B The Hypothesis in Science Education

In an unsystematic pilot study, I briefly reviewed 13 high school-level science textbooks (7 biology, 4 ecology, 1 chemistry, 1 geology) to see how they cov­ered the Scientific Method and the hypothesis.

Perhaps the omissions are not surprising. Perhaps it is understandable that pre-college textbooks that cover a single scientific area, such as biology, focus on the factual content of that science or how the facts are gathered observationally, instead of being deduced from controlled experiments. Textbook publishers, it seems, may have little incentive to worry too much about broader issues of the philosophy of science. National, nonprofit educational organizations, on the other hand, should be above narrow commercial or social interests; indeed, they could help mold those interests.

A major source for national science education policy is the National Science Teachers Association (NSTA),⁴ which is the largest organization of pre-college science teachers in America. A special focus of NSTA training is on the Next Generation Science Standards (NGSS), which were developed by a consortium of national scientific societies⁵ including NSTA. The NGSS is intended to serve as a guide to teaching science from kindergarten through high school (K-12).

13. B.1 Teaching the Hypothesis at Pre-College Levels

Being a grade school science teacher must be one of the hardest teaching jobs. In addition to the usual classroom and curricular hassles, the science teacher has to stay abreast of the blinding pace of change in science and its technical dimensions—the websites, computer apps, and devices that affect access to scientific information—all the while keeping a weather eye on state and local regulations and being alert to societal sensitivities. And science teachers must be prepared for push-back from students who come to class primed to disbelieve their subject—resistance that few, say, arithmetic teachers have to counteract.

When it was rolled out 2013, the NGSS program was the first major over­haul of science education standards since 1996. NGSS is founded on the prin­ciple that students must be actively engaged with science problems, discovering and testing their own ideas—“Less Memorizing, More Sense Making!” NGSS identifies core themes in individual science disciplines, frames each theme in age-appropriate contexts, and links themes across disciplines via “cross-cutting concepts.” NGSS emphasizes what scientists actually do (e.g., “use and develop models”, “get evidence to investigate and test concepts,” “construct explanations”) rather than on what scientists know. NGSS recommends assessment of student achievement through performance measures, not multiple-choice exams that encourage empty “teaching to the test.” And it stresses that scientific knowledge is provisional, that it represents the best information that we have at the moment, and that it is constantly updated as new evidence comes in.

13. B.1.a NGSS, NSTA, and the Hypothesis

In NGSS, the concept of scientific thinking is taught through examples and is reinforced throughout all grades. While this attention is much needed, there are some areas of concern. For instance, while NGSS and NSTA correctly attach great importance to scientific words such as hypothesis, prediction, theory, law, etc., the essays on these topics lack unity.⁶ I'll skip over theories and laws because⁷ they are conceptually similar to hypotheses—conjectural, testable, falsifiable, and provisional generalizations—and because there was no consensus on how to define them. Most importantly for my purposes, theories and laws are either rare or absent from certain sciences (e.g., biology, neuroscience), or they have com­pletely different connotations in, for example, psychology and sociology than they do in physical sciences. In contrast, all sciences recognize hypotheses and predictions as valid and applicable concepts.

Unfortunately, the NGSS/NSTA philosophy doesn't take a definitive stance on what “hypothesis” and “prediction” mean or on the relationships between them. This is a problem because you can't fully comprehend the Scientific Method or what is involved in testing hypotheses without being clear on what these terms mean. Now at first, these look like simple, easily correctible oversights: define the words and get on with more important matters! Unfortunately, because of a fundamental NSTA precept that I'll get to shortly, it may not be easy to fix them.

13. B.1.b Definitions

Although the word “hypothesis” is mentioned several times in the NGSS pro­gram description, it is not unambiguously defined. As an example, one statement of high school objectives says that students should know that “A hypothesis is used by scientists as an idea that may contribute important new knowledge for the evaluation of a scientific theory”; another notes that, “Scientists often use hypotheses to develop and test their theoretical explanations.” So far, so good, but what is a hypothesis? The authors don't say, and the essays on it elsewhere in the program don't always agree.

