Low profile downlights

[EFLImpudence]

I don't think your quantity analogy is appropriate.
I'm not sure why "contents not less than 8oz" is unrealistic.

"Not less than" or "250ge" is specific and people know it is illegal to supply less than that figure.
On the same basis they might think (possibly unrealistically but that's what it says) that 30,000 hours was indeed a minimum and that is why they get upset when several fail quite quickly.

Also, on the basis that the hours figure is an average must mean it is thought that most of the lamps will make it to reasonably close to that figure - some just under; some just over.
If it is an average, it must be concluded that lamps which fail relatively quickly will have counterparts which last almost twice as long. This would be unreasonable so it really means that a few will fail relatively quickly but most will last a bit longer than the hours figure.

It would seem there is no satisfactory way to do it.

"Most of these lamps will last 30,000 hours but we have no idea how long this one will"

 

[Notch7]

"Most of these lamps will last 30,000 hours but we have no idea how long this one will"

I like that!

anticipated lifetime figures for lamps is a bit of marketing blurb. I wonder what level of testing is actually done.

 

[JohnW2]

I don't think your quantity analogy is appropriate.

It was an attempt to illustrate that people seem to have accepted the concept of average-based figures, even when they are not explicitly described as such (other than by a symbol, the meaning iof which very few people know).

I'm not sure why "contents not less than 8oz" is unrealistic. "Not less than" .... is specific and people know it is illegal to supply less than that figure.

Exactly. As you say, in those days it was illegal for the contents to be, say, less than 8oz, which is the reason why it was unrealistic for manufacturers, and why the manufacturers lobbied to have the law changed. Under than old system, and given the appreciable (often pretty wide) spread of manufacturing tolerances, the only way that manufacturers could attempt to avoid breaking the law (hence avoiding possible litigation, or even prosecution) was for the average contents to be considerably more than the "not less than" labelled figure. Hence a "not less than 8oz" box of cornflakes might have had to have been manufactured so as to have an average contents, of 10oz, 12oz or even more (depending upon 'how sure' they wanted to be that they would 'almost never' {'never never' is probably impossible} be breaking the law) - which was commercially unrealistic.

...

or "250ge" is specific and people know it is illegal to supply less than that figure.

No, and that's the whole point of the 'e'. With e-labelled contents, one cannot break the law by selling a particular box of cornflakes which has a very low content. With the e, the amount of the contents is defined statistically/probabilistically. A contents of 250ge implies that the average weight of a large number of samples of that product will be about 250g and guarantees that a certain percentage (commonly 95%) of individual boxes will contain 250g±x% (x% commonly being 5%). Implicit in that probabilistic definition, and 'legal', is the fact that (if the production spread is 'symmetrical') 2.5% of boxes will contain less than 250g-x%.

Hence, if x=5% and the required probability 95%, then if it could be shown (by weighing the contents of a large number of boxes) that more than 2.5% of boxes contained less than 237.5g, then the law would have been broken. However, the fact that you or I had bought a box which contained only, say, 220g or 230g would not necessarily mean that the law had been broken, and certainly would not be a demonstration that the law had been broken.

Also, on the basis that the hours figure is an average must mean it is thought that most of the lamps will make it to reasonably close to that figure - some just under; some just over.

I suggest that you probably need to think again about that one! Where did that 'just' ('just under' and 'just over') come from? In many situations, there is a very wide spread on both sides of an 'average'. I haven't a clue as to what the actual figure is, but if the 'average population' of a country was, say, 50 million, that doesn't mean that there aren't countries with under 1 million or over 1 billion.

If it is an average, it must be concluded that lamps which fail relatively quickly will have counterparts which last almost twice as long. This would be unreasonable so it really means that a few will fail relatively quickly but most will last a bit longer than the hours figure.

Why would that be 'unreasonable'? I've had "1000 hour" incandescent bulbs die in the first few days, and others which have kept going for thousands of hours.

