Is the ‘Medical Model’ Right for Education Research?

In “Straight Talk with Rick and Jal,” Harvard University’s Jal Mehta and I examine some of the reforms and enthusiasms that permeate education. In a field full of buzzwords and jargon, our goal is simple: Tell the truth, in plain English, about what’s being proposed and what it might mean for students, teachers, and parents. We may be wrong and will frequently disagree, but we’ll try to be candid and ensure that you don’t need a Ph.D. in eduspeak to understand us.

Today’s topic is the relevance of the “medical model” in education research.

—Rick

Jal: We recently discussed “evidence-based” practice and its limitations. That discussion almost always raises good questions about the comparison with medical research and what we might learn. So, it seemed worth talking a bit about the “medical model” and whether it has applicability to education research.

It is easy to see the appeal. You develop an intervention, you test it with Group A, you compare the results with matched group B that didn’t receive the intervention, and you see whether A has fared better than B. Compared with other forms of research, this method more clearly allows the researcher to see whether it is really the intervention that created the effect, as opposed to all the other potentially confounding factors. Appealing, isn’t it?

But the track record of these methods in education—and other social-sector fields—is, on the whole, not strong. Why not? There are many reasons, but let’s start with the most obvious one: The medical model of research is well-suited for answering research questions where one variable can be easily isolated, holding external interference to a minimum. When we do drug trials, we are essentially testing biological processes, and if we have a sufficiently sized sample, we can see whether a certain vaccine produces the intended results.

But if we are instead testing a math program, there are a lot more variables involved. Does the teacher understand the new approach? Does she believe in it or does she think the older way was better? Does she need to learn new math herself, and if so, are there opportunities to do so? Are the opportunities and professional development offered “high quality”? Additionally, there is the classroom environment to consider: What is the composition of her class, and how well are they responding to a new approach?

Social physics is a much more uncertain science than Newtonian physics. The result is that when new interventions are rolled out at scale, there is usually wide variance in results. And that’s because some people figure out how to make the intervention work in their contexts and others don’t.

Rick: I like where you’re going with this. While I come at the question a bit differently, I think we wind up in similar places. As I see it, the medical model is usually referring to the kinds of randomized control trials (RCT) used to test something like a new vaccine. But that’s an obvious oversimplification. After all, public-health research offers another version of the medical model, and it focuses on questions that don’t lend themselves to RCTs—like the health benefits of needle exchanges, seat belt laws, or masking.

Public-health research lacks the precision of a personal medical intervention, which means any benefits are more contingent on complex, tough-to-predict behaviors. Here’s what I mean: When we say that a vaccine “works,” we mean that a precise dosage administered under specific conditions is likely to have a predictable physiological effect for a particular individual.

When we say that a needle-exchange program “works,” though, we mean that making needles available in a community led to an overall improvement in health outcomes for a given population. But we rarely know precisely how the program was run, exactly who benefited, or what specifically changed due to the program. This means it’s tougher to know exactly why a program was successful or how to replicate its success.

Put another way, the challenge you describe is due to the fact that a lot of education questions are more akin to public-health research than to testing a vaccine. Will hiring more counselors help students? Well, it depends on who those counselors are, what they do, and how good they are at their job. The same uncertainty applies when it comes to teacher training, accountability, or school discipline.

Now, where I think we might disagree is that I do think there are instances where the classic, narrow notion of the medical model translates pretty cleanly to schooling. That holds when research concerns a targeted intervention administered to particular individuals under controllable conditions. There’s not a lot that fits there, but I would put certain elementary reading strategies and approaches to math tutoring in that bucket. What’s your take on that?

Jal: I do think there are situations where the model is more appropriate. In particular: 1) when the goals are widely agreed upon; 2) when the outcome measurement is clear; 3) where there is some significant underlying understanding of the basic processes involved; and 4) where the area is largely modular, meaning that you can improve it without needing to change other aspects of schooling. Both of your examples fit those criteria.

If we were more cognizant of the limits of the medical model, we might also be more open to the advantages of the variety of different modes of research. Case studies can help us understand the mechanisms by which things work—showing specific instances of what to do and how. Research can also frame problems or put new issues on the table, which in turn can lead to different action. Perhaps, as famous studies of research suggest, the most powerful effect of research in the long run is to shape the very categories in which people think.

One other idea worth considering is how the medical model holds constant features of the background which may be the most important thing to change. For example, if features like the knowledge, skill, and selectivity of the teaching profession vary across nations, then it seems reasonable to conjecture that almost any intervention that is teacher-dependent will vary correspondingly, working better in nations where teaching is more professionalized. But an RCT in one country can’t pick that up, because it takes the quality of the teaching profession as a given. And it may be at this deeper level of how the sector is constructed that really matters. What do you think, Rick?

Rick: I love that four-part definition. It should be on every superintendent’s wall, prior to their next vendor meeting. And great point about what we can hold constant when making comparisons.

As a political scientist, I’ve always been struck by studies which seem to presume that political behavior is consistent across time and place, without obvious regard to things like institutions, trust, or technology (kind of like a chemical reaction). There are plenty of scholars who discuss their findings with casual assurance, as if they were immutable truths. They aren’t, of course, and readers who don’t realize that can be led astray (see, for example, democracy building in Iraq).

When it comes to education, I’m left thinking that the value of the medical model is highly constrained. One takeaway is the value of embracing a full array of research methods, as you suggest.

But there is real value in the medical model, too, and I fear that too many education researchers have used the aegis of qualitative methods to excuse advocacy posing as scholarship. Indeed, large swaths of the education research community seem actively hostile to the kind of disciplined research implied by the medical model and contemptuous of narrowly tailored, outcome-specific science. The ethos of the American Education Research Association, for instance, seems to be that scientific inquiry is epistemologically suspect and that it’s just fine for “qualitative research” to serve as a wink-and-a-nod euphemism for tenured advocates to promote their political, social, and cultural agendas.

It was in response to this performative culture that Russ Whitehurst so aggressively championed the medical model when he launched the Institute of Education Sciences. While I think it’s fair to argue that IES may have embraced that model too tightly at times, it’s easy to understand that impulse when confronting an academic culture that rejects the medical model’s strict self-restraint because too many researcher-advocates prefer to feel unfettered.

Jal: I’m a lot more hopeful about qualitative research than you are, Rick. And I think if we spent more time listening to students and less running numbers, we would make more progress. But let’s leave that for a future conversation.