What is an experiment?
-
"An experiment is a test under controlled conditions that is made to demonstrate a known truth, or examine the validity of a hypothesis." (From a
![[WWW]](/wiki/classic/img/moin-www.png)
web site at Drexel by Brandon C. Beltz, Edna Monroe, and Denise Williford.) This definition is fine, but not so useful for our purposes. Let's elaborate.
-
An experiment is an alteration of a system presumed to be (otherwise) stable, and observation of the results. This requires a control -- a copy of the system that hasn't been altered -- so we can compare the altered system and the control system.
-
For an experiment to be any good, you have to have a well defined hypothesis, and you have to have some kind of plan for evaluating the results.
-
In mathematics, the definition of experiment might be broader.
-
What's not an experiment but still empirical?
-
Just making observations or measurements of unmanipulated systems.
-
Why do we do them?
-
way of examining all variables systematically
-
variables examined in isolation can be seen more clearly
-
experimentation lends itself well to formal or rigorous falsification -- experiments are all about falsification
-
You can't falsify things you create, and something something
-
How do they do what we want them to do?
How are models like or unlike experiments?
Oops, maybe we should be talking about how we use models instead of about the models themselves. After all, we're not comparing some Netlogo sim to the stuff in the test tube. We're comparing it to the whole process of putting stuff in the tube and adding something and watching the color change.
-
unlike
-
Models are a simplified version of nature; the experiment is nature itself. So the thing you're changing might not affect as many other variables as in nature.
-
Experiments
-
Experiments are definitely (generally) about falsification. Models aren't always, depending on how they're used.
-
like
-
You can do an experiment (as defined above) on a model just like you can on nature.
-
They both give answers or conclusions -- or they can both falsify hypotheses -- based on observations of a system's behavior.
-
They're both structured on (limited by?) the scientist's notions of what variables are or aren't important. That is, they're both made of variables, and variables are defined by the scientist.
Can models substitute for animal experimentation? For nuclear testing?
-
Models might make sense for education -- reducing the number of viviseted frogs in high-school bio classes
-
But what about for research?