You also seem to be assuming that individual figures are necessary distributed similarly on both sides of the 'average', which is far from necessarily the case. Average lifespan may be around 85 years, with people dying at all ages from 0 upwards, but never more than about 110 years - so there is a spread of 85 years below the average, and only about 25 years above it.

Kind Regards, John

 

[EFLImpudence]

No, and that's the whole point of the 'e'. With e-labelled contents, one cannot break the law by selling a particular box of cornflakes which has a very low content.

Ah. Sorry. I did not realise the main point, so I would say most people don't realise that, either.

I suggest that you probably need to think again about that one! Where did that 'just' ('just under' and 'just over') come from? In many situations, there is a very wide spread on both sides of an 'average'. I haven't a clue as to what the actual figure is, but if the 'average population' of a country was, say, 50 million, that doesn't mean that there aren't countries with under 1 million or over 1 billion.

No, but that's not the same.
A country might have 1 million and another 99 million.
My point was that I would think it unlikely that a lamp could last twice as long because one failed instantly, but...

Why would that be 'unreasonable'? I've had "1000 hour" incandescent bulbs die in the first few days, and others which have kept going for thousands of hours.

...you seem to think that is possible.
This just makes the hours figure even less helpful.

You also seem to be assuming that individual figures are necessary distributed similarly on both sides of the 'average', which is far from necessarily the case. Average lifespan may be around 85 years, with people dying at all ages from 0 upwards, but never more than about 110 years - so there is a spread of 85 years below the average, and only about 25 years above it.

That was my point. but you seem to be contradicting yourself.
If 10% of lamps fail quickly then that means the actual life of the other 90% is increased a little.
If 50% of lamps fail quickly then that means the actual life of the other 50% is nearly doubled - I thought not likely but you have said different - the 1000 hour.

It seemed far more likely to me that with a couple of exceptions most lamps would be within a reasonable percentage of the average.

 

[JohnW2]

Ah. Sorry. I did not realise the main point, so I would say most people don't realise that, either.

Fair enough. As I said, I doubt that hardly any people realise what the e symbol means, but I imagined that most of them probably realised that what was being quoted was an 'approximate/average' figure, rather than a "must not be less than" one - but maybe I'm wrong about that. If you were to buy, say, 1 tonne (or 1 m³) of sand, gravel or something like that, would you not expect (and accept) that that was an 'approximate' figure, rather than a "guaranteed not to be less than" one?

My point was that I would think it unlikely that a lamp could last twice as long because one failed instantly, but... ...you seem to think that is possible.

I do. It's surely true of very many products - we're always hearing anecdotes about nearly new ones that have failed long before they were expected to, and also anecdotes about things which are 'still working fine' years or decades beyond the end of their expected/intended lifetime, aren't we?

This just makes the hours figure even less helpful.

As I said, I think the figures are 'helpful'/useful, provided one does not over-interpret (or incorrectly interpret) them. A "3000 hour" incandescent bulb is "very likely" to (but not guaranteed to) last a lot longer than a "1000 hour" one. However, given that a rogue "3000 hour" one might only last for 5 minutes, there's a limit to what one can read into these figures.

That was my point. but you seem to be contradicting yourself....If 10% of lamps fail quickly then that means the actual life of the other 90% is increased a little.

I think you may be overlooking the fact that the average we are talking about here is a median, not a mean. As a result, all that matters is which side of the average (median) values are on. By definition, the median is the value which splits the figures into "50% below" and "50% above" that median. If the median were 25,000 when 50% of the individual figures were 24,999 (and one was 25,000), if nothing else changed, the median would still be 25,000 if that 50% of 'below median' ones changed from being 24,999 to being, say, 200. Similarly ....

If 50% of lamps fail quickly then that means the actual life of the other 50% is nearly doubled

You are again thinking of means, not medians. If 50% of lamps (well, technically 50% minus 1) failed within (to be extreme) the first 100 hours, the median could still be 25,000 if all of the others had lifespans ranging from, say, 25,000 to 25,100 - so no need (with medians) for any 'nearly doubled' ones. The mean would then be around 12,500, but the median (which is what is being quoted for lamps) would still be around 25,000.

