Nevertheless, the scientific method works fairly well in natural science. It is, however, less effective in social science and the humanity due to two reasons.
First, the research target is human itself. Each individual is unique. Experiments on one person cannot be exactly reproduced on another person, which violates the “repeat” scientific method. Furthermore, the same experiment may not generate the same result if it is applied to the same person at different times. Human may not be objectively isolated. The person may simply interact with the rest of the world (or have time to think about the experiment) and draw a different conclusion in the second experiment. Thus, many disciplines in social science and humanities turn into researching the characteristics of a group of people, or the common parts of people. Statistics are widely used in deriving useful conclusions. It is precisely because the characteristics of a group of people change much slower than each individual person in the group. Experiments can be meaningfully “repeated” in a short period of time.
Second, some disciplines, such as history, focus on past events. We have only one chance to do the experiment, that is the time the event happens. It so happens that some information may be lost as time passes by. Later generations may only base on the remaining information and guess the cause or reason. We can only do such experiments “repeatedly” if we invent time machines, go back in time repeatedly, and objectively evaluate the event without interfering it. It is however not likely under the currently scientific discoveries (mainly general relativity). Because we cannot repeat the event, statistically, the error of such judgment is one. Moreover, people may well have predetermined judgment before analyzing the event, and thus be biased.
The distinction between natural science and social science is somewhat blurred in quantum mechanics. In quantum mechanics, the observer cannot be isolated out of the equation even for physical quantities. The quantities change so fast1 that quantum mechanics has to rely on statistics and focus its average over time. Both due to the scientific method “repeat”.
I sketched the limitations of “repeat” in scientific method above. Below I will focus on another basic scientific methodology, which people more or less focus less. It is Occam’s razor. It is often stated as “All other things being equal, the simplest solution is the best.” This principle is often used to reason the non-existence of God[3]. If the assumption of the existence of God does not help explain the phenomenon, the assumption is cut out. The majority of the scientific community prefers to rely on explanations that deal with the same phenomena within the confines of existing scientific models[3]. This is the basis of Atheism: the scientific method cannot support the existence of God. Due to Occam’s razor, the God does not exist. Please note, however, if one day God can be fit in the scientific models, because of the same Occam’s razor, God may be assumed existence. But can God fit in the scientific models, even in the far far future?
1This is not an entirely accurate representation, but a more vivid description.
[3] Wikipedia: Occam’s razor.