The 'scientific method' is the cornerstone of modern science. Pioneered by Sir Francis Bacon (1561 - 1626) as 'Inductivism', it is the underpinning of all knowledge in the physical sciences and is used whenever it is suitable in other fields of science1.
The philosophy of Inductivism has been attacked, with some success, by numerous scientists and philosophers such as David Hume (in 1748) and TS Kuhn (in 1996). However it has also been justified as many times by its usefulness.
The scientific method consists of four simple steps and, when used correctly, gives us an applicable understanding of some aspects of the universe.
The four steps of the scientific method are:
Observation - of some measurable aspect of the Universe.
Hypothesis (a guess) - of a property of the universe, based on observation.
Prediction - of something that should hold true if the hypothesis is correct.
Experiment - to test the prediction.
The steps can be repeated indefinitely and any repetition will either support, refute or modify the existing theory.
A very simple example could be that the scientist observes that something falls off a desk. The hypothesis - 'Things fall when not supported by other things' - springs to mind. From this hypothesis the prediction - 'If I drop something else, it will fall' - can be made. And finally the hypothesis can be tested by doing experiments to verify or contradict the prediction, ie, the scientist drops things.
In addition to the four steps, the scientist needs to analyse the prediction, the experiment and the results to ensure that they fulfil certain requirements, some of which are explained later in this entry. If the use of the method is acceptable, a decision can be made. In the above example, the experiments showed that of the things available for testing not a single one failed to fall when not supported by anything, so the hypothesis is published as the 'Theory of Unsupported Falling', including a record of predictions and experiments.
In science a theory is not the same as in common usage (ie, 'an unproved assumption'), but is an explanation of observable phenomena which has been tested and not found to be wrong, yet. Before a theory is accepted by the scientific community, the research should be duplicated by someone independent of the group or individual doing the original work. In the case of the 'Theory of Unsupported Falling', the independent group decides to try an experiment with a helium balloon and so proves the theory false. The hard work put into the original research is not in vain though. Those first observations and experiments, together with the experiment with the helium balloon can all be used as observations in a new round of using the scientific method. New cycles through the method could eventually produce the 'Theory of Unsupported Falling of Objects Heavier than Air on the Planet Earth', which would be confirmed by independent researchers.
Starting over again like this is the normal way to use the scientific method and one of its main strengths. The example also shows one of the most common developments of a theory, the narrowing of its scope.
The scope of a theory is the limits of where it is applicable. 'The Theory of Unsupported Falling' started out with the universe as scope. Testing it with something lighter than air or in a spaceship disproved the old theory and replaced it with one with a narrower scope, the falling objects had to be heavier than air, and the location had to be Earth.
The Scientific Method and Truth
Nothing is ever proven to be absolutely true in science and, together with the use of the word theory, even on something like the Theory of Gravity, which almost anyone would consider a truth, this leads many people to think science isn't the answer. This then annoys scientists who tend to forget that although a scientific theory works for all practical and theoretical purposes, and we're 99.99% sure it is correct, it can never be proven 100%.
Further muddying the water is the scientific use of 'law' as a synonym for 'theory'. A theory that has withstood all tests scientists have been able to conceive and perform for many years is often called a law. An example of this is The Three Laws of Thermodynamics. But although calling it a law makes it look like an assertion that it is 100% correct, there is still that tiny chance that new observations will prove it incorrect in some manner. The entry Lies, Damned Lies and Science Lessons shows how some theories that have been disproved still remain useful.
Human senses and human thinking aren't perfectly suited for science. The requirements of the scientific method are meant to compensate for this, so it is important to know what they are and the pitfalls involved when they are ignored.
The requirements for the initial observations are... none. The observations are what the initial hypothesis is based on and the process of the scientific method ensures that this hypothesis doesn't become an accepted theory without measuring up to the rest of the requirements. However, proper observations (see the section on 'Prediction and Experiment') are more likely to yield a useful hypothesis, and properly recorded observations and a logical explanation of how the hypothesis was derived from those observations makes it more useful to other scientists.
Hypothesis and Prediction
The hypothesis will, unless proven incorrect, become the theory and must therefore have certain characteristics. The most important characteristic is that it must be 'falsifiable', that is, it must be possible, by experiment or a new discovery, to show that it's inaccurate. The predictions must have the potential to prove the hypothesis wrong.
The hypothesis 'things fall' is falsifiable and therefore scientific; it can be proven wrong, as with the helium balloon experiment mentioned earlier. The hypothesis 'pigs can fly' isn't falsifiable; that pigs do not fly can be demonstrated again and again, but that doesn't prove that they can't, just that they, in the experience of the researchers, haven't yet. So the hypothesis 'pigs can fly' isn't scientific.
Does that mean science accepts pigs can fly? Only sort of, because the hypothesis 'pigs can't fly' is scientific2. It can be falsified by having a pig fly. And as long as all experiments show that pigs fall gracelessly to the ground, scientists will happily claim (with high probability) that pigs can't fly.
Prediction and Experiment
Humans are bad observers and anyone can be fooled by optical illusions or be tricked by inaccurate memories. Humans are also prone to wishful thinking and spurious logic. When formulating the hypothesis, this isn't necessarily a problem, but when making predictions, performing the experiments and evaluating the results, it is. The purpose of the requirements applied to the hypothesis, prediction and experiment(s) is to overcome these obstacles so that the truth (remember that nothing is ever proven absolutely true in science) behind sometimes flawed observations is revealed.