- I thought not likely but you have said different - the 1000 hour. It seemed far more likely to me that with a couple of exceptions most lamps would be within a reasonable percentage of the average.

If we were talking about the weight of the contents of a box of contents, the actual diameter of a nominally 6mm bolt or the number of cocktail sticks in a 'box of 1,000", then I'm sure you'd be right. However, when it comes to talking about something subject to so many factors/variables (some pretty random) as the lifespan of a lamp/bulb, I think ones has to expect far more variation on either side of the average. We've all had some lamps/bulbs (or many other things) that have only lasted for a very short time, and others which seem to have lasted 'for ever'!

Kind Regards, John

 

[EFLImpudence]

Fair enough. As I said, I doubt that hardly any people realise what the e symbol means, but I imagined that most of them probably realised that what was being quoted was an 'approximate/average' figure, rather than a "must not be less than" one - but maybe I'm wrong about that. If you were to buy, say, 1 tonne (or 1 m³) of sand, gravel or something like that, would you not expect (and accept) that that was an 'approximate' figure, rather than a "guaranteed not to be less than" one?

No, not at all, I would expect and demand a tonne - or any other weight.

I do. It's surely true of very many products - we're always hearing anecdotes about nearly new ones that have failed long before they were expected to, and also anecdotes about things which are 'still working fine' years or decades beyond the end of their expected/intended lifetime, aren't we?

Ok. That's fair enough, but then in fact the life expectancy is unknown and random.

As I said, I think the figures are 'helpful'/useful, provided one does not over-interpret (or incorrectly interpret) them. A "3000 hour" incandescent bulb is "very likely" to (but not guaranteed to) last a lot longer than a "1000 hour" one. However, given that a rogue "3000 hour" one might only last for 5 minutes, there's a limit to what one can read into these figures.
I think you may be overlooking the fact that the average we are talking about here is a median, not a mean. As a result, all that matters is which side of the average (median) values are on. By definition, the median is the value which splits the figures into "50% below" and "50% above" that median. If the median were 25,000 when 50% of the individual figures were 24,999 (and one was 25,000), if nothing else changed, the median would still be 25,000 if that 50% of 'below median' ones changed from being 24,999 to being, say, 200. Similarly

Ok. but I think it still unsatisfactory.
I'm not sure I, or anyone else, should have known it was median and not mean unless that is how it is always done.
I either missed it or it is never stated.

You are again thinking of means, not medians. If 50% of lamps (well, technically 50% minus 1) failed within (to be extreme) the first 100 hours, the median could still be 25,000 if all of the others had lifespans ranging from, say, 25,000 to 25,100 - so no need (with medians) for any 'nearly doubled' ones. The mean would then be around 12,500, but the median (which is what is being quoted for lamps) would still be around 25,000. If we were talking about the weight of the contents of a box of contents, the actual diameter of a nominally 6mm bolt or the number of cocktail sticks in a 'box of 1,000", then I'm sure you'd be right. However, when it comes to talking about something subject to so many factors/variables (some pretty random) as the lifespan of a lamp/bulb, I think ones has to expect far more variation on either side of the average. We've all had some lamps/bulbs (or many other things) that have only lasted for a very short time, and others which seem to have lasted 'for ever'!

Fair enough - for median - but how many people would know that?

 

[ban-all-sheds]

I don't know how they get away with such a statistic.

Because they define it.

If the OP has two fail in a relatively short time, does that mean the other two will definitely last that long?

No? so an entirely meaningless lie.

No, 4 is a statistically insignificant sample size.

 

[EFLImpudence]

Because they define it.

Where?

 

[ban-all-sheds]

if the 'average population' of a country was, say, 50 million, that doesn't mean that there aren't countries with under 1 million or over 1 billion.

