This is where a the bell curve and 3 sigma standard come into play. Below I made a rough assumption of what one would look like for the bursting pressure of PVC pipe by simply adding some numbers to a already existing graph I found on the net, shown below.
Basically the height of the curve denotes the percentage of a given event occurring. In our case, the possibility of the pipe bursting. The most common percentage is listed in each 8 portions the curve has been divided into. And the further 'right' you move on the graph the higher results the given situation produces. In our case, a higher burst pressure. As you can see I went ahead and listed some numbers to give a idea on what the bursting pressures would roughly look like. Pretend as if I went out and tested 10,000 pipes for there bursting pressure and what I got is what is listed on the graph.
Since the height of the wave occurs at 250PSI, that also means that the median bursting pressure is 250PSI. Although a major fluke in production that produces weak chemical bonds in the PVC occurs .13% of the time resulting a low bursting pressure [0-40PSI], and another .13% of the time a uber-strong bond is formed by the PVC resulting in pressure withstanding capabilities of 500+PSI. Both these instances are rarely accounted since of there rarity. The other 6 portions are more plausible, and increase with there likeliness as there height in the curve increases.
So if you were to go test a section of pipe according to this data the pipe would most likely burst at around 250PSI. But you dont want to approach anywhere close to that median in fear of failure. So to accommodate for this fear they will rate the item at a negative deviation of the median. Something such as pipe is held to high negative deviations since failures of such can result in serious injuries or death. While something such as rubber band is not, since what happens when a rubber band snaps? Nothing, so it doesnt need that high of standards. Problem being some companies will get greedy and rate a product at a lower deviation to make there product appear better. But most industries can venture into -4 deviation on a 3 sigma standard since in the pipe industry that would require you to make a 1" thick pipe rated for 10PSI to be absolutely sure it will not fail.
Chances are that pipe companies have tested huge amounts of pipe for there failure point and had a few fail under there claimed safety figure, but the chances of someone having such fail on them is so low, its accepted. I mean there wouldnt be a pharmaceutical drug on the market if they were held to a -4 deviation on a 3 sigma standard.
This is why we avoid even approaching the stated pressure rating of pipes since as we can conclude that there is proven flukes that may end in our possession. The results of a failure in our instance is not acceptable, hence why its held to such high standards. Damn I hate when I have to sit here and cover something so extensively before someone understands...
[Ill admit the bell curve actually graphed for our situation would be steeper, but you get the idea
EDIT- Ive edited the post a billion times in a attempt to get the picture to work, and finally got it working. Its now working, and not a worry. MrCrowley, thanks for the heads and now I will delete your post since it has no matter in this discussion.