One requirement is that the predictions must support the hypothesis. With a simple hypothesis, this might seem obvious; with the hypothesis 'Things fall', the prediction 'Things will fall' is right there, but with more complex hypotheses it becomes difficult. Because humans aren't natural logicians, a prediction that almost, but not quite, fits the hypothesis can be difficult to spot. To help find errors it is important that the predictions are formulated in a way that is not ambiguous.
When performing the experiment, the most important requirement is that the observations should be of something that can be measured, and that the inaccuracy of the measuring device and the inaccuracy in reading off the result is taken into consideration.
Evaluating Experiments and Drawing Conclusions
Drawing conclusions isn't usually mentioned as a step, because it's supposed to be simple. If the prediction supports the hypothesis, and the experiment tests the prediction, and the result is positive, the hypothesis holds as a theory. Unfortunately there are a number of pitfalls along the way.
The most commonly encountered of these pitfalls are:
Using 'common knowledge'.
Ignoring unexpected results.
Accepting unexpected results.
Using faulty logic.
'Common knowledge' does not make experiments unnecessary. Making assumptions is something scientists can do at the hypothesis stage, and the results must be tested afterwards. The scientific method doesn't allow for taking anything for granted, except trusting the results of other scientists when they have been independently verified and accepted as a valid theory3.
What happens if one skips the testing to go straight to theory? Well, with a lot of luck the theory will hold, but if it doesn't you will look a fool4, and in some cases horror may ensue. For instance 'everyone knows' water puts out fires, but pour water on a grease-fire and steam blows burning grease everywhere, pour water on an electrical fire and you may electrocute yourself, and pour water on a fire with reactive metals and you get a violent explosion. So the safe bet is 'never assume'.
Not only are humans bad observers, their tools are inaccurate as well. This is a known fact and scientists try to compensate for it by knowing exactly how accurate their measurements are. This was mentioned in relation to experiments, but, regardless of this examination and scrutiny, errors sometimes creep in, and knowing this it can be tempting to disregard aberrant data and leave them out of the record. However they could in fact be an indication that the hypothesis needs to be changed. On the other hand, no-one wants to throw out a hypothesis because of background noise in the measurements, and there have been cases were accepting weird results led to new hypotheses and great debate, until it became clear that error in the measurements was to blame.
How do you know when it's safe to ignore the odd error, and when it's an indication of something new and exciting? Well, you don't, so make sure you record it, and check it twice, and then, if you think you can disregard it, mention why.
And finally it is time to shoot a hole in the 'Theory of Unsupported Falling of Objects Heavier than Air on the Planet Earth'. The scientists behind this theory took a shortcut through either circular logic or applied common sense. The question here is how did they define 'heavier than air'? Actually, they didn't, but they defined the opposite - 'lighter than air' - and very cleverly, too:
'Lighter than air objects' are those that, in presence of Earth and the atmosphere, do not fall5.
Why this makes the theory useless is left as an exercise for the reader!
The Case Against
If David Hume and other great minds after him have successfully attacked the scientific method, why is it still in use? 'Because it works' couldn't possibly have satisfied any serious opponents, could it?
The case against the scientific method in general, comes down to two main points:
Nothing is objectively verifiable.
Induction is not objectively valid.
Both of these are true and they are often used to argue that science is completely false. Those who argue thus often seem to be right, because some of their opponents forget (or never learned) these basic facts and argue from the point that the scientific method is objectively verifiable and/or objectively valid.
The first point is the easiest to deal with and the scientific methods approach has already been described. Nothing is considered completely true, the inaccuracies of the observations are estimated6 and the result is supporting evidence of a theory, rather than proof of it.
The second is trickier. The scientific method relies on observations on a few cases to hold for all cases, and for things that happened yesterday to happen tomorrow. Just because this has held true so far doesn't objectively mean it has to be true tomorrow. And if it isn't the scientific method isn't capable to deal with it, the scientific method cannot be used on itself.
The scientific method is used also when the basic rules cannot be followed. This means an even stronger case against can be made in those instances. This, however is too big a subject for this entry (a cheap cop out, yes, but true nonetheless), but an example may make the problem, if not the solutions, clearer.
Psychology deals with humans and human behaviour. A pass of the scientific method on a psychological problem could go as follows:
Observation - people today are more violent than before.
Hypothesis - action movies make kids violent.
Prediction - showing kids action movies should make them more violent.
Experiment - show kids action movies, see if they become more violent.
That looked simple enough and experiments like that have been performed, but there are several problems with drawing conclusions from such an experiment.
Observations and measurements are not simple. Determining whether or not people are more violent now than before is difficult enough and likely depends on interpretation of statistics. But drawing a conclusion means having accurate measurements of the 'violence-level' of kids before and after showing them action movies. 'Violence-level' is not a simple physical property than can be measured with an instrument with a nice pointing needle and a scale, the researchers have to invent both the scale, the method for measuring and the definition of what 'violence-level' is.
Creating an hypothesis is also a problem; human behaviour is often very complex, but the hypothesis has to be simple enough to be tested. Since the cause of violent behaviour in children is likely not to be due to a single cause like watching one Arnold Schwarzenegger movie, psychology often turns to statistics, which gives a less narrow view, and lots more opportunities for human interpretation.
The argument for the use of the scientific method here is perhaps just that it makes theories halfway comparable.