One essay claims that a hypothesis is a “prediction with an explanation.”⁸ In addition to its confusing wording, this definition makes it hard to follow advice on how to conduct certain classroom exercises. A section describing a labora­tory exercise says that students are supposed to formulate their hypotheses “after the lab [while] doing the data analysis” and another instructs that during data analysis “is not when the hypothesis is to be formed”⁹ and requires that students

state their hypothesis before the exercise. Several articles explain that you make a prediction after you have a hypothesis, but these articles don't explain how the prediction relates to the hypothesis or clarify what testing a prediction has to do with the hypothesis.

According to one essay “A hypothesis is an educated guess about the cause of a phenomenon,” whereas “a prediction is an educated guess about the ex­pected result of a specific experiment.”¹⁰ This sounds fine and the authors lu­cidly discuss the nature of scientific thinking, even mentioning falsification. Then, at another moment, we read that, when it comes to hypotheses, “there's no guessing involved.”¹¹ You wonder how readers seeking enlightenment can put it all together.

A key principle in scientific thinking is how predictions are related to hypotheses. One NGSS-compliant program called Hypothesis-Based Learning (HBL)¹² seems to agree, noting that, “remarkably there is much confusion con­cerning definitions of hypothesis and prediction.” HBL says that a “hypothesis is an explanation,” that “a prediction makes the hypothesis a subject for rational thought,” and, finally, that a prediction is “given by compound ‘if-then' logic,”¹³ (i.e., if the explanation is true and such and such experiment is done, such and such results will be observed). This is consistent with the answer that I advo­cate (Chapter 1), although it is incomplete and, regrettably, is the last sentence of the essay.

Vagueness surrounding foundational principles may contribute to an­other problem in science education, as a vignette in Teaching for Conceptual Understanding in Science¹⁴ related by science educator Page Keeley suggests. Using an anonymous checklist of 14 possible definitions of “hypothesis” (six good ones and eight poor ones), Keeley queried a group of middle school teachers about their understanding of the term. Nearly everyone in the group strongly preferred two of the poorer descriptors: “an educated guess” and a state­ment “used to prove whether something is true,” and they were “adamant” in their choices. Keeley argues that teachers need to be aware that these phrases perpetuate students' misuse of words and, presumably, contribute to their mis­understanding of science. Though admittedly small and informal, the sample results suggest that some science teachers may lack the background to lead students to a nuanced understanding of scientific thinking. The NSTA website should be first and foremost a reliable resource for science teachers.

In short, the genuine value of NSTA/NGSS is diluted to an extent by an abun­dance of divergent views and an absence of direction as to which ones are most worth listening to.

13. B.1.C Off-the-Shelf Science Education

To help science teachers manage the curricular demands of NGSS, educational companies offer ready-made teaching plans that not only summarize scientific advances and principles and present topics and materials for classroom use, but also lay out guiding philosophies for teaching science. Argument Driven Inquiry¹⁵ (ADI) and Claims-Evidence-Reasoning¹⁶ (CER) are two examples.

Both ADI and CER teach that empirical evidence is critical in formulating and defending a scientific conclusion.

Given their promising premises, I was surprised to learn that neither the Scientific Method nor the hypothesis are even mentioned, let alone taught, in either program. The topics are not forbidden, as I was assured by an author of one of the program guides; individual teachers can bring them up if they want to, but the concepts have no official place in the curriculum. The written materials create the impression that an optimal outcome is a group of children who have all contributed to the discussion. In itself, getting all students involved is a highly commendable goal. Yet the message that everyone's point of view is as good as everyone else's can be a slippery slope. (It reminds me a little of the time when, in early grade school, my son returned from a sports-oriented field day with an eye-catching ribbon proclaiming that he had won a “Participant!” award.) In sci­ence, there are better and worse ideas, and the worst ones are discarded. If we don't acknowledge this, we foster societal debates that end inconclusively with a statement that “there are scientific supporters on both sides,” even when there are 97 highly qualified experts on one side and 3 vocal opponents on the other. Sometimes you have to make a decision in favor of one side, and the goals of har­mony and participation must give way to the critical thinking that takes place within the Scientific Method.

