what they really mean? that's what we have statisticians for. I mean would you expect us all to have medical knowledge before we go to a hostpital just so we can tell the doctor/nurse when they are doing something wrong?

Ok now I believe you are female
That's at least 2 for sure. Cojak being the other one.

[janey]

Ok now I believe you are female

and why's that mister!?!

[sasguru]

and why's that mister!?!

If you don't know, there's no point explaining.