"The word average is one of the most dangerous words in the universe" Which "Average" are we referencing when we state the word? Arithmetical mean, Geometric Mean, RMS, Median, Mode then allowing for standard deviations and non-standard deviations. Sometimes these figures hover around a similar approximation yet often they do not. ... They can be used to advise us of how typical things might be yet they can/are often used to deceive us both intentionally or accidently.
All very true. However, since we live in a world in which almost everything (both 'animal' and 'mineral') is subject to some degree of variation (commonly fairly 'random'), we are totally reliant on ;averages' (of whatever sort) as a means of describing/reporting the 'typical' properties of things (dimensions, weight, numbers, life expectancy etc. etc.)
However, as you say, averages can easily be misunderstood/misinterpreted, and are sometimes deliberately used to deceives - and the same is true of 'percentages'.
In some contexts, mean and median 'averages' may be similar - indeed, they will be identical if the pattern of 'failures' (or dimensions, or whatever) is 'symmetrical' around the average. The more 'symmetrical', the greater will be the difference. However, as I've said, the main difference is a practical one - testing to establish a median survival of, say, 1,000 hours need only take a bit over 1,000 hours (about 6months), whereas to determine the mean survival of the same batch of products might take 20 years or more (having to continue until the last item fails).
How means can 'mislead' was brought home to me when, many moons ago, I was researching my family history. We are inclined to believe that average human life expectancy has, in most countries, increased dramatically over the decades and centuries - and that is undoubtedly true in terms of mean life expectancy. However, when doing my research, I noted that they great majority of my ancestors lived for roughly the same amount of time as they do today, most dying between 65 and 85. I also noted that even thousands of years ago, there are Biblical references to human lifespan being "three score years and 10" (70 years).
The explanation, of course, was that there was a very high rate of child mortality, with many babies dying withing days or weeks of birth, and probably less than half surviving to adolescence. This large number of very short life-expectancies dramatically reduced the
mean survival - but the
median survival has not changed all that much - if the median is, say, 70 years, when those who survive less than 70 years die when they are one day old, then that median will still be 70b years if all those who dies befoore 70 die at, say, 69.
Traditional filament lamps that we have employed for years had stated average life expectancy. So far so good (not withstanding what I wrote above) yet how were those averages obtained anyhow? Often when doing destructive tests to see life expectancy of filament lamps they quoted times on/off (the mark/space ratio) of being several hours at a time, how often do we use filament lamps or any lamp at all in large hourly cycles. We often switched the on for a few mins then off again for a few mins or a few hours. All these short duration switching on and off really stresses filament lamps and throws out of kilter any results, reliable or not, for figure gained by three or four hour constants during test runs.
Very much so. It was pretty rare for filament bulbs to fail
other than at 'switch on' (when filament was cold and hence low resistance), so testing should obviously seek to reflect the pattern of on/off/on switching in normal service. One cannot really expect them to test under countless different ';usage conditions' (which would probably just confuse users, anyway!)(,so I imagine they use a 'usage pattern' which they believe, rightly or wrongly, is typical of 'common usage' - but maybe they've got that wrong.
I have to say that, at least in my case, although we were probably much more 'careful' (because of 'running costs') in the days of incandescent bulbs, we don't do much 'switching on and off' of lights these days, other than perhaps in loos and bathrooms. Lights in 'living areas' tend to be left on continuously during non-daylight hours, and those in bedrooms just switched on/off once per day.
With oven lamps etc, you also have obscene temperature differences and air flow thrown into the mix.
Indeed. As I've always said, oven lamps/bulbs are a very exceptional case,and one would hope that bulbs designed for this use are tested using something approaching the pattern of real-world use. However, as I've said,I'm not convinced that all lamps/bulbs used in ovens are necessarily designed and manufactured specifically for only that application - in which case some might only be tested under totally inappropriate conditions (for an oven lamp/bulb).
Kind Regards, John
I would suggest that leaving them on days/weeks/months/years gives the only real expectation of life and gives us a true best figure to work with, the only way we can beat that is to leave them switched off, they last forever then.
This is true of filament lamps and perhaps more so of fluorescents too. LEDs I`m not sure of, they are still "relatively young".