Equipoise and ethics in educational research

Abstract

Does the moral requirement that medical research comparing the effectiveness of two treatment methods be done only when there is community level equipoise also apply to research in teaching and learning comparing the effectiveness of two instructional methods? This article argues that it does. It evaluates three claims that the requirement does not apply to research in teaching and learning. One is the idea that the equipoise standard mixes up the ethical rules for practice with those for research. So it applies neither to research in medicine nor research in teaching and learning. The second is the idea that research in teaching and learning is different than research in medicine. The ethical basis for the equipoise requirement in medical research does not exist for research in education and so does not apply. Finally, the point is sometimes made that satisfying the equipoise requirement can be outweighed or more than compensated for by other factors when evaluating the ethics of research. For example, the knowledge gained about the comparative merits of different methods of teaching and learning might be so significant that it offsets any moral demand for equipoise or uncertainty.

Introduction

Imagine you are doing some research to find out whether one way of teaching a course or a particular topic in a course is better than another at promoting student learning. For example, suppose you are comparing a distance learning teaching method where there are no classes for students to attend with the regular teaching method where there are class meetings (Jalaluddin, 2019; Schunn and Patchan, 2009; Shacher and Neumann, 2010). Or instead, imagine you are comparing two methods of instruction in introductory statistics. One includes the use of computer-generated simulations to estimate things like the width of confidence intervals while the other is much more traditional and uses formulas for the same purpose (Chance et al., 2017; Lock, 2018). As a last example, assume instead that you are comparing the benefits of ‘massed’ instruction on and practice in a particular skill or technique with more distributed or ‘spaced’ teaching and practice of it by student learners (Carey, 2015: ch 4; Rohrer and Taylor, 2007). In your research you allocate one group of students to one of the two methods you are evaluating and another group of students to the other method. You measure their learning in some valid way that works for both methods of instruction. Then you tally up the results of your investigation and perhaps report these results to others at a conference or in a journal article.

What might make this kind of clinical trial research unethical, something you should not be doing from the moral point of view? One answer is that it would be wrong to do if you had not obtained the informed and truly voluntary consent of the students who are subjects in your research (Emanuel et al., nd; Firth, 2020: ch 8; Sanderson, 2010). Then, at least it is widely supposed, the research would be unethical. But suppose you had this; assume that this consent is needed and that you obtained it. Could the research still be ethically wrong to do with the student subject pool you have available for your research?

According to a common idea in medical research, at least if it applies equally to research in education, it also will be morally wrong to do this kind of research in many circumstances if there fails to be in the relevant expert community a state of uncertainty or ‘equipoise’ about the effectiveness of the methods being compared in the research (Freedman, 1987; Hill, 1963; London, 2017). In health care, the methods being compared for effectiveness are often different treatments for an illness. Does one treatment method produce a better rate of cure for a particular illness – for example, breast cancer or a serious COVID19 infection – than another? In medicine, the relevant expert community is the community of medical practitioners and, in particular, specialists in the field even if they are not themselves medical researchers. If the investigation compares treatments for breast cancer, for example, the relevant community is practicing and respected oncologists. The uncertainty required on this equipoise idea is that the members of this medical professional community must be collectively uncertain or undecided about which of the treatment methods being compared in the research is most effective. ¹ If the community firmly believes that one of the treatment methods is better than the other (or others) then there is no equipoise. Any research which compares the method that is believed to be better with another that is thought to be worse by giving some subjects the less effective treatment is unethical and should not be done. In teaching, the relevant expert community will be experienced instructors in the subject field even if they themselves do no educational research. If, for example, distance learning is being compared with classroom learning in the teaching of introductory mathematical symbolic logic, the expert community will be teachers of introductory mathematical or symbolic logic courses. The uncertainty required on this equipoise standard for ethical educational research is uncertainty in the community of symbolic logic instructors about which of these two methods of instruction will be more effective at getting students to master the relevant concepts and skills. If there is a generally held conviction that one method – perhaps it is classroom teaching – is better than the other it is being compared with in the controlled trial, in this case distance instruction without classroom interaction, then the needed equipoise is absent. Any research that requires giving some student research subjects what is thought to be the less beneficial teaching method is immoral and should not be conducted, even if the student subjects have provided their consent.