My concern is that, if neither national norms nor state educational agencies are clamoring for instruction in critical scientific thinking, then educational companies will not want to risk potentially rocking the boat by bringing it up. The short-term pedagogical gains may come at the cost of a scientifically less savvy populace.

13. B.1.d Don't Find “Fault”

Why can't we just tell the students what they need to know about the rigors of hypothesis testing, falsification, rejection, etc.? The problem is that, under the NGSS/NSTA aegis, we can't do that, at least in so many words. A central tenet of the NSTA educational philosophy is that science teachers must strictly avoid “fault words”¹⁸ (e.g., right, wrong, true, false). It's as if you weren’t allowed to call out “warmer” or “cooler” to the blindfolded child searching for the prize in the kids’ party game.

Perhaps to help students get over their anxiety about rejection and to be more willing to offer their own hypotheses, the NSTA/NGSS programs teach that hypotheses are never rejected or falsified, they are either “supported” or “un­supported” by evidence. Obviously, we don’t want to traumatize students, and it is true that working scientists often say that evidence supports or is “consistent with “their hypothesis. It is also true that, if enough evidence goes against a hypo­thesis, science throws in the towel, “unsupported” becomes “falsified,” “rejected,” or “wrong.” Avoiding “fault words” obscures the ultimate purpose of vigorous hypothesis testing and falsification, which is to get to the Truth as best we can.

13. B.1.e Abstract Ideals Have Concrete Significance

Are we supposed to avoid fault words because they allude to ideal states (e.g., Truth) that are not attainable? This, too, seems misguided. “At some point” in a scientific investigation the hypothesis must be rejected. Naturally, rejection won’t usually occur at the first sign of trouble. As I’ve discussed elsewhere in the book, scientific decisions can be complicated. The shining ideals of “Conclusive Falsification” or “Scientific Truth” may be literally unattainable, but the mere fact that we can’t attain them shouldn’t prevent us from teaching students about them. We don’t dismiss the ideals of “Honesty,” “Justice,” and “Freedom” simply because we can’t always achieve them.

Trying to account for scientific progress without calling on the concepts of right and wrong, true and false, etc., subtly endorses the myth that confirmatory evidence is the goal of science. This brings us back to the realm of inductive rea­soning and away from the search for scientific Truth.

13.B.1.f Right, Wrong, and the Real World

There are two additional reasons—one practical, one abstract—not to avoid fault words when talking about fundamental concepts. Practically speaking, students who enter the scientific workforce expecting their cherished opinions to be treated with kid gloves will be in for a rude shock, but they, at least, will eventu­ally get straightened out. The greater danger is that we’re educating the public to consider all hypotheses with some empirical support as being equally sound. If the standards of truth and falsehood are not in place, even as abstract goals, then the basis for making societal decisions that depend on scientific outcomes is on shaky ground. If we’re not taught about better and worse hypotheses, then we not equipped to judge the claims of science that affect society. And we’re more likely to believe that the mere existence of conflicting evidence indicates there is a stalemate and that no real-world decisions are warranted until “all” of the data are in.¹⁹ All the data will never be in, and everybody’s opinion is not equally valid.

In summary, many pre-college students may not be getting a realistic picture of the hypothesis or its place in scientific thinking. But perhaps this is under­standable; perhaps a little sugar-coating is necessary to make science palatable at that stage and is justified by the increased level of participation children show when they’re shielded from fault words. Are materials aimed at older students more honest?

13. B.2 Critical Thinking for Advanced Students and

Science Consumers

Among the resources relevant to scientific reasoning are the online sites I men­tioned earlier and books that deal with “critical thinking,” including scientific thinking. written for post-high school students and others. Many of the books on critical thinking advance the view that anyone can learn to think carefully and skeptically. Typically, scientific thinking, especially as it relates to the hypo­thesis, is somewhat of a tangent to their main thrust, which can range from dis­tinguishing science from pseudo-science,^20,21 to establishing the philosophical underpinnings of critical reasoning,²² to describing the skills needed for rational decision-making under uncertainty.²³ The hypothesis is one possible approach to critical thinking, and science is one arena of many in which critical thinking might be applied. Therefore, although these books make valuable contributions they aren’t part of the science curriculum.

