Why your science needs a theory
I have worked as a scientist for about ten years now and I have gone back and forth on the usefulness of theory in all research. For a long time I was fascinated by phenomenological reporting. I didn’t want to ask a question as much as report what it is that was observed. I felt like this connected me to some objectivity that work grounded in theory may lack. As a social scientist who comes from a physics background, I found this intellectually stimulating. Social science theories often times can come across as murky and obtuse. They don’t always offer directly quantifiable measurements and can seem less rigorous. They sometimes lack any sort of quantifiable statement. It can begin to seem quite reasonable to throw out the theory altogether and simply report the numbers that I have gathered for the social system under study. I even had some idea that by simply reporting enough facts I could arrive at a theory by considering a lifetime of phenomenological reporting.
The problem with this line of thinking is that all data is gathered with some question in mind. Even if the data is sitting in a database somewhere never with the original intent to be used for science. Someone at some point said, “We have some use for this data so lets collect it like so.” This line of thinking also only really is sustainable if someone is doing all the data collection for you and you are simply mining it. Experiments can not be built without some driving theory. Neither can simulation. Eventually I came to the conclusion that every scientific problem needs a theory.
Many people are likely shocked that a scientist sat around and thought about whether or not they need a theory. It is obvious that science and theory are intrinsically linked. This is made clear in any introductory science textbook.But paper after paper is published that simply reports correlations. They report that they have found some connection between two or more variables, and lay it out in the paper. “We compared A to B finding that when A is greater B is also greater.” The issue is that without an explicit theory there is no explanation, no connection to any sort of grounding that can contextualize the work. There are no boundary conditions to describe how the results presented can and cannot be used.
It also should be noted that there is no scientific investigation devoid of theory. Instead the theory is implicit. In my field of physics education research there is a very famous paper:
Hake, Richard R. “Interactive-engagement versus traditional methods: A six-thousand-student survey of mechanics test data for introductory physics courses.” American journal of Physics 66.1 (1998): 64–74.
In this paper Hake demonstrates across many different instructional settings that if you teach physics using “interactive engagement” (this means students do things other than passively listen to lecture), then the students learn more physics concepts than if they sit passively in lecture. There is no explicit theory stated in the paper. There isn’t any sort of explanation as to why when you organize the classroom in an interactive environment students learn more than when you organize the classroom as a passive environment. Hake simply states that he believes students learn more if they do things instead of just listening. To be fair to Hake, Hake’s goal with this paper wasn’t to push forward our understanding of how student’s learn, but to change university education to benefit students. Hake’s goal with this paper was to shift policy as much as it was to shift our understanding of learning science. But because there is no explicit theory in the paper, we must implicitly gather what the outcome should be. We lack any sort of explanation as to why students learn more by doing except that our common sense would indicate this to be true.
All this being said, I like this paper. Hake presented a very clear result that had profound implications on the field. A great deal of subsequent work has been successfully built off of the result that Hake demonstrated with this paper. But because this paper lacks a theoretical grounding, you are only left with the results. Interactive engagement improves science education. But what happens when this is not the case?
In Computing Education there is a problem known as the “Rainfall problem”:
Design a program called rainfall that consumes a list of numbers representing daily rainfall amounts as entered by a user. The list may contain the number -999 indicating the end of the data of interest. Produce the average of the non-negative values in the list up to the first -999 (if it shows up). There may be negative numbers other than -999 in the list.
Created by Elliot Solloway, this problem seems to be a reasonably solvable problem for students who have completed a semester in programming. Except that students are very bad at solving this problem. Students don’t necessarily get better when you introduce interactive engagement techniques to teaching programming. Thus, while the result of Hake’s study may lead us to believe “just add interactive engagement to solve introductory science learning”, we quickly see that this doesn’t actually work all the time.
What would a theory by Hake look like? I dunno. I’m not Hake. My point with this article isn’t to generate a new theory of learning. It is to highlight how important it is to have a theory when presenting results. Theories provide boundary conditions on claims you can make with your results. A good theory, even in social science, provides testable and measurable claims. It describes the relationships between the different actors within the system you are observing and even will dictate what are actors and what is the system. A theory can describe latent variables that cannot be directly observed in the environment. In summary, your science needs a theory.
