Tuesday, March 13, 2012

Science and Certainty

I got into a discussion on a LinkedIn discussion board and thought I'd share the exchange.

The other person wrote:
"There is very little absolute scientific fact in most of the scientific evidences referring to global warming. Science is only a study of things. I look for the most logical and rational scientifically determined evidences. I look for words such as "could, may, possibly, are believed to be etc. as red flags for possible deceit. The more sensational the phrase that comes after any of these vague red flag words, the more I sense there is deceit."

My response:

I understand good science to be something that is essentially statistical. It does not deal in absolutes. An honest scientist will rarely give an absolute conclusion, but instead will qualify the conclusion to respect the limitations of the evidence on which it is based.

This is why the reproducibility of results was a cornerstone of early scientific work. This is what differentiated it from philosophy or theology.

Chemistry is a good example of a pure science. Mix chemical A into chemical B and a certain result occurs. There appears to be a direct cause-effect relationship.

Now a scientist at some other research institution conducts the same experiment and achieves the same results.

Statistically, the conclusion becomes more certain, not ABSOLUTELY certain. With each repetition of the test the certainty increases.

This is what might be called "pure science".

Unfortunately a lot of research is done within a natural context that makes it impossible to achieve that level of certainty. It may not be possible to control variables (Economics). Combinations of elements may be too numerous for our ability to compute them accurately (Atmospherics).

Good scientists, working in those less than ideal contexts, will recognize that their conclusions cannot be stated as absolutely as test tube chemistry. They will qualify their conclusions. They will point to correlations that MAY be cause-effect relationships...

Of course, that kind of language, which may be totally appropriate for scientific work, also is more easily abused and misused, to deceive people, for whatever reason.

Somebody making the statement, "X causes Y" can easily be checked by outside sources, and possibly shown to be wrong.

Somebody making the statement, "In 90% of samples observed a change in X correlated to a change in Y...", may be making an accurate scientific statement, correctly reflecting the level of certainty of the evidence. Or that person may be dishonestly connecting 2 things that are in fact completely unconnected. It's a lot harder to tell which is which.

In general I think it's a mistake to expect absolute certainty in most scientific conclusions. Most science does not occur in the ideal conditions that you find in test tube chemistry.

I think it is critical for scientists to do 2 things, make their conclusions precisely reflect the level of their evidence, and openly acknowledge the limits of their evidence.

Many good scientists do this, but too often, when their work gets reported in the media the finer details and subtleties of their work are not communicated and instead, the media superimpose a level of certainty that is not appropriate.

Excuse me for making this so long but some ideas are just too difficult to compress into a few sentences. If you made it to the end, thanks for reading.