Four books do cover the scientific thinking and the hypothesis in detail. The textbook by David Glass (Chapter 10A) is intended for university science students, and Stuart Firestein’s books (Chapter 10B), although not textbooks, are also intended mainly for post-high school science readers. As I’ve criticized the way these books cover the hypothesis, I won’t add anything here, except to state the obvious: they do not provide the reliable instruction regarding the Scientific Method that I, and I believe the respondents to my survey, are looking for. One textbook that does aspire to provide formal instruction in scientific thinking is the subject of the next section.

13.B.2.a Is Astrology a “Marginal Science?”: The Program of Giere, Bickle, and Mauldin

In their influential textbook, Understanding Scientific Reasoning, philosophers of science Ronald Giere, John Bickle, and Robert Mauldin²⁴ propose a step-by- step program to help post-high school students and nonscientist readers under­stand and evaluate scientific findings. The book is loaded with examples of how to apply their program to historical and contemporary scientific problems, and it offers study questions as well as advice for understanding scientists’ decision­making process. Principles of experimental design, including randomization, double-blinding, statistical practices, and decision-making under conditions of uncertainty, are covered in depth. Giere et al. are devoted to the proposition that scientific thinking is a skill that anyone can master with practice, and their book is intended for people wishing to acquire that skill, including college students. Their treatment of the hypothesis is somewhat abstruse and misleading, al­though that is a minor objection. There are more serious drawback to the pro­gram, which we’ll get to. First, the program itself.

13. B.2.b Models, Hypotheses, and Claims

For Giere et al., there are three kinds of hypotheses: theoretical, statistical (their statistical hypothesis is different from mine), and causal. All science depends on models which, they add, can be theoretical entities. In fact, “thinking scien­tifically about anything requires constructing a model.” A model depicts a real- world phenomenon, although a model is not the same as a hypothesis. Rather, a claim that a model accurately depicts a phenomenon is a theoretical hypothesis. In other words, scientists do not try to find out if a model is true; they try to find out if a claim that a model is true, is true. Imagine that we want to test our conceptual model that “Grass is green.” We would do experiments, not to discover if grass is really green, but to find out whether our claim that “Grass is green” is true. Only a philosopher could love such a distinction, and I suspect that it has never entered the mind of any experimental scientist.²⁵ In any case, Giere et al. do state that “all general scientific claims are hypotheses.” By comparing a model’s real or concep­tual properties, or its predictions, with the world, we can determine whether the model “fits” the world and, thereby, whether the claim about it is true or false.

13.B.2.C The Program

If you want to understand a scientific finding, according to Giere et al., you apply a six-step algorithm, which is modified slightly for the various kinds of hypoth­eses and decision processes that they evaluate throughout the book. This is how it works for the theoretical hypothesis:

1. Identify the aspect of the real world that is being studied.

2. Identify the theoretical model whose fit with the real world is at issue.

3. Identify a prediction based on the model to find out what data should be obtained.

4. See if data that bear on the model is available.

5. Ask if the data agree with the prediction. If the data do not fit the predictions, reject the model. If they do fit, go to step 6.

6. Ask if there is another plausible hypothesis that could explain the fit be­tween the data and model as well as or better than the present hypothesis. If there is no other hypothesis, accept the present one provisionally. If there is another one, either reject the current hypothesis or decide that the situation is inconclusive. (Emphasis added.)

Giere et al. consider the idea that science is distinguished by a Scientific Method to be “doubtful, at best,”²⁶ and they don't mention either Karl Popper or the principles of demarcation or falsification. They do say that science is meant to “explore how the world works” by doing experiments and making observations designed to “help [scientists] decide which of several possible ways the world might work is the most like the way it really does work.” They state that science tries to understand the world, though they focus attention on exploration and observation, which for them leads more-or-less directly to the creation and eval­uation of models.

