In my recent NSTA talk, I advocated a view of the scientific method that did not include the hypothesis. What blasphemy! I felt like Galileo speaking out against the Church or something. But let's face it, hypotheses are stupid and irrelevant for science in our modern age. At best, they are an artifact from the past that has long, long lost its purpose.
Now, I may have ruffled some feathers but I want to point out I'm not the only one - Douglas Llewellyn's session at NSTA, "The Role of Argumentation in Inquiry" session also threw hypotheses in the trash. Additionally this excellent compilation of quotes just published on brainpickings (many from actually scientists!) makes many of the same points I make here. My favorite view of the scientific method, over at Understandingscience.org, doesn't emphasize hypotheses either.
So what is my point? To summarize, I argue that there are three main reasons why hypotheses should not be a part of science education:
- They aren't used in all scientific disciplines equally, or at all.
- When used, they aren't a necessary part of the process or the focus (questions are the focus).
- Educationally, teaching hypotheses makes an otherwise intuitive process more formal and unfriendly.
The history of hypothesesLet's start with a little history lesson. Where did our fascination with hypotheses originate? It's not exactly clear, but it seemed to occur somewhere between the logical positivists (science philosophers who believed science was a series of logical steps) and Karl Popper (famous philosopher of science, noted for saying that science doesn't prove anything, it only falsifies hypotheses). After Popper, you're not doing science if you don't have a hypothesis. Or so many teachers and scientists claim.
Interestingly, hypothesis advocates are most common among biologists, and Karl Popper's greatest interest was in the biological sciences. So it would seems that was where he got his ideas from. Maybe if Karl Popper was obsessed with physics, we wouldn't have our hypothesis obsession. But we have what we have.
1) All sciences are not equalThere are 3 types of ways that sciences use hypotheses. Let's analyze them in turn.
Hypotheses can be...
a) ...neither necessary nor helpful. There are plenty of fields in which hypotheses just aren't used because they would serve no purpose. Much of physics comes to mind. Imagine quantum mechanics making a prediction on the nature of a particle's momentum and position. A series of equations are predicted, the measurements are made, and the data are fit to the proposed equations. The equations either match or they don't match. Where is the hypothesis? You could say that the hypothesis would be "the equations fits the data". But that hypothesis would be utterly unhelpful, and therein lies my point. Even if a hypothesis can be formed ad hoc, it's being formed ad hoc, meaning it clearly is not of any use to the physicist during their experiment. It's just something you tack on afterward, like a fancy bow taped to a present - all form and no function.
Another great example: I know an atmospheric geologist who has published papers on measuring various transfer coefficients in our world, such as the rate of transfer of CO2 between the atmosphere and the ocean. His results have been used in models and they are generally considered good, publishable science based around a very clear question (ie. "what is the coefficient of ..."). He could have formed a hypothesis about what sorts of values to expect, but that wasn't necessary nor would it have guided his research. You might as well just measure the value.
And last, there's always exploratory research, which investigates phenomena which we don't even know enough about to develop good questions or hypotheses (what lives in the deep ocean? What's buried beneath the ice on Jupiter's moon?). Exploratory research of course does not use hypotheses, but many people say this science isn't "real", or that experimental science "must" be hypothesis-driven.
b) ...helpful, but not necessary. A great many types of experiments fit into this category, in which hypotheses are a helpful guide to research. In these experiments, hypotheses help put your questions in context, but even in these then they are not technically necessary. In ecology, a field in which I have done quite a bit of research, this is generally the case. I suspect this is actually true for most other fields. Coming up with hypotheses can force one to think about what sorts of answers they'd expect to their question, and WHY they'd expect that answer. Answering the why forces you to put your question and results in context of other scientific knowledge, and thus serves a useful purpose. But although helpful, these hypotheses are hardly necessary, and the same contextualizing can be achieved in the questioning, without explicit reference to hypotheses.
c) ...necessary and helpful. There are some people who have claimed that hypotheses are absolutely necessary to their research. I am yet to be convinced that this is actually so, but even if it was, this is most definitely a small percentage of all scientists. I think there are some people whose research question is so broad that it cannot really be addressed until a specific hypothesis is developed, giving them something in particular to test. But I'd argue even in this case, its simply a matter of further refining their question.
