Probably every introductory science text in the world includes a first chapter section on the scientific method. Discussion of this approach to scientific investigation is so ubiquitous that it is easy to come to the conclusion that this is the description of how science works.
Unfortunately, this conclusion is not only erroneous, it also leads to some confusion about scientific investigations which don't follow the protocol of the Scientific Method.
So if this isn't the "way science is done," why do teachers and texts make such a big deal over it?
The answer is that, while science can be done (and often is) following different kinds of protocols, the description of the scientific method includes some very important features that should lead to understanding some very basic aspects of all scientific practice. Some of these would be:
With these purposes in mind, let's take a look at this hairy old icon of scientific process. Just what is the scientific method?
The scientific method is generally described as a series of steps. Though we speak of a scientific method, you will find that different sources list slightly different steps, though they all turn out to mean precisely the same thing. In your text, the list is:
This is a pretty good representation, so we'll go with it.
Scientific problem solving involves two basic types of reasoning, generally called induction and deduction.
Induction involves gathering together a collection of bits of data--observations, experimental results, whatever kinds of information are available--and formulating a generalization which reasonably explains all of them. This is analogous to the formation of a hypothesis. You make a set of observations, then hypothesize an explanation which accounts for all of the observations.
You can see why forming a hypothesis is sometimes described as forming an "educated guess." It's a guess in the sense that you are devising an explanation, but it's educated because (1) it must be reasonable (ie, sensible) and, (2) it either has to be consistent with what we already think we know, or it has to include a very good justification for deciding that what we think we know is wrong. This is a vital kind of self-policing. One of the most significant strengths of scientific knowledge is the degree to which it is self correcting, and this is one part of that. No matter how good an idea is, if it violates the centuries worth of hard won knowledge we've accumulated, there must be extremely good reason (based an a lot of evidence) to accept the new idea and throw out all of the old ones.
Deduction begins with a generalization. Predictions are made based on the generalization, and those predictions are challenged. This, in essence, is the testing part of science.
One can't say enough about how important this aspect of science is. "Testability" is a requirement for any useful scientific concept. But, of course, there are many, many different kinds of ways to test predictions. Some require laboratories and lots of expensive equipment, but many don't. History's scientists have demonstrated immense creativity in devising ingenious ways to challenge their predictions.
If you go back to that list of steps, you can see the first part of the scientific method is inductive; the rest is deductive. Most scientific investigations, whether they follow the scientific method's protocol or not, shift back and forth between induction and deduction.
One final note about induction and deduction. Read the descriptions of these two reasoning processes again, then think about Sherlock Holmes, renowned as a deductive genius. Note that the collecting of clues (observations) and the formulating of a suggested solution (hypothesis) is induction, not deduction. Somehow, though, "great inductive genius" doesn't have quite the same ring to it.
It would be impossible to adequately discuss the methods that scientists use to test their ideas, but there are some general categories and some important restrictions we can examine.
The most straightforward way to challenge a hypothesis is by taking the direct approach. This is often not possible, but we can use this as an avenue to look at some important restrictions on testing protocol.
As a rather silly example we can play with, consider that you have collected a bunch of observations about geese. You've combined the observations that (1) geese live around and swim in water, and (2) they have webs between their toes to hypothesize that webbing between the toes of geese improves the efficiency of their swimming ability. In other words, the webbing between the toes makes them able to swim faster than they would if they had feet like, for instance, a robin. This is the hypothesis you want to test. Note that you've already begun behaving like a good scientist.
Something else you need to keep in mind.
So now you're ready for your most direct test. Your hypothesis is that webbing between the toes improves the swimming speed of geese. It's not enough to simply toss a goose in the water and measure how fast it can swim. You need to demonstrate that the webbing improves its speed. So this leads us to a really important aspect of a well designed experiment like this:
So you can see that this testing stuff has quite a few limitations and controlling factors involved. Scientists are constantly on the lookout for flaws in their experimental designs. And for flaws in other scientists' as well, since every scientist is dependent upon the work done by everyone else in his or her field. This is another aspect of the self-correcting nature of science. Mistakes get made--but they also get found and corrected, because all work scientists do is scrutinized by other scientists.
What about other ways to test your hypothesis that aren't so direct? There are a lot, but a couple of kinds merit a moment's attention.
One indirect way to test a hypothesis like this is by using models. In the case of our goose, we could construct robot geese. Or we could reduce the problem even further, to consideration of the physics involved when the goose's foot cuts through the water, and confine our simulations to the foot and the water. And in this age of exploding electronic technology, computer simulation is an increasingly important aspect of experimental activity.
Useful as simulations can be, we need to keep the old computer programmers' adage GIGO in mind. GIGO is an acronym for "Garbage In, Garbage Out." Essentially, it means that the information you can get out of your system can be no better than the quality of information you put into it. In terms of our goose experiment, the more like real goose feet our models are, the more the results of our test will tell us about real geese.
One final area of testing needs a bit of attention. Our original observations about our geese were that they (1) swim and (2) have webbed feet. We could take a road trip and observe as many bird species as we could, looking for those that live in water, and those that have webbed feet. Then we can compare these two groups, and see how much overlap there is. In other words, we could check to see what percentage of water birds have webbed feet, and what percentage of webbed birds live in the water. This is called correlation, and can be a useful kind of information. A high degree of correlation suggests some kind of causal link between the two things we're comparing.
However, correlation data are very dangerous, and need to be considered with a considerable degree of skepticism. Correlation does not imply causation. In other words, you can't conclude that, since living in the water and having webbed feet have a very high correlation, one must have caused the other. It might be so, but the correlation can't tell you that; you need other kinds of evidence to demonstrate the causal relationship, if there is one.
A simple example can demonstrate this easily. If you were to collect a bunch of information about elementary school children from grades one through six, you'd be astounded to discover an extremely high correlation between mathematical ability and shoe size. If correlation implied causation, you'd be wracking your brain to figure out whether being good at math makes your feet big, or whether having big feet makes you good at math.
In reality, of course, neither of these is true. As children get older, they get bigger--including their feet. And as they get older, they also go further in school, and get more instruction in math. A sixth grader's feet will be bigger than a first grader's, and her math skills will be better, too. Not because one causes the other, but because both are affected by two other, related factors: age and amount of education.
Eventually, all the testing and experimenting produces something in the way of conclusions
This is one of the ways in which the typical depiction of the "Scientific Method" is most misleading. Publications of conclusions, etc., typically happen along the path of the investigation, since scientific investigation simply doesn't come to an end.
If we reduce real science to a simplistic depiction, here are the three kinds of things that can constitute the "Conclusions."
There are two vital pieces of information buried in this list of conclusions, and we'll wrap up this long essay by listing them:
© 2000 College of DuPage
Center for Independent Learning (630) 942-2185
Updated 13 Sept 2004