This image, while reasonable in some respects, depicts science as a soft, largely passive endeavor that does not engage in aggressive searching for answers to pro­found questions. Scientists gather up data through experiment or observation and then match the data with model predictions; they never critically challenge a model, let alone try to falsify it. And any bit of data seems to be as good as any other; the main thing is whether the data agree or disagree with a model. Implicitly, the more agreement the better. Giere et al. point to a few caveats (agreement counts only when such agreement would have been very unlikely “if the hypothesis were clearly false”²⁷), but at heart their program prizes verification and confirmatory evidence as progress. They do not spend much time talking about how you choose between models, although they do bring up Expected Utility Theory, which we'll get to, and prefer a model that fits a collection of data better than does another model.

These weaknesses are perhaps debatable and, in any case, are not the real problem with Giere et al.'s philosophy. The inconsistency between tolerance and indecisiveness, on the one hand, and unreachably high standards, on the other, is a tip-off to its more serious drawbacks.

13.B.2.d Concerns About the Giere et al. Program Giere et al. feel that it is largely a waste of time to try to distinguish systematically between science and nonscience or pseudo-science. Certain human activities— religion, art—are obviously not science, they admit, but other fields are on a legit­imate fringe of acceptability. Their willingness to consider scientific claims to be undecidable, rather than false, leads the authors to accept Freudian psychology, astrology, extraterrestrial visitation, reincarnation, and extrasensory perception (including clairvoyance) as marginal sciences. For Giere et al., any field that uses “models to represent the world” and that makes an “appeal to empirical data” to support its “hypotheses” may qualify as marginal science.

If what we know about extraordinary claims is not convincing enough to warrant incorporating them into genuine science, or if is there is another plau­sible model for existing data, then Giere et al. insist that we must withhold final judgment pending more information. Not wrong, these claims exist in limbo awaiting stronger support. Evidently, we are supposed to ignore evidence that a “marginal science” (like astrology) has been rigorously debunked, discredited, or requires violation of the known laws of physics, as several of them do. The authors' reluctance to take a firm stand and declare these belief systems to be sci­entific nonsense is alarming and not only because they give aid and comfort to devotees of the supernatural.

In striking contrast to their readiness to accept astrology and reincarna­tion as marginal sciences, the authors place impossibly high demands on le­gitimate scientific hypotheses. Using “global warming” (anthropogenic global warming; i.e., man-made global warming—increasingly and more appropri­ately known as “climate change,” although we'll stick with their term to avoid confusion) and the “greenhouse effect” caused by atmospheric carbon di­oxide (CO2) as a case study, the authors consider two competing hypotheses to explain recent warm weather trends: (1) global warming and an alternative (2) natural temperature fluctuations, reportedly duplicated by a model that does not incorporate atmospheric CO2. Giere et al. conclude that the exist­ence of two plausible competing hypotheses means that we can't reasonably draw any conclusion; that we're obliged to hold out for “undeniable evidence” before making a judgment.

The net result of having the two widely divergent standards—one too lax and one too strict—is to expand the zone of undecidable “marginal” science. In ef­fect, Giere et al. sanction placing newspaper column astrology and rigorous cli­mate science into the same category.

And that is not the only area in which their program flirts with intellectual ir­responsibility. They do not mention the mass of evidence that is consistent with the man-made global warming hypothesis or that the overwhelming majority of climate scientists accept that this change is caused by human activity.²⁸ They do not discuss the different standards of validity that science adopts depending on the purposes of the information—basic or applied science—it gathers. They do not discuss the role that consensus plays in deciding scientific questions. However, the most pernicious fallacy in their argument is that the truths of sci­ence are determined by the criterion of undeniability. Remember, this is the standard that they set for man-made global warming, and it flatly contradicts their statement that all scientific conclusions are provisional. If a statement is provisional, it is “deniable” at some level.

