Tabby,

I'm just a cell phone salesman and radio tech, but I came from a QA background where it was all about numbers. What I learned was that numbers seldom are hard and fast as it relates to things in nature but really are a range, from less than mean to well over the mean. In fact, it is by examining the range and distribution of numbers and values that we can determine if something is natural or contrived. Whenever too many members of a population are clustered too close to the mean, or when the mean is skewed to one side of the median or the other, then it can be said that the values are not even random but are not in fact a true representation of the natural state of the given population. It might be that the sample was too small or that the categories sampled were contrived to demonstrate a premise rather than the theory coming from the random selection of the population for sampling purposes.

This is why a statistic like "most affair end within two years" is so hard to grasp. We can all identify those samples that have lasted longer than two years or that have lasted less than that number. But what this means is that some affairs are really only a ONS and others, an equally low number, in fact probably lower, turn into long lasting even healthy marriages. That does not contradict the reality that the majority of all affairs will lie within some point between a ONS and two years.

There is another number that is often forgotten when looking at a sample of all affairs and that would be the fact that not all affairs even begin. Some even get close to being an affair without actually crossing the line because of the good boundaries of those involved. Some never happen because of other factors such as fear etc. These too are part of the population.

And that is why I have singled out our friend Zelmo on this matter. His distribution is skewed and shows curtosis, that is, it limits the tails of the distribution such that it can be assumed that the sample was not random or the wrong categories were assigned to the values measured.

Wherever randomness exists, which is everywhere in a natural setting, it is not some chaotic scramble that has no meaning. Randomness can be clearly defined and is always the same for all things. It is the "bell curve" used to analyze grades in school. It is a Gaussian distribution or it is not truly random. There are those things that are outliers, that fall well outside the range of what might be called normal but the shape of the curve must always be the same to be random.

The larger the sample, the more randomness it represents. Too small of a sample cannot describe a complex variable to any kind of certainty. Too large of a sample makes the task tedious and mind numbing often causing the one looking at the numbers to get lost in the details without recognizing any kind of distinguishing pattern. But if a sample is truly random, then it will always show a range of values that is always the same shape, always clustered evenly about the mean and the mean will have the same value as the median.

That is how you can spot bovine excrement in peoples numbers or theories. If it's too perfect to be true it can't be true.

And that dear friends, is the reason we use terms like typical, most, some, many and often instead of all, always and every.

Mark