I will here consider three questions or issues about applying this equipoise standard for medical research to the ethics of research in education. ² One is the idea that it is a mistake to think that ethics ever requires equipoise or uncertainty concerning the merits of methods being compared in a research investigation, either in medicine or in education. It mixes up the proper ethical standards for practice with those for research. The second is the idea that while equipoise may indeed be a requirement for ethical medical research at least in some circumstances, research in teaching and learning is different. The ethical basis for the equipoise requirement in medical research does not exist for research in education. Finally, the point is sometimes made that satisfying the equipoise requirement can be outweighed or more than compensated for by other factors when evaluating the ethics of research. For example, the knowledge gained about the comparative merits of different methods of teaching might be so significant that it offsets the moral demand for equipoise or uncertainty. It is easy to imagine this happening with research on life-saving medical techniques or treatments like a vaccine preventing a serious communicable disease (Emanuel et al., nd). But perhaps it can also happen with research on teaching and learning methods.

Issue 1: Research isn’t practice

The first issue is whether the moral requirement for equipoise or uncertainty about the effectiveness of the different methods compared in a controlled trial research study – whether in medicine or in education – is totally misguided. The thinking here is that the ethics of practice in medicine and in teaching is different from the ethics of research in each of these fields. The equipoise requirement mixes these up, supposing that rules about what methods are ethically allowed in the therapeutic setting for health care patients or in the teaching and learning setting for students also apply to what methods may ethically be used on subjects in the research setting (Miller and Joffe, 2011; Miller and Brody, 2003). If this is correct, or so the reasoning goes, equipoise is never an ethical requirement for either medical research or research in teaching and learning. It is just plain misguided.

Certainly practice and research are different enterprises. For instance, they have different goals or purposes (Miller and Brody, 2003; Miller and Joffe, 2011; National Commission for the Protection of Human Subjects of Biomedical and Behavioral Reseach 1979). The aim or goal of therapeutic medical practice is improving the health state of a particular patient or group of patients, those a medical professional is now treating (Miller and Joffe, 2011; Miller and Brody, 2003; National et al., 1979). And the purpose or aim of teaching is increasing the skills or understanding of a topic for a particular student or group of students, those a teacher is currently instructing (Biesta and Stengel, 2016; Noddings, 2016; Scheffler, 1968 at ch 3). On the other hand, the goal of research into the merits of different medical treatments is generalizable knowledge about their effectiveness at bettering the health states of future patients, not, or at least not primarily, the individuals who are subjects in the present research (Miller and Joffe, 2011; Miller and Brody, 2003; Munson, 2012: ch 2). Similarly, the goal or purpose of research concerning the effectiveness of different methods of teaching and learning is also generalizable knowledge relevant to improving the skills and understanding of a field by future learners, not necessarily those individuals who are subjects in the current research (Dynarski, 2015; Mitchell, 2014: ch 1).

But what follows from this difference in the goals or aims of research and practice when it comes to ethics and the equipoise standard? Not much, either for research and practice or any other pair of enterprises. Different goals or reasons for two activities doesn’t always mean that they are governed by different ethical rules (Miller and Weijer, 2006a). The function or purpose of eating the food available at a feast is not the same as that of preparing the food for the feast in the first place. At least according to ethically motivated vegetarians, for example, it would be morally wrong to include meat both in the eating and in the preparation of the feast (Powell, 2016). This ethical requirement does not seem to mistakenly mix together the practices and different purposes of eating or consuming food with its preparation. Or for another example, the purpose or goal in performing a song is or can be different than the purpose or goal in composing it in the first place. Nevertheless there are ethical requirements these two activities have in common. Neither allows for a certain kind of lying or misrepresentation. Singers miming a recorded voice during a performance are doing something wrong if they present themselves as not doing this. And someone who claims to have created or composed a piece of music but has copied it from somewhere else has equally done something wrong of the same kind. The claim that this is an ethical failure in both of these enterprises doesn’t require mixing up one activity with the other. The root of the problem here is that the goals of enterprises or the different goals or purposes of different enterprises aren’t guaranteed to prove much about their ethics (Miller and Weijer, 2006b).

