The Scientific Method

First, what is science? I like this definition by James Randi:

Science is best defined as a careful, disciplined, logical search for knowledge about any and all aspects of the universe, obtained by examination of the best available evidence and always subject to correction and improvement upon discovery of better evidence. What's left is magic. And it doesn't work. --

Scientists use a process to construct a realistic, repeatable, accurate model of the natural world called the scientific method. This method is used because scientists realize that personal beliefs and biases can influence interpretation of data and experimental results. This only works if you accept that there is an objective reality that is the same for everyone.

It works something like this:

1. Observation and description of a phenomenon.

2. Formulation of an hypothesis to explain the phenomenon. A good hypothesis contains a testable, measurable, plausible mechanism which may cause the phenomenon.

An hypothesis is a best guess, working assumption or prediction of what causes a phenomenon. Without doing an experiment the hypothesis doesn't explain anything. Automobile mechanics make hypotheses all the time in diagnosing what is wrong with your car. The hypothesis has to be plausible. Stating the car won't start because it's the first Tuesday in April doesn't cut it as an hypothesis.

3. Use the hypothesis to make predictions.

An experiment has got to be properly designed such that it excludes personal
bias and includes the proper experimental controls. You cannot include some
data obtained from an experiment and exclude other data from the same
experiment just because it does or does not support the hypothesis.

4. Test the hypothesis using experimentation and modify the hypothesis based on experimental results.

5. Repeat steps 3 and 4.

Proper controls are an absolute necessity in a properly designed experiment. The control and experimental groups must be identical in all aspects except for the introduction of a variable or suspected causal agent into the experimental group. Effects seen in the experimental group should not be seen in the control group.

Members of the control and experimental groups should be assigned randomly. Even better are randomized, double blind studies. In this type of experiment neither the researcher or the subject knows whether they are in the control or experimental group until the end of the study.