Much of Understanding Scientific Reasoning addresses scientists’ and policy makers' obligations to act despite being faced with uncertain facts. The authors advocate using Utility Theory (i.e., Expected Utility Theory; see Chapter 12) and again they use man-made global warming as an illustration. To apply the theory, you have to assign values to variables, such as the cost of taking action (e.g., trying to stop global warming vs. doing nothing) and of different outcomes (e.g., global warming occurs or it doesn’t) and the probabilities of each outcome happening. You plug these values in a simple matrix of costs and probabilities, and choose the option having the greatest expected value²⁹. However, Expected Utility Theory calculations are meaningless unless they’re based on realistic assumed values. The expected utility of doing something depends on both the cost of the outcomes and the chance that the outcomes occur. You multiply the apparent cost times the probability to determine the expected utility of an ac­tion. Hence, even an outcome with only a tiny probability of happening could be worth avoiding if the possible costs of inaction are high enough. You may believe that it is extremely unlikely that you’ll die at a young age (the proba­bility of your early death is very small), but if your spouse and children would suffer greatly without your financial support (the potential cost of your dying is very high) you may decide to buy life insurance even though you’re young and healthy.

When it comes to man-made global warming, Giere et al. cite “consider­able controversy among scientists,” effectively tabling the question of whether it is even real or at least making the probability of it’s happening appear rela­tively small. The worst costs of man-made global warming that they entertain are associated with “flooding of major cities” and “the destruction of farmland.” This sounds as if it could be fairly bad, but the vague and limited assessment of damage seems to diminish the likely costs of global warming happening. Thus we have a dubious and not especially high, probability that global warming is real coupled with costs that are ill-defined, though perhaps manageable; in sum, there is probably nothing to worry too much about.

How would their example change if they took into account actual evidence that “around 95% of climate researchers actively publishing climate papers”³⁰ agree with the conclusion that global warming is being caused by human ac­tivity? This would tend to make the probability of global warming seem sig­nificantly higher. How about the total costs of global warming? Giere et al. do not consider, let alone assign values to, potential costs associated with mass human starvation, worldwide strife, or widespread animal extinctions that could result from global warming. Such costs are likely to be extraordi­narily high but hard to assess in the abstract. A more tangible assessment was provided by the administration of US President Donald Trump, not known as a bastion of liberal nervous Nellies on global warming, which released a US government report that estimates that the costs of global warming to the US economy amount to hundreds of billions of dollars, perhaps as much as 10% of the US gross domestic product, by the end of the century.³¹ This is a genuine prediction that the American people will have to pay significant out-of-pocket costs.

Rather than stressing the requirement for realistic estimates of probabilities and costs, Giere et al. breezily conclude that “What would be required to make the decision process [regarding man-made global warming] look different to policy makers is undeniable evidence both that doing nothing will lead to warming and that doing something will prevent it [emphasis added].” Perhaps the authors were being ironic and meant to imply a distance between what they themselves believe and what they anticipate that “policy makers” think? It is not clear. What is clear is that they do not call attention to the fiction of “undeniable” evidence, and their omission takes on ominous overtones, as they conclude that, without undeniable evidence for both positions, the only rational decision regarding global warming is to do nothing.

I do not know whether policy makers have read or been influenced by Giere et al.'s book, but it has clearly been read by many ordinary citizens. The discus­sion of man-made global warming is a perfect example of how something as seemingly minor and esoteric as a mistaken view of scientific reasoning could have far-reaching real-world consequences. Collectively, citizen opinion of sci­ence has a huge effect on how seriously we take the alarms about global climate change. Citizens who are poorly informed about how science works cannot make the best decisions.

13. B.3 Suggestions for Improving Science Education About the Hypothesis³²

1. Scientific educational organizations should develop and promulgate con­sistent core definitions of the key terms and principles of scientific thinking, especially those related to hypothesis, prediction, and testing.

2. Remove “fault words” from the taboo list and instead describe the process of science and explain how the fault words fit into the quest for scientific progress.

3. Stress that it is not wrong to have an idea that turns out to be wrong; that science progresses by replacing worse ideas with better ones.

4. Distinguish between the objectives of applied science and basic research science and emphasize that applied science is guided by the best scientific information available, while basic (“pure”) research seeks to achieve the ideal of Truth.

5. State that the goals of pure science to have all of the data and to achieve Truth are ideals that inspire us to understand nature to the greatest degree possible, even though we cannot actually achieve perfect knowledge.

6. Explain that it is a mistake to judge the actions of applied science by the goals of pure science.

13.