Science and Falsification of Data

[Simon_Jester]

Strictly speaking, you cant. When you are doing climate science or any statistics you are working with samples. If I take a sample of temperatures from thirty years ago and compare them to equivalent modern temps (and bear in mind, you are taking a sample, not the entire population of temperatures from all locations at all points in time), I am creating a probability distribution. I can prove something is so likely to be different, that it is stupid to disagree... but I have not absolutely proven that the temperatures are different. It is sort of like claiming "If I lean against this wall, I will pass through it". There is a non-zero chance that you will in fact pass through the wall thanks to quantum mechanics. However, the chance is so small you can laugh at it.

At this time there are so many different sources, using different methods, that all reach the same conclusion, that a statement like "it is generally warmer today than 30 years ago" is pretty much in the same region as your falling through the wall example. Now, I would be willing to admit that there is some uncertainty in the temperature record of the last few decades, except it is so negligibly tiny in comparison to the temperature difference as to be safely ignorable.

While I agree, I cannot have you making incorrect statements about science or statistics.

The trouble with this is that when you're speaking in English (as opposed to the mutant dialect used by Popperian philosophers), saying "there is a non-zero chance that you will pass through the wall" is more misleading, and more incorrect in the language you're using, than saying "you will not pass through the wall."

Human beings, even scientifically literate ones, have no concept of what a probability of one in a trillion or less means. It's completely foreign to their experience. When speaking in normal English, the statement "X might happen" implies a realistic chance that X might happen, that if we wait a reasonable amount of time X will actually happen to someone, somewhere. For low enough odds, odds that can be described quite well by mathematics, that's not realistically possible... at which point, for purposes of human cognition, the probability of the event occuring is exactly 0.

Even odds as high as one in a million are hard for human beings to grasp, as witnessed by the popularity of lottery tickets. All your brain really processes is "I might win the lottery." It takes considerable mental discipline to stop and realize just how crappy your chances are.

So when talking to people who are prone to misinterpret the meaning of saying "might" while talking about vanishingly small probabilities (i.e. practically everyone), it's perfectly reasonable to say "will not" instead of "probably will not." If you run full tilt into the wall, you will not tunnel through it, you will smack into it, and that's the end of matters. Any other outcome is so unlikely that it is not worth taking the time to consider, and mentioning other outcomes serves only to mislead, unless we are discussing the subject of quantum tunneling specifically.

[Zed]

Edit: deleted.

[Anguirus]

I have the impression that people are pulling statistical data on probabilities out of their asses.

I beg your pardon? Would you please clarify this statement?

[Alyrium Denryle]

So when talking to people who are prone to misinterpret the meaning of saying "might" while talking about vanishingly small probabilities (i.e. practically everyone), it's perfectly reasonable to say "will not" instead of "probably will not." If you run full tilt into the wall, you will not tunnel through it, you will smack into it, and that's the end of matters. Any other outcome is so unlikely that it is not worth taking the time to consider, and mentioning other outcomes serves only to mislead, unless we are discussing the subject of quantum tunneling specifically.

I have had good success actually describing those probabilities to people, particularly when talking about homeopathy. They hear "grain of rice diluted in the solar system" and all of the sudden the tiny probability makes sense to their minds. Sometimes you have to scale that probability appropriately with something they (the average pleb) can identify with. Such as the size of the solar system. Sort of like how people have a problem with talking about distances in meters, until I tell them that 100 meters is a little bit bigger than a football field, then they have it pretty damn well.

[Simon_Jester]

OK, but that's the form that works when you're in a position to deliver a quick lecture for the sake of explaining yourself. You don't always have that luxury.

[Starman7]

To answer the original question of "do they still have a place in science", I would say they might have a place teaching, giving ethics lectures, or reviewing for scientific journals, but I would never again trust that person to carry out a major scientific study.

Going off of the previous anecdote about a scientist who got trashed for a scientific fraud investigation vs. one who got off free on murder, I would say that murder is the greater crime (assuming the fraud doesn't lead to faulty products/devices which could kill people), but that scientific fraud is the less forgivable crime. Science needs to be 100% honest to live up to its name.