Now perhaps there is something else than their different goals or functions that gives the two activities of practice and research distinct ethical standards. Suppose this is correct. Suppose there is something X and it means that the ethical rules for assigning subjects to different medical treatments or teaching methods in research aren't the same as those for assigning patients different medical treatments or pupils to different instructional methods in practice. It still won’t follow that the basis for the equipoise standard in the ethics of research is always undermined by this point. Consider the kinds of cases the inventors of the equipoise standard for medical research were concerned about (Bradford Freedman, 1987; Hellman and Hellman, 1991; Hill, 1963). These are cases where the researchers were also at the same time physicians and the subjects in the research investigation were at the same time patients of those running the investigation. A relatively recent example of this is a study comparing the effectiveness of arthroscopic surgery with placebo surgery for the treatment of pain from osteoarthritis in the knee (Howick, 2018: ch 7; Moseley et al., 2002). The subjects used in the research were knee surgery patients of the principal investigator and continued to be so during the course of the research. For learning and instruction a similar thing often happens. There are many cases where the researchers were also the course instructors and the research subjects were at the same time their students. For example, at least some studies of the comparative effectiveness of various ‘intelligent’ computer tutors for LISP computer programming, high school level algebra, and analytic geometry, used students as research subjects while their teachers were at the same time part of the investigating team (Anderson, 2015: ch 9; Ritter et al., 2007). If in their normal practice setting physicians and other medical professionals are ethically required not to select certain methods of treatment for their patients because they are widely thought to be deficient methods then they are still required to do this when their patients become at the same time subjects in their own research investigating these treatment methods. If in their normal practice setting instructors are ethically required to avoid the choice of certain methods of teaching their subjects because they are widely thought to be ineffective then they are still required to do this when their students simultaneously are subjects in research they are running which investigates these flawed procedures. In other words, community-level uncertainty concerning the comparative effectiveness – or, rather, the lack of effectiveness – of the methods of treatment or teaching being tested in a research investigation is needed, at least in these kinds of cases.

But what about those occasions when the research subjects aren’t patients being treated by physicians conducting the research or they aren’t students being taught by instructors leading the investigation? This certainly can happen. In the field trials of the Salk polio vaccine in the early 1950s the research subjects were generally not patients being treated for polio or at risk of polio infection – or anything else for that matter – by those conducting the investigation (Meldrum, 1998). This is a medical example. Of course, it sometimes happens in the educational setting, too, that the research subjects are not students of the researchers. An investigation comparing training with two different representations of the important parts of a reasoning or argument text used paid research subjects who were not students being taught the topic by the investigators (Dwyer et al., 2012).

Even so, the existence of these cases is no support for the argument that the equipoise rule mixes up or illegitimately extends the ethics of practice to the ethics of research and, therefore, that it is always a mistaken rule. At best, they are instances where the ethical standards for practice and those for research, either in medicine or learning, are not both applicable at the same time.

Issue 2: Educational research is different

The second issue is that even though equipoise or uncertainty is ethically required of medical research at least in those instances where the subjects are also patients being treated by the investigators, research in teaching and learning is different. It is different in a way that means that community-level equipoise is not needed in research about methods of instruction even in those cases where the subjects are at the same time students being instructed by the researchers.

In health care it is often said that the equipoise requirement comes from a more basic ethical rule concerning the relationship between a medical professional and his or her patients. A physician or any other medical professional has a fiduciary or trust-based duty of medical care for his or her patients (Hellman and Hellman, 1991; London, 2000; Miller and Weijer, 2006). Medical professionals have to offer to their patients whatever treatment is believed by the relevant practice community to be best for their patient’s condition; this is the normally expected standard of care, When there is a community-level conviction about what is the best treatment or standard of care, any research study which assigns to some subjects who are also a physician’s patients a treatment which is not up to this standard clearly violates this fiduciary duty or trust (London, 2000; Mcleod, 2019: ch 2; Munson, 2012). When the research subjects are at the same time the investigator’s patients the only way out is to run just those research investigations where the practice community does not have a settled view about what is the best for patient care and treatment. This is exactly what the equipoise or uncertainty requirement says must happen, at least when research subjects are also patients being treated by the investigators.