For a human being, I have an above average number of arms, legs and eyes.

 

[JohnW2]

No, not at all, I would expect and demand a tonne - or any other weight.

It sounds as if you think ion terms of "Bakers' Dozens"! Do you therefore expect (and think it would be fair to the suppliers) that, given the inevitable errors in relation to measurements and losses etc., that they should supply considerably more than 1 tonne to 99% of customers who have ordered (and paid for) 1 tonne, in order to avoid a situation in which <1% of customers might get slightly less than 1 tonne (i.e. a reversion to the "not less than" days.

Ok. but I think it still unsatisfactory. I'm not sure I, or anyone else, should have known it was median and not mean unless that is how it is always done. I either missed it or it is never stated. Fair enough - for median - but how many people would know that?

It's a very standard, and I would say sensible, 'average' to use for this sort of thing, not the least because a mean could be even more misleading.

A major problem with means is that they are appreciably influenced by markedly 'outlying' values. Imagine a situation in which most lamps had a life of about 25,000 hours, but, for whatever reason, a few percent had lives of 100,000+ hours, some perhaps much longer than that. The mean could then be considerably greater than 25,000, despite the fact that nearly all of them had a life appreciably less than the mean figure that was being quoted. Would you really prefer to be told the (I would say, very misleading) mean in that situation?

There's a similar problem with historic human life expectancy, but because of issues 'at the low end', rather than the 'high end'. Over the past few centuries, mean lifespan has increased dramatically, but median lifespan hasn't gone up all that much. Even hundreds of years ago, people who reached adulthood commonly lasted until their 60s/70s/80s (even the Bible, speaks of a lifespan of "three score years and ten", thousands of years ago) - and, for what that's worth, that has certainly been what I've seen in tracing my family back to the 1500s. The mean used to be so much lower than it is now because of the very high rate of infant and child deaths - if the lifespan of the rest is, say, around 70 years, it does not take an awful lot whose lifespans are less than 1 year (or less than 5 or 10 years, to pull the mean down to considerably lower than 70, even though that is the age that 'most' people reached. On the other hand, the median is little affected by those 'very young deaths'.

Kind Regards, John

 

[JohnW2]

BAS undoubtedly does have an 'above mean' number of arms, legs and eyes - but, unless he is an unbelievably rare oddity, he will not have an 'above median' number of arms legs or eyes - so, yes, another illustration why medians are often more sensible/useful 'averages' than are means.

 

[333rocky333]

Theres quite a lot of slim lights nowadays, the problem is more where to put the Driver that powers them and more so if theres insulation in the void
https://www.google.co.uk/search?q=l.....69i57j0l5.9403j0j4&sourceid=chrome&ie=UTF-8

 

[EFLImpudence]

It sounds as if you think ion terms of "Bakers' Dozens"! Do you therefore expect (and think it would be fair to the suppliers) that, given the inevitable errors in relation to measurements and losses etc., that they should supply considerably more than 1 tonne to 99% of customers who have ordered (and paid for) 1 tonne, in order to avoid a situation in which <1% of customers might get slightly less than 1 tonne (i.e. a reversion to the "not less than" days.
It's a very standard, and I would say sensible, 'average' to use for this sort of thing, not the least because a mean could be even more misleading.

I disagree, and so, I would think, do Weights and Measures.
Does it apply, for example, to fuel?
Some beer glasses are oversized and marked with the correct volume. The optics are checked for accuracy.
You know why the Bakers' Dozen came about but it is no longer necassery with modern weighing and charging equipment. With them people can be charged the correct amount for underweight items and the same for overweight ones IF they agree; if not the excess will have to be removed or given free.

A major problem with means is that they are appreciably influenced by markedly 'outlying' values. Imagine a situation in which most lamps had a life of about 25,000 hours, but, for whatever reason, a few percent had lives of 100,000+ hours, some perhaps much longer than that. The mean could then be considerably greater than 25,000, despite the fact that nearly all of them had a life appreciably less than the mean figure that was being quoted. Would you really prefer to be told the (I would say, very misleading) mean in that situation?

