Tuesday, February 06, 2007

Hated Statistics Class

In the evolution/creationism debate, we often see an argument in which the creationist puts together computations of various processes, and by multiplying out the probabilities of each of these events occurring, forms some long number, claiming that it is extremely unlikely this process could happen. Therefore, the claim is made, evolution is extremely improbable.

The premise being that there are two choices: Evolution or Creation, and if the creationist can make evolution less possible; by virtue of the only other choice, creation must be more plausible.

Now, I may not be well-versed in evolution, but I am well-versed in arguments. And this one smells rotten.

Statistical odds are derived by reviewing relevant data. We could sit by an intersection in the road. Over the course of time, by counting the number of cars that go through the intersection, and comparing how many went to the right, and how many went to the left, we can determine the “odds” of a car going to the right. We can count red cars, two-door cars, cars with broken windshields. We can use a whole variety of factors to make the odds of a red, two-door car with a broken windshield going to the right as compared to the left.

And we can look at other intersections. How are the odds affected if by going to the right, one heads to the country, and going to the left, one heads to a city? What if there is advertising to go one way or another? What if there is a signal?

And, after careful observation, and utilizing the information obtained, we can safely place our new McDonald’s on the intersection that is more likely to be traveled. (Thus affecting the odds for the next statistician that comes along.)

So the creationists look at the odds of how evolution could happen. Often what we see is some proposed process or time line, and the number of mutations or changes in generations that must happen, and the chance that these changes could both occur AND occur in the time allotted.

Eventually they pop out with some number that seems incredibly huge, saying the chance of this occurring is 1 in 10 to the 2000th, or a 10 with 2000 zeros behind it. A number so large, it takes two people to read it.

Now, I seriously question that it is a true dichotomy between evolution and creation. While admittedly I cannot see a third alternative, 200 years ago, people could not see evolution as an alternative, either. 100 years ago the theory of relativity or quantum mechanics would be laughed at as science fiction. For all I know, 200 years from now, a new theory will be proposed and demonstrated, and evolution with creationism become footnotes with geocentricism.

I further question the method of obtaining the statistics on evolution. One item that is often over-looked is that the processes are considered sequential (one after another) when they could be happening simultaneously. Like determining how long it takes to fill a stadium with 50,000 people one at a time, as compared to a stadium with 50 entrances able to take one at a time. Another item is that some processes may necessarily entail a change, so it is NOT another completely wide-open field. Like rolling a rock down a hill, if it goes to the right of a bump, the next foot of travel will necessarily be the right of that bump. It is not as if every instant, the rock could go right or left.

But even assuming we DO have only two choices. And assume that the statistical odds of evolution happening are 1 in 10 to the trillionth billionth power. A 1 with a trillion billion zeroes behind it. A number with so more zeros than a person could count in a life time.

Does this minuscule, teeny-tiny, vastly remote possibility make creationism more likely? Well, to know THAT, we have to know what the statistical odds are for creationism to do the same thing.

And that is where the problem rears its head. We have no way to obtain the statistical odds of what a God would or would not do. Statistics are derived from observation. We can’t observe God. Ergo—no statistics on God.

Look, what are the odds that God will have oatmeal for breakfast tomorrow. Come on. This is an easy one. No proteins, or processes or speculation as to the earth’s condition 450 Million years ago. Oatmeal. Breakfast. Oh, what’s that you say? God is on a supernatural plane, so we cannot see what He eats or when? Right—no way to determine whether it is a 1 in 10 or 1 in 10 the trillionth billionth power as to whether God eats oatmeal.

Or God sneezes. Or God speaks. Or God does anything, because we have no way to observe the raw data to come up with the statistic in the first place!

On the one hand we have evolution with a 1 in 10 to the trillionth billionth power chance. And on the other we have “No information available.” Simply put, because of the definition of God, we can’t use the same method we can with observable nature.

So what good is this argument? None, really. As we cannot make the comparison. Sure, we understand that if evolution has a 1 in 10 chance, and creationism has a 1 in 9 chance, using this argument creationism is more likely. Evident as pie. But instead we have evolution with a 1 in some astronomical number chance and creationism with no way to calculate its chance at all.

To use a euphemism, we are comparing apples to oranges.

But a creationist never goes there. A creationist never explains that the same method it is using on evolution can never be used on creation.

I suggest we put the shoe on the other foot.

“There is a 1 in 10 the trillionth trillionth billionth power that creationism could occur.”
“Where did you get that statistic?
“Good question. Where do you propose I look for statistics as to the probability of creationism?”

As a trial lawyer, one thing we learn is that people hold stronger to a conclusion they derive themselves. If I point out every step and every inference of every step, coming to a strong point, the jury will politely listen and retain the information. But if I lead them every step of the way, and leave the last point tantalizingly out there, they come to the same conclusion, but since they “figured it out,” it is a more firmly held belief. Because it is theirs, not something someone told them.

I get the same impression here. The creationist points out all the remote possibilities, piles them on, comes (eventually) to this grandiose number, compares it 747’s in junkyards, and then stops. The person is supposed to come to the conclusion that therefore, evolution is so remote, creationism must be more plausible.

No creationist ever provides the statistics to compare. Because, by the very definition of God being unknowable, no statistics are available.

If both possibilities cannot use the same method, it is useless to use it to determine what is “more probable.”


  1. The problem with the creationist statistics argument is quite simple: it ignores cumulative adaptation.

  2. Dagoods, why couldn't you say that the more improbable evolution is, the more probable creation is? That way, you wouldn't have to evaluate the statistical probability of creation in isolation.

  3. ephphatha,

    Assuming those were the only two possible choices—you could. (You know I do not agree that those are necessarily the only two choices, but for this purpose, it doesn’t matter.)

    But what gain is it?

    I find a piece of evidence that makes evolution 10 times less likely. Arguably 10 times less probable. This makes creationism 10 times more probable. So which is now more probable? Depends on our starting position, true?

    If evolution had a 1:1000 chance; my piece of evidence would now make it 1:10,000. But where does creationism move from and to? What if creationism started off at 1:1,000,000? Now, it is at 1:100,000. STILL less probable.

    The trap they have sold you on is that is that the statistical odds are equal. How can that be a known, if we do not have any data by which to determine the statistics on one side?

  4. I think you have a good point there, Dagoods.