For example, take the question "What is restricting the range of grey wolves?" Of course, there are many factors that could limit the range of grey wolves, so to answer this question, one may argue a hypothesis is required, like "mountains limit the range of grey wolves." But, I argue, that "hypothesis" is in actuality a new research question, one so clearly worded that the "hypothesis" is unnecessary and unhelpful.
|View of science||Without hypotheses||With hypotheses|
|Big question||What limits the range of grey wolves?||What limits the range of grey wolves?|
|Subquestion/hypothesis||Do mountains limit the range of grey wolves?||Mountains limit the range of grey wolves|
|Data||Grey wolf populations do not extend over mountains.||Grey wolf populations do not extend over mountains.|
|Conclusion||The range of grey wolves are limited by mountains.||The range of grey wolves are limited by mountains.|
2) Questions are the focus, not hypothesesI often ask fellow scientists, "who is your favorite scientist, and why?" The answer I get 90% of the time is that "I like so-and-so, and he always asks such great questions." Coming up with great research questions is what leads one to greatness in science, moreso than coming up with good hypotheses. This shouldn't exactly be surprising, since we've already established that not all scientists even use hypotheses.
I challenge you to find a scientific publication that does not include a specific reference to a hypothesis. Depending on your field, this may be more or less difficult, but rest assured you will find one eventually. Now try to find a paper that doesn't include a research question. I bet you might have a bit more difficulty here.
The point is that actual scientists tend to judge quality science through effective questioning, rather than effective hypothesizing. Hypotheses are really just expected answers to questions. If your question is well phrased, though, a hypothesis is so obvious as to be useless. You don't need a hypothesis to figure out the best way to answer a question. Educationally, we should be helping student's refine their questions to be more helpful and exact (as scientists do), rather than helping a student rescue a helplessly vague question by requiring a hypothesis.
What about null hypotheses? With no hypothesis, you can't run statistics, right? This is where Karl Popper comes in, because we can never prove our null hypothesis true, we can only prove it false. We fail to prove it false enough times, and we eventually sort of accept that the converse is true, right?
Wrong. Here's the funny thing about null hypotheses. We've created a funny set of statistics that tests whether data differs from a "null hypothesis" (meaning, nothing's going on). In a strict mathematical sense we can only prove this null hypothesis wrong, but never prove it right. And since Karl Popper's writing meshes with many scientists' understanding of statistics, this seems to work really well. The issue with this is twofold, though: first, we are inventing statistics that actually allow us to prove alternate hypotheses true, rather than fail to prove a null hypotheses false (Bayesian statistics, for those interested nerds). So, we shouldn't base our philosophy of science on a passing trend in statistics. Second, when we prove a null hypothesis false, that means something is going on, and we actually "prove" that our scientific hypothesis is true! This is contrary to Karl Popper's idea!
The take-away here is simple: the statistical use of hypothesis does not correspond to the use of hypothesis in the scientific method. So basing your philosophical understand of science on statistics will only read you down a road of paradoxes.
3) Hypotheses are a terrible educational toolScience is intuitive. Just look at our history: the first sciences started with the ancient Greeks, with astronomy. We've been doing science in earnest for at least the last 400 years. Yet, it is only in the last 100-150 years that we've really tried to define the philosophy of science. We then asked the question "What is the Scientific Method?" and as of today, we still don't have a complete answer. How can we have engaged in science for over 2000 years, yet still not have a complete, satisfactory description of that process? The only way this paradox is possible is that science is so intuitive that we can practice it without really even knowing what it is we are practicing.
If you asked me what science is, my answer is simple- the act of science is the act of answering questions about our world in a convincing manner, based on objective data. How we collect data and make a convincing argument varies greatly from field to field, and that's why we have so many different fields of science. But the commonality to all fields is the questioning and answering; it's simply the methods of answering that differ.
The main role that hypotheses play in education is to make the process of science less intuitive. They add an unfamiliar technical term and another step to science, to make it more of a recipe and less of an inquiry. To a beginner, the scientific process seems like an ancient ritual that must be practiced in just the right way or lest yours prayers for data go unanswered ("You forgot the hypothesis! You need to go back a step and get one before you can move on with S.C.I.E.N.C.E."). If we tell kids that science is just a way of answering questions about the world it's less intimidating to them because, well, they've been doing that for years! ("You mean people have found a way to do that better? I usually just pester my mom for an answer or use Google or Wikipedia. Tell me more!")
ConclusionI've had fellow scientists tell me that if your research doesn't have a hypothesis, then it's not science. When I tell them that my research doesn't have hyoptheses and explain my research to them, they usually say something along the lines "Well, if you think about it this way, then you can say you were testing this sort of hypothesis, so your research does have a hypothesis." As if there was an implicit hypothesis that I was never aware of, steering my research like a guardian angel and validating my science. But that's my whole point. If you can do a valid experiment without ever using or realizing you're using a hypothesis, then what is the point of the hypothesis? It's clearly not a vital component of the scientific method while we're doing it.
And so, if hypotheses aren't being used to guide scientists in their research, why do we expect them to offer any better of a guide for our children, who are learning the process for the first time? Why add another term and formality to an already overly formal process? Is that likely to make kids more eager to become scientists?