If this fiduciary or role-related duty of care a medical professional has for his or her patients is the correct foundation for the equipoise requirement is there a similar fiduciary or professional duty of pedagogical care that an instructor has to his or her students? If not, there may be no equipoise or uncertainty requirement for research in teaching and learning. There would be no foundation or basis for the requirement in this domain. There are two ways of trying to answer this question. One is to look at statements that are explicitly about the relationship between a teacher and his or her students and, in particular, statements of course about a teacher’s duty to select certain instructional methods. The other is to look at more general discussions concerning the nature of fiduciary or professional-client relationships and then, if the relationship between an educator and student is an instance of this, to try to derive from these discussions something about the association between a teacher and a student and particularly about any moral duty an instructor has in choosing a teaching method.

While it is more about course content than methods or techniques chosen to teach this content, Strike and Soltis’s (2009) well-known textbook on ethics and teaching is an example of the first way of answering the question about instructional obligations (ch 6). They imagine the case of an instructor who believes that the subject matter she has been asked to teach will not be easily learned by her pupils. It is therefore, to her way of thinking, a bad choice. Among other things she defends her belief by pointing to widely accepted and well-evidenced views in the professional teaching community that this is a bad choice of material for pupils at the particular school level concerned. Soltis and Strike appear to accept that her professional duty as an educator is to take these settled judgements from the expert teaching population into account in selecting material to teach to her students. Other writers in the philosophy of education say similar things but specifically about the choice of instructional methods and not the choice of subject matter. For example, Harvard University philosopher Israel Scheffler (1968) says that instructors have a duty to choose pedagogical strategies that, at least in the judgement of the professional teaching community, have a reasonable chance of producing learning in their students (ch 4; Noddings, 2016: ch 3). So when there is a settled opinion of the professional teaching community about what is an appropriate teaching method an educator has a professional obligation to be following this in his or her pedagogical practice.

What about the more general approach to answering the question about a duty of care in the choice of teaching methods that an instructor has to his or her students? A fiduciary or trust relationship is ‘one in which one party (the fiduciary) exercises discretionary power over the significant practical interests of another (the beneficiary)’ (Miller, 2011: sec IIA). The relationship between a medical professional and his or her patients certainly is an example of this (Bernabe et al., 2014; Binik and et al., 2011; Miller and Weijer, 2006; Ryman, 2015 ). Medical professionals have the power or authority to select what will be offered as treatment methods for the medical conditions of their patients, obviously an instance of an important interest for a patient. They have a fiduciary duty to use this power to choose a treatment method that improves the patient’s medical condition and so advances the patient’s interests (Bernabe et al., 2014; Ryman, 2015). In making this choice, their discretionary power or authority is limited by what the professional community thinks is appropriate for the patient’s medical condition. Can similar things be said of the relationship between an instructor and his or her students? Instructors have to some extent the authority or power to choose the teaching methods they will use. At the university or college level, this power typically is much greater than at the elementary or secondary school level. Their choice can certainly affect the practical interests of their students in learning certain skills or material – or at least that’s what we suppose. Instructors are morally required to use this power to promote these material interests of their students rather than hinder them. But their discretion in the selection of instructional methods is restricted by what the professionally accepted standard of instructional care is.

So it looks very much like a parallel does exist between the situation of the researcher physician and the researcher instructor. If the foundation for the equipoise requirement in medical research is a physician’s continued fiduciary or trust-based professional responsibility for the medical care of his or her patients even when they become subjects in his or her research (Bernabe et al., 2014) then the basis for a similar equipoise or uncertainty demand for research in teaching and learning is an educator’s fiduciary or professional responsibility for instructional care, a responsibility that remains when students are subjects in his or her research.

But according to some, while there is indeed an equipoise requirement, it does not derive from this duty of care. The parallel duties of care exist, it’s just that the duty does not provide the foundation for the equipoise requirement, certainly not in medical research and perhaps not in instructional research either.

For example, medical ethicists Miller and Weijer (2006; Weijer et al., 2015) argue that community-level equipoise must exist for a controlled trial comparing two medical treatments for their effectiveness to be ethically acceptable. However it is because the subjects in such a trial trust that researchers will not be allowed by the state or the government or the law to run a controlled trial where community-level equipoise does not exist. Their trust is in the behaviour of their state authorities or the government and not that of their treating physicians. Subjects in medical research trials that are morally sound are volunteers; their free and informed consent has to be secured. Without their participation as subjects the research can’t be done and the valuable knowledge about the effectiveness the medical treatments being compared can’t be produced. In exchange, the state or the government is morally bound to ensure that subjects are not taken advantage of. This means they can reasonably believe that the expert medical community is not convinced prior to the research being completed that one of the medical treatments being compared in the trial is worse than the other. While it is not clear how this different foundation will support the community-level equipoise requirement in medical research, a similar story can be told for instructional research. ³ Subjects volunteering for controlled trials in education can similarly trust that the state or the government or some other authority – perhaps an educational institution hosting the research – will not permit a researcher to exploit them. But this can happen only if the research satisfies the community-level equipoise requirement, at least if Miller and Weijer’s reasoning about medical research is sound.

