Reproducibility of Experimental Research

The purpose of any experiment is observation. But it is not a casual observation, loosely related or inconsequential. It is an observation made to use as evidence in the process of building up a case, that case being the hypothesis concerning the connected­ness between selected events of nature. As a witness in a court case is expected to give the same answer to the same question asked by the trial attorney, time after time, so the experimenter expects, from that part of nature that is “fenced in” by his experi­mental setup, to yield the same response to the same stimulus created in his experimental condition. The experimenter acts in a certain way in terms of his experimental variables, and he expects nature to react the same way, time after time, whatever that reac­tion may be. In addition, if the experimenter creates the same condition elsewhere and repeats his actions, he expects the reac­tion from nature to be the same as in the previous site of his experiment. Thus, the observation made by the experimenter is expected to be independent of both time and place.

There is nothing to assure the experimenter that this require­ment “should” happen based on some inevitable, more funda­mental principle, or law of nature. That the sun has come up every morning all these years is no guarantee that it “should” hap­pen tomorrow also. Instead, it is an act of faith, with which we all function. The experimenter, likewise, needs to proceed with this faith in repeatability, for without such faith, no experimental research is possible. The philosopher Mill termed this, without antecedent, the Law of Uniformity of Nature.

Somewhat connected, but in a different respect, is what is known as the reproducibility of experiments. Everything that is published is not holy. Suppose that after some time and in another place, another experimenter wants to make sure that under reasonably similar circumstances as those published, his actions in terms of variables and the reactions of nature toward those actions will be the same as he found reported. To the extent that this latter experimenter faces no surprises, the experiment is considered reproducible. Now, in addition to time and place, we may include the experimenter himself as the third element, from which the experimental observation should be independent. This stipulation is necessary for the growth of science, since, as we have said previously, science is a complex, but ordered, ever­growing structure of knowledge, involving the work of many people widely distributed in place and time. It is a self-imposed responsibility of each scientific worker to see that his work is reproducible, and only then is the work made public through whatever means. The experimenter finds himself to be a trustee.

In passing, we should note that in philosophically precise terms, nothing is reproducible in nature since every event is unique, and nothing ever repeats. “The moving finger writes and, having writ, moves on.” In view of this, words like “reproducibil­ity,” “repeatability,” and “same,” among others, are used loosely, and we tolerate, even accept, them. Instead, using words like “similar” or “identical” may be more proper.

Source: Srinagesh K (2005), The Principles of Experimental Research, Butterworth-Heinemann; 1st edition.