The Method of Science
What is science? At first glance, it should be easy to define. It is what scientists do in laboratories, wearing white coats, carrying test-tubes around. It is experiments, testing hypotheses and theories to see if the results are what the theory predicts. It is the whole industry of people delving into the natural world, studying it, experimenting on it, observing it, and coming to verifiable conclusions about what the world is and does.
It does not take much investigation into the practice of science, however, to start to be a little confused about what it actually involves. A scientist could be a person who follows tornados around and records wind speeds, goes to the bottom of the ocean in a highly pressurized diving unit, or never does anything more adventurous than look at a computer in an office. The theories a scientist works with could be chemical formulae, mathematical functions, or descriptions of prehistoric artifacts. They might concern times in the long distant past or billions of years into the future. A scientist might work on the scale of the sub-atomic or the galactic, with the digestion of an ant, the mating behavior of a buffalo, or the rate of change of a forest. Would an observer from another planet easily group all these studies into the one category? In what sense are all these different activities the same thing?
If the objects of study, the kind of theories, or the scale of investigation cannot be the thing that binds together all these activities, in an important sense the method of investigation can. Science still, essentially, works on the same principles described by Bacon centuries ago. What unifies all the different practices that go under the heading “science” is that its discoveries are meant to be based on collecting empirical data from the world. Once data are collected and analyzed, then theories can be pursued that might predict new ways of understanding how the world works. It is assumed that there will be widespread regularities that will explain on a general level what happens in the particular circumstance.32
In some issues, this is a relatively straightforward matter. What is the structure of an atom? At one time, atoms were thought to be solid and have no smaller parts. Through experimentation, that theory was shown to be false, and gradually a picture of an atom containing a nucleus with an orbit of electrons was developed. In time, after further experimentation, that picture was also discovered to be too simple, and much smaller parts again were discovered within the nucleus. Through a process of experimenting, theorizing, testing, and more theorizing, more and more things were and still are being discovered about the structure of matter.
For other areas of scientific study, however, it can be much more difficult to come to agreement about the proper scientific conclusions to be drawn. It can be very difficult to isolate which factors are the relevant ones that need to be studied, especially when things like human politics are affected by scientific conclusions.
To pick a currently contentious issue, it is observed that over time the average temperature of the earth is rising. That observation is based on data gathered in many different ways such as measurements from the atmosphere and oceans, from studying ice cores for changes in the way ice has formed over centuries, and even human records from past times. The observation part of science is painstaking and detailed. Any conclusion that the earth’s temperature is rising must be based on many, many observations in all sorts of different conditions. Only then can the conclusion be regarded as reliable.
What, however, explains the rise in temperature? That is another issue again. One theory is that human production of carbon dioxide, pumped into the atmosphere, causes global warming. Another theory is that the earth naturally goes through cycles of heating and cooling and that we just happen to be in a heating phase at the moment. Which of these theories we take to be true will have massive implications for public policy about energy production. If human activity is responsible for the rise in temperature, then we are in grave danger that the more we act in the same way the further the temperature will rise to the detriment of us all. If, on the other hand, we are simply going through a natural cycle and human activity has nothing to do with it, then we need not worry; the cycle will reverse again in time, and the earth will cool.
When a theory can have massive effect on whole industries, a great deal of pressure is placed on science to come up with a definitive answer. This goes against the nature of most scientific enquiry, which is slow, painstaking, and careful, gathering information and testing it within strict limits. The whole world system can hardly be tested in a laboratory. It is a very complex matter with a huge number of variables. Conclusions must be drawn with great care. It can be very frustrating for policy-makers that science works this way, but making sure that conclusions are strictly drawn from evidence is the very strength of science.
Science, then, at its most basic is a method for finding out how the world works. It is based on certain principles that inspire confidence in the knowledge it produces. It depends upon actual observations, noted, tested, repeated, and confirmed. It requires tests to be repeatable by other people in different conditions. Scientific conclusions are not meant to be based on the secret observations of one individual or on isolated opinions. Science is based on the idea that we can, in principle, work out knowledge that is objective because it is observed by many different people and tested in multiple ways. For many, that kind of knowledge is the only knowledge worth having. It is trustworthy precisely because it does not depend on who a person is, but how they carry out their task.
But can we be sure? How can we know that this way of understanding the world is reliable? Ever since science began as an activity, there has been a parallel field of inquiry that keeps asking such questions. We put a lot of trust in science in our modern, Western world, and the fact of our advances in technology seems to justify that trust. But how can we be sure? Throughout the eighteenth and nineteenth centuries, philosophers have pondered these questions because even if observations are made carefully and painstakingly, there is still the risk of error and uncertainty. Moreover, in principle there still remain certain insurmountable problems in being sure that empirical science is the way to true knowledge.
David Hume, for instance, pointed out the problem of induction: you may be able to frame general laws from multiple instances of something, but you can never be sure that the next instance will not contradict your law. Immanuel Kant proposed that we can never know for sure that our senses are telling us what the world is actually like, for all our sense data are necessarily “filtered” by the innate categories of thought we have in our minds. Despite this, however, science could still be held to be progressively approaching truth. For instance, William Whewell (who coined the word “scientist” to replace “natural philosopher”) claimed that confidence in science was justified on the grounds that disparate laws in separate fields could come to be explained by the one overarching law, just as Newton’s theory of gravitation had explained such separate things as the motion of planets and the tides on earth.33
The problem is, every time we try to tie down what it is that makes science a reliable way to find certain knowledge, problems crop up. The idea that science makes observations, then inductively constructs theories, and then tests the theories has difficulties at each of those points.
One Response: Falsifiability
The philosopher Karl Popper (1902–94) was one who sought to distinguish carefully what it was that made something scientific, and his solution has been widely quoted in scientific literature.34
Popper was struck by the way in which Einsteinian physics, a radically new theory that challenged the old, made extremely precise and “risky” predictions, such as that light from stars would be bent by the gravitational attraction of the sun, something that could be tested during a solar eclipse. This was tested in 1919, and the prediction made from the theory of general relativity was confirmed. This, to Popper, seemed to be something qualitatively different from, say, Freudian psychoanalysis which explained things only “after the fact,” not making risky predictions. It is easy to find confirmations of a theory if you make the theory all-encompassing enough. Confirmations mean something only if the prediction was unlikely or risky. There has to be a chance that the prediction be incorrect for the confirmation to be worth anything. Only if the prediction might not have come true is the theory being genuinely tested. What matters is not just that the theory is confirmed, but that it be in principle falsifiable. So Popper came to his famous conclusion that what makes a theory scientific is its falsifiability, refutability, or testability.
This general view has a lot of power, and Popper’s work has been condensed to a slogan of “falsifiability” which has appealed to many people as a good description of the essence of science. Whatever your field is—whether astronomy, biology, or archaeology—the important thing is that your theories should be such that they can be tested by the data. The real world must have the last say. Unfortunately, it is in the very practice of science that this description starts to falter. The problem is, scientists often do not just accept what the data say. In practice, what do you do if the experiment fails? Does this really prove your theory was wrong? Or was your experimental technique wrong? Or was the equipment faulty? Or was one of the auxiliary hypotheses wrong? For instance, if the experiment on Einstein’s theory had shown no bending of light around the sun, would this have proved Einstein wrong or the theory of optics on which the measuring equipment was based? It is possible to stay with a theory even if the experimental data are going against it at the moment, for the theory may have other strengths such as its coherence within the mathematical framework or its overall explanatory power.