Alex London offers yet another basis for the equipoise standard in medical research (London, 2007). It derives not from the personal duty of care a medical professional owes to his or her patients but from the general moral requirement we all have to count equally the significant or important health interests of everyone, and so of course, those who are the subjects in medical clinical trial research. According to London, this general moral requirement means that medical treatments offered to anyone in either a practice or a research setting should not ever be anything worse than what the medical community would accept as competent or acceptable treatment. Will this line of reasoning support exactly the equipoise standard in medical research, as London thinks? It will certainly support an ethical rule which says that in a clinical trial comparing two medical treatments for their effectiveness neither of these treatments should be thought by the medical community to be less effective or worse than an acceptable treatment. On the other hand, the equipoise standard says that a trial should not include as an investigated treatment one that is generally believed by medical experts to be worse than another even if this worse treatment is still one that is acceptable. Something that would fill the gap here is the premise that if one treatment is believed worse than another then it is automatically an unacceptable or less than competent treatment. If the gap can be filled in this way, is there a parallel foundation for an equipoise standard in research in teaching and learning? If we all owe it to one another to count equally the interests many have in learning subject matter and skills that may be important or significant for our lives then instructors should never use teaching methods that are believed by the expert instructional community to be unacceptable or deficient. This applies in both teaching practice and research. So, given the gap-filling premise, it follows that a trial comparing two instructional methods should not include as an investigated method one that is generally believed by teaching experts to be worse than another. Two instructional methods may be compared in clinical trial research only if the relevant community of expert instructors is undecided about the comparative merits of the methods and they are both acceptable or competent methods of teaching.

Issue 3: Equipoise isn’t absolute

The final and third issue accepts that there is a moral requirement in both medicine and teaching to do only research that can satisfy the equipoise or uncertainty standard. But, or so it is said, this moral requirement can sometimes be broken or infringed when outweighed by other considerations. Sometimes this point is made using the terminology of ‘conditional’ and ‘absolute’ or that of ‘qualified’ and ‘unqualified’ moral rules or duties. A conditional or qualified duty is one that can be outweighed in certain conditions. An absolute or unconditional or unqualified rule or obligation can never be outweighed or trumped by other considerations. That this kind of outweighing can happen for other rules than the one requiring researchers to comply with the equipoise standard is a common point in ethics. For example, medical professionals and teachers have a moral obligation to keep certain information about their patients or students confidential (Kipnis, 2006; Scott, 2008). But this is often thought not to be an absolute rule, one that can never be rightfully broken. If serious harm is believed to result to innocent third parties from keeping confidential a specific piece of information then ethics may permit and even require medical professionals and teachers to break confidentiality and reveal the information (Bersoff, 2014; Bozzo, 2018). Confidentiality is outweighed or trumped by something more important, a moral obligation to prevent serious harm coming to others. But is there something that can sometimes or even always outweigh the moral requirement for equipoise concerning the assignment of treatments to subjects in a research project? The widely cited Belmont Report, for example, says that the principle of beneficence obliges us ‘to recognize the longer term benefits. . .that may result may result from the improvement of knowledge and from the development of novel medical, psychotherapeutic, and social procedures’ when considering the ethics of an investigation comparing different medical treatments (National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, 1979). Perhaps the obligations of beneficence can therefore sometimes outweigh the equipoise requirement, at least in medical research. And certainly it is hoped that educational research will similarly produce long-term benefits to learning. Could the duty of beneficence and the anticipated benefits to come from some research project trump the equipoise requirement in research in teaching and learning?

