Monday, May 4, 2009

Science - What It Is and Is Not

I too often encounter people who fundamentally lack an understanding of what science is, and this leads to problems in communication and understanding other ideas. Many people lump all kinds of modern technology and knowledge into their definitions of science, and that is woefully inappropriate. I meet middle school and high school students who tell me that they love science, when they actually just enjoy trivia about animals and volcanoes. Hippies (for lack of a better term) often accuse science of causing cancer and pollution and climate change. Some religious fanatics selectively cite a few news reports about some corrected or disproven theory or other as they proclaim that science tells us lies and misinformation, so we should ignore it. Science is not chemicals or information or beliefs or news reports or advice or a group of people.

Science is the method by which we determine the predictability of phenomena given specific conditions.

That's it. Science is a method that takes advantage of a number of experimental and statistical techniques to let us figure out the probability that phenomena Z will happen given conditions A, B, C, D.... As early as second or third grade we are taught the "scientific method" of 1) asking a question, 2) forming a hypothesis, 3) testing the hypothesis, 4) analyzing the data, 5) drawing conclusions (or some variation of these steps depending on your school district).

Example: 1) Does smoking cause lung cancer? 2) Since a true experiment with random assignment would be unethical, we will resort to some retrospective and longitudinal observations of smokers and non-smokers for X amount of years, and see who gets more lung cancer. 3) We collect the data. 4) (these numbers are made up) .03% of the non-smokers got lung cancer over X years, and 3% of smokers got lung cancer. 5) Since this is not a true experiment, the results are correlational, but we have a strong theoretical construct for causality, since there is no good data supporting the idea that a predisposition for lung cancer causes smoking, and we know from cellular studies that the chemicals in the smoke damages cells. So, smoking was associated with a 100x relative rate of developing lung cancer, which amounted to 3% of the smoking participants over X years. Further variation as a result of quantity smoked, age, sex, race, family history of cancer, nutrition, etc... should be analyzed by future studies.

In the example above, science was the method used, involving an observational technique, to determine the predictability (3%, or 100x relative risk) of an outcome (lung cancer) based on a condition (smoking for X years).

Some fields have good track records of coming up with theories with near-100% predictability. Chemistry and physics are good examples. We know how a whole lot of chemicals behave in given conditions, and they always behave that way in those conditions. Newtonian physics is pretty darn consistent for most of our purposes, but it threw us for a few loops at relativistic speeds, extreme densities, and a few other conditions. If you drop a ball, it will accelerate at the same rate and direction every time as long as you are in the same place. Quantum physics is a different beast.

Other fields have more trouble, usually because it is too hard to account for the thousands of conditions that contribute to the outcomes. Psychologists are happy to understand even a quarter of the variance in most phenomena. Medical doctors try to give the most probable diagnosis given the scientific research on various symptoms, then prescribe the most probably effective treatment (if they're knowledgeable of the research, though they can also fall back on whatever the drug company representatives bribed them to give). When we deal with probabilities below 100%, even our best practices are wrong sometimes, and that is where some people get confused and offensive, especially if they are accustomed to believing in things they were told are 100% certain, or are overwhelmed by complexity and ambiguity. We will be wrong sometimes, and that is not the fault of science, it is the fault of a complex world in which some things are affected by thousands of variables. People are especially troublesome with our many different genes and experiences (which can affect gene expression).

I have heard so many times, "Those scientists don't know what they're talking about! Science says one thing, then the opposite!" That is usually not what is happening. The media simplifies the often complex results of research, and people's brains simplify the news even more to make it easily memorable and usable. One day we find out that a food increases the risk of stroke; the next day we learn that it decreases the risk of colon cancer, but the complexity of the situation is lost in the simplification to "bad" and "good", and people start thinking inaccurate things about the process, what happened, and what science is.

No comments:

Post a Comment