Assumptions are statements that are NOT supported by evidence. Use assumptions as premises only when NECESSARY. Identify and communicate assumptions you must make.
DEFINITION: Assumptions, or "beliefs," are statements that are not directly supported by empirical evidence.
For example, there are Three Main Assumptions of Science (TMAS):
1) The universe is real.
2) Humans can accurately perceive the universe.
3) Natural processes are sufficient to understand the universe
We do NOT have direct evidence for the TMAS! Debating assumptions such as "The world is real" may be appropriate for philosophy. However, instead of debate, scientists typically accept the TMAS and move on.
Scientists seek to understand the universe while making as FEW assumptions as possible.
A main goal of science is to create better and better models of the world that are based on as few assumptions as possible. Each assumption in a scientific model is a potential source of error, because assumptions can easily be incorrect. Ideally, scientists would be able to conduct experiments and build models using only the TMAS.
However, in practice, scientists are often forced to make additional assumptions. For example, a researcher may only have access to 5 males and 15 females in Group A, and 11 females and 9 males in Group B. The researcher may assume that males and females have comparable physiology for the specific question that the researcher is asking. Based on the assumption that physiology is the same for males and females, the researcher may compare outcomes of Group A to Group B despite the sex differences between the groups. Evidence from other studies can contribute to the conclusion that an assumption is reasonable.
Scientists inevitably make some assumptions when conducting research. Most experimental research cannot reasonably avoid (test) EVERY assumption required to perform an experiment. Moreover, scientists (like all humans) make assumptions that they are not even aware of (Gould, 1981). Therefore, purely descriptive or objective communication is simply not possible. The best that scientists can practically do is to identify the assumptions that they make and to honestly communicate all potentially important assumptions to their audience.