Imagine that community-level equipoise is not present for a particular research comparison of teaching methods T_A and T_B. That is, the professional community is sure that method T_A is better or more effective than method T_B. This why equipoise forbids doing any research which assigns some subjects to the inferior T_B. Of course, if the research comparing T_A and T_B is done it might be discovered that the standard professional opinion is incorrect and that T_A and T_B are equivalent in effectiveness. This as a matter of fact what happened when distance learning without classroom interaction was compared to classroom learning in the teaching of introductory mathematical symbolic logic (Schunn and Patchan, 2009). The expert community of teachers of introductory mathematical or symbolic logic courses was pretty much convinced that classroom learning would be more effective. But it turned out that the two methods of instruction were equal at producing student mastery of the skills involved. The conventionally accepted teaching method T_A could continue to be used, at least as an alternative to T_B. A switch to T_B would not increase learning. Of course, it could no longer be claimed to be better than T_B in terms of learning but no change in teaching practice is called for or required to benefit future students. So this research produced no long-term benefit, at least for learning. ⁴ In this case, the obligations of beneficence surely don’t outweigh equipoise and thereby make the research permissible.

But it might instead be discovered that the common expert opinion about the comparative effectiveness of T_A and T_B is the complete reverse of the truth. While the expert teaching community thinks T_A is better than T_B, it is actually T_B that is more effective at producing learning than T_A. This is what happened when the benefit of the ‘massed’ teaching and exercise of a particular skill or technique was compared with more distributed or ‘spaced’ instruction and practice of it by student learners (Carey, 2015: ch 4). While massed instruction and homework was widely supposed to be more effective, it turned out that the spaced or distributed system is in reality better. In this case, future students would benefit from a change in teaching approach to what was discovered to be the better instructional and skill practice method (Carey, 2015: ch 4). In this case, it might be thought that the obligations of beneficence to these future learners – many, many, of them perhaps – outweighs the equipoise requirement and makes the research morally permissible or even morally mandatory.

But this way of thinking won’t work, and for two reasons.

First, recall that the imagined discovery that teaching method T_B is as matter of fact better than T_A is a surprise. The widely held expert opinion before the research was done and the discovery was made was the opposite. Requirements for moral choice and behaviour are based on reasonable beliefs about what is true, not on what is perhaps later discovered to be actually true (Morrow, 2017: ch 4; Timmons, 2013: ch 1). So if the obligations of beneficence dictate anything then based on expert opinion about what the facts are they say don’t do the research comparison in this kind of situation. The research will, according to expert opinion, assign some learners a deficient method of instruction and disadvantage them without compensating benefit to any future leaners because it will not yield worthwhile useful results. Beneficence doesn't here disagree or conflict with the equipoise standard, let alone outweigh it.

Second, there is a problem with the thinking that parallels a familiar criticism of utilitarian moral theories. The benefits to the learning of future students gained by a change in teaching practice based on the imagined research results may in total be very large but are unlikely to be life-changing for any particular learner. In many cases, it will be small. The total is large on the assumption that many many future students will gain. What moral difference does it make that while the total benefit is large the benefit to any single individual is small? This is where the familiar criticism of utilitarian moral theories comes in. These theories obligate the infringement of important duties and rights when the resulting total benefits to the members of a population are great enough in total even if the benefit to each member of that population is very small (Hooker, 2014: sec 1). Imagine that causing or not relieving a long-lasting great pain to someone who has done no wrong or even a few individuals none of whom have done anything wrong will bring a very small benefit to each of millions of individuals. Here is a specific example of this from philosopher Thomas Scanlon (1998: ch 6):

Suppose that Jones has suffered an accident in the transmitter room of a television station. Electrical equipment has fallen on his arm, and we cannot rescue him without turning off the transmitter for 15 minutes. A World Cup match is in progress, watched by many people, and it will not be over for an hour. Jones’s injury will not get any worse if we wait, but his hand has been mashed and he is receiving extremely painful electrical shocks. Should we rescue him now or wait until the match is over? Does the right thing to do depend on how many people are watching – whether it is one million or five million or a hundred million?

Since the total benefit produced is overwhelmingly greater than the suffering of the person caused pain, utilitarian consequentialists are compelled to say that the right thing to do is to leave Jones in pain. But surely this is a mistake. Similarly if this is what the obligations of beneficence require then that is a mistake too.

Either the obligations of beneficence to future learners do not really dictate pursuing the research comparing T_A and T_B when community level equipoise is absent or it does and it’s mistaken in prescribing a sacrifice of the equipoise standard and the fiduciary duty instructors have to their current students which is its basis.