Creationist Statistics

Statistical Sampling 101 …Suppose you find a big box of marbles. You reach in and grab six marbles at random. When you remove the marbles, you discover that each marble is either red, green, or blue…This sample is so small that is may not be representative of all the colors in the box…However, you keep going until you've pulled about 1,000 marbles out of the box. You look at them all, and still find only red, green, and blue ones. There are still some marbles left in the box. What colors would you guess they are?
Explore Evolution, p. 31

What Explore Evolution is really saying here is this: We do not see abundant trace fossils because there were no animals to make them. Therefore Cambrian animals had no ancestors, but were suddenly "created." This is wrong in that it fails to consider the type of animal that existed at that time. Marbles: from  Wikicommons Marbles: from Wikicommons

But in an additional sense, this example shows why every scientist must have a working knowledge of statistics. Yes, the likelihood is that the next marble produced in this scenario would be red, green, or blue. But if among the thousands of marbles, you had planted a white marble, then even though this white marble was present, it would be very unlikely to be found. That's not the same thing as saying it could not be there. In fact, not until every single marble was removed could you say with any assurance that this hypothetical box contained only red, green, or blue. There might have been 1 white, 2 black, 3 purple, and so on. But the very small numbers of these odd colors make it very unlikely that would they be found in a random draw.

There's an old joke that goes like this:

A businessman, a philosopher, and a scientist are on a train traveling through the countryside. They see a solitary black sheep grazing in a field.

The businessman says, "Sheep are black."

The philosopher says, "At least one sheep is black."

The scientist says, "At least one side of one sheep is black."