What you say might be correct in such circumstances but I do not understand how you could judge such figures, in most people's circumstances, to be other than completely irrelevant and therefore meaningless.

There's a similar problem with historic human life expectancy, but because of issues 'at the low end', rather than the 'high end'. Over the past few centuries, mean lifespan has increased dramatically, but median lifespan hasn't gone up all that much. Even hundreds of years ago, people who reached adulthood commonly lasted until their 60s/70s/80s (even the Bible, speaks of a lifespan of "three score years and ten", thousands of years ago) - and, for what that's worth, that has certainly been what I've seen in tracing my family back to the 1500s. The mean used to be so much lower than it is now because of the very high rate of infant and child deaths - if the lifespan of the rest is, say, around 70 years, it does not take an awful lot whose lifespans are less than 1 year (or less than 5 or 10 years, to pull the mean down to considerably lower than 70, even though that is the age that 'most' people reached. On the other hand, the median is little affected by those 'very young deaths'.

That is true, as I also discovered looking into ancestry, but again then the figure is virtually meaningless.

 

[JohnW2]

I disagree, and so, I would think, do Weights and Measures. Does it apply, for example, to fuel?
Some beer glasses are oversized and marked with the correct volume. The optics are checked for accuracy.

I'm certainly no authoprity on such matters, but I was under the impression that when Trading Standards people check the accuracy of pub measures, petrol pump metering, scales in shops etc., the requirement (for compliance with the law) was that the measuring equipment being used in the pub, petrol dtation, shop or whatever had top be accurate to within "plus or minus X" - not "plus, but not minus X" - but maybe I'm wrong.

You know why the Bakers' Dozen came about but it is no longer necessary with modern weighing and charging equipment.

Oh, I didn't think that was the origin. I'd always been under the impression that it arose because penalties for providing (by an error of counting) only 11 items when 12 had been ordered (and paid for) were so severe that bakers routinely gave 13 items when 'a dozen' had been ordered, 'just to be sure'.

What you say might be correct in such circumstances but I do not understand how you could judge such figures, in most people's circumstances, to be other than completely irrelevant and therefore meaningless.

If all you told was an 'average' (which was, in fact, a mean), you wouldn't know how it had come about. My point was that a small number of very low or very high figures can seriously distort a mean (rendering it a potentially misleading 'average'), whereas that is much less of a problem with medians.

That is true, as I also discovered looking into ancestry, but again then the figure is virtually meaningless.

Means are often pretty meaningless, particularly in cases such as I mentioned - a mean lifespan which results from 'averaging together' a lot of people who lived less than 10 years and a lot of people who lived for more than 50 years (with not a lot in between), resulting in an 'average' of, say '30 years' is almost totally meaningless (for most purposes - see **). On the other hand, median lifespan is very meaningful, since it indicates the age which was achieved (or exceeded) by half of the people who were born.

** means are not only useful but essential if one is going to use them to undertake calculations which involve the concept of things like 'person-years'. For example, if one wants to calculate the total cost of keeping X people in prison, if one multiples the mean number of years of stay by the cost of keeping a person in prison for a year, one will get a 'correct' answer (which one wouldn't if one used the median stay). Actuaries in the insurance/pension etc. areas use this approach all the time.

As for my ancestry, once one gets early than 20th century, I would say that something like 30% survived until adolescence - but, if they did, relatively few died before 50/60 (those that did being mainly women who died in relation to pregnancy), and plenty got into their 70s or beyond - i.e. not that much different from now.

Kind Regards, John

 

[ban-all-sheds]

BAS undoubtedly does have an 'above mean' number of arms, legs and eyes - but, unless he is an unbelievably rare oddity, he will not have an 'above median' number of arms legs or eyes

I'm pretty sure I do.

I wonder if you mean "mode"?