I have to say, your assessment of the scientific process is incorrect. We are not going out to “prove” a hypothesis. See it more as an evidence gathering operation. We are supposed to be weighing information gained in an experiment, in the context of a hypothesis. Normally your null hypothesis is no difference between groups (usually control and treatment). When we get a significant difference, there is evidence to reject the null hypothesis, not that our proposed hypothesis is true. The null hypothesis still could be correct but in this instance, the way we have approached the test we have confidence to reject. When we continually test a system over and over, and we gather enough evidence against the null hypothesis, then we can conclude in fact that our proposed hypothesis is true, well as true as anything can be in sceince. I view the theory of evolution in this light. While still a theory, enough evidence has been presented that I can confidentially reject the null hypothesis (that evolution has not occurred or is not occurring). Now setting type 1 (alpha) and type 2 (beta) error levels is another big problem in sceince that needs to be discussed further.
1 Like