Continuum progress of science is driven by people. It is the result of their curiosity, dedication, and creativity. In this perspective, although the scientific investigation itself in principle should be objective and rigorous, the truth that it reveals about the universe reflects continuous human endeavor and changing perspectives. It represents, to some degree, the manner how humanity sees the world and the degree of manipulation in materials and abstract ideas that a society is capable of achieving at certain moments in time.
There are, however, many ways to carry out an investigation. Some are more effective than others. Some likely will not work. My reflections on this matter over the years has allowed me to condense a few ideas of my own. These ideas are certainly subject to further refinements in time.
I think there are in principle two crucial stages of an investigation into how anything works. The first is observation. This must be done in care and in accordance to the evolution of the subject in its reference frame of space and time. This observation (or sampling in a sense) sometimes may also affect the subject, which further complicates the investigation, but nevertheless; this kind of disturbances should be kept to a minimal, effects always noted and in the end, some comparisons should be made to ensure objectiveness. In an abstract sense, this stage of the investigation is about the collection of data from the subject. From these data collected, one or more abstract models of the subject are constructed (the choices of these models may be initially random). This abstraction is based on previous experiences of similar subjects. In the case that there are no prior experiences, this will be the first collection to start.
Second, based on these abstract models, a second stage of the investigation may begin, which is experimentation. Experimentation is necessary and is the artificial perturbation of the subject with purposely designed goals. Experimentation must be done with comparisons and must have a set of controls to be compared with. Experimentation begins with an insight based on the already established abstract models, and in principle is done with a clear purpose to distinguish the subject from these models. It either verifies some of these models or isolates the subject from the set of many abstract models. Designing experiments require some creativity and intuition. Its reasoning takes into consideration many factors: available resources, tools, time and space allocable, and likely much more. Experimentation is fundamentally the only way to achieve the goals of an investigation, which is to obtain new or improve existing abstractions regarding how the world really operates.