A strategy for building knowledge, common to the process of science in many disciplines, goes like this:
- We make observations of the natural world. (We often see patterns in these observations that catch our attention.)
- We then try to create plausible explanations (a model) to account for what we observe. (A "plausible" explanation is one that seems reasonable; it makes at least some sense based on previous experience. A good explanation would usually be based on well-established physical principles.)
- Then we use our explanation to make predictions about what we should observe in a new situation.
- Then we make observations in a situation like that new one.
- Finally, we compare the new observations with our predictions. (That is, we use the new observations to test the explanation.) If the new observations are close enough to (that is, "agree with") the predictions, then we say that the new observations confirm (support) the explanation.
However, new observations that agree with an explanation don't prove the explanation. Why not?
First, to prove a statement means to demonstrate that it is true with complete certainty, without any doubt. (Mathematicians, for example, use rigorous logic to prove abstract mathematical relationships, such as the Pythagorean Theorem.)
There are several reasons why new observations that agree with an explanation don't prove the explanation, and they have everything to do with degree of certainty. Here are a couple of important reasons:
Either way, we can never say with certainty that agreement between new observations and a prediction might not have been at least partly a lucky accident.
- There could be a problem with (error in) the new observations. (We can never be 100% sure about observations, and hence no scientific test can be 100% certain.)
- There could be another, alternative explanation that no one has thought of and that accounts for the observations better than yours does.
On the other hand, although finding observations that agree with your predictions does not prove your explanation, it does increase the likelihood that it is correct, and that helps us build knowledge.
In science, there are always alternative explanations for any set of observations, though some are much more likely than others. And even if there is more than one "true" explanation, they might not all be equally important. That's why, when we test an explanation and it is confirmed (supported), we still have to test alternatives (which might be correct too, or instead). With diligent work over time, scientists find that the likelihood of some explanations increases while it decreases for others. Eventually, the most likely explanations become widely accepted.
Reasoning from Evidence in Science: Some Terms
To prove a statement means to demonstrate that it is true with complete certainty, without any doubt. (Mathematicians, for example, use rigorous logic to prove abstract mathematical relationships. Unfortunately, in science, which relies fundamentally on observations of the natural world, proof is not possible.)
To disprove a statement means to demonstrate that it is false with complete certainty, without any doubt. (Mathematicians do this, too. Scientists can't.)
- confirm (or support)
To confirm or support a statement means to present evidence that is consistent with the statement, or agrees with predictions made from the explanation, thereby increasing the likelihood that the statement might be true. (It's not proven, though!)
- disconfirm (or not support)
To disconfirm or not support a statement means to present evidence that disagrees with, or is not consistent with, the statement, or disagrees with predictions made from the explanation, thereby increasing the likelihood that the statement not correct. (It's not disproven, though!)
Home |*| ANNOUNCEMENTS |*| Syllabus
Assignments, Labs, Quizzes, Handouts, etc. |*| Forecasting |*| Links