In science we propose, test, and debate theories. Typically, after loads of reading and/or observing, a theoretical model is built. Theoretical models describe the relationships between things in the universe and allow us to ask questions that can be tested (a.k.a., testable hypotheses). A testable hypothesis often takes the form of “changes in variable X (predictor) will result in changes in variable Y (outcome)”. We can then challenge the model by repeatedly testing the relationship between X and Y with multiple, random cases of X, different levels of X, X versus No X, X compared to X2, Xcetera…
If you can demonstrate that the change in variable X (predictor) is responsible for the change in variable Y (outcome) then you can be confident in the existence of a relationship between X and Y. In the physical sciences (especially in physics), the predictive power of variable X is often very high, if not, almost perfect. For example, if I’m on Earth and I throw a rock into the air (X), it will fall (Y), and VOILA! The theory of gravity is born – OK, it’s a little more complicated than that, but you get the idea. These theoretical models are so powerful that we call them laws because, given certain ubiquitous physical parameters, it’s unlikely they will ever be falsified, i.e., every test will result in the same outcome. Rock go up, rock fall down – on Earth…given the standard laws of physics.
The reason the social sciences (e.g., Psychology, Sociology, etc) get so much hate is because it’s very difficult to perfectly predict outcomes when dealing with phenomena as complex as human thought, behaviour, and social interaction. Sure we can hypothesize that, according to the central limit theorem, roughly 68% of students writing an exam will fall within one standard deviation of the class average. However, it is unlikely that I would be able to predict Suzy’s exact mark. Sure I could predict her mark using her past scores, social economic status, GPA, parent’s education level, and gender—and it would be a pretty damn good guess, but it won’t be perfect. There will definitely be error in my prediction because her performance on an exam is an extremely complex phenomenon. There are just too many variables involved each of which are complex themselves. A theoretical model for perfectly predicting exam scores (i.e., without error) for each student is impossible to create, or it’s beyond my level of education. The laws of physics, on the other hand, perfectly predict that when I throw a pebble into water, the surface of the water will ripple. Go test it. I bet I’m right.
But just because it is hard to predict outcomes when dealing with highly complex phenomena it does not mean that we should discontinue applying the scientific method to complex things, or that what we learn from such endeavours is not useful. I may not be able to predict Suzy’s exact mark, but educational psychology has taught us many things that can help improve our learning environments, and educators can apply and refine these methods using a scientific approach.
More importantly perhaps, the social sciences help us to debunk theories by demonstrating, undoubtedly, when a relationship does not exist between two complex variables. A great example of this is actually the topic of my Master’s thesis: Eyewitness memory. Decades of research have shown that eyewitness memories are not only fallible and unreliable, but are often wrong. This is further supported by the progress made in using DNA testing to exonerate innocent prisoners (The Innocence Project, 2014). Of course, this does not mean that every eyewitness account should be dismissed, but shouldn’t we question the reliability of an eyewitness if it means potentially avoiding a false conviction? Due to the increasing research in eyewitness memory and misidentification, courts have been ever more cautious about the weight they give to eyewitness testimony (Loftus, 2013). Therefore, in this case, knowing that X (eyewitness memory) does not predict Y (accurate culprit identification) has improved the human condition considerably.
I guess the reason we love to hate the social sciences is because of the way their results are often interpreted and communicated. The last discovery you probably learned from the world of Astronomy is that we’ve found a new “earth-like” planet in a nearby star system. YEPEE!! WE LOVE ASTRONOMY! But if you watch CNN, the last discovery you learned from the social sciences might have been that low blood sugar makes couples more aggressive (Source). Then you tested this model by thinking back to the last time you were really hungry and realized, “my partner and I don’t get more aggressive when we’re hungry – it depends on blah blah blah.” Quickly, you dismissed this study and labeled all social scientists as idiots. Also, nobody likes to believe that his or her behaviour can be predicted. Prediction is a close relative of predestination, and predestination undermines the importance of choice and free will. And we don’t like that. And we shouldn’t.
Most social scientists are very conscious of the fact they are studying extremely complex phenomena (citation? I hope so). We draw our conclusions carefully by looking at the magnitude of the effect in question, the generalizability of these effects, the limitations inherent in the methodologies being used, etc. We describe relationships between variables in detail and some of them are strong, but never perfect. Therefore, we try to avoid using terms like “caused” or “makes” (like in the above article about blood sugar). Funding is also limited in the social sciences (as it is in other fields of research), so the research agenda is limited to what the money can buy. That being said, quacks and frauds exist in the social sciences as they do in other fields, and pointless research abounds (especially at the graduate level because pointless questions are easier to do a thesis on than important ones).
The takeaway message: We can and should apply the scientific method to understanding complex phenomena, but those who choose to do so have a difficult task before them. It’s important that we, as social scientists, do not forget to apply sound scientific methods in our research. Moreover, communication and interpretation has to be done with a high degree of fidelity. Finally, we have to remember that the scientific method can be applied to any testable hypothesis, no matter how complex. So don’t hate the game, hate the players.
Loftus, E. F. (2013). 25 Years of Eyewitness Science……Finally Pays Off. Perspectives on Psychological Science, 8(5), 556–557. doi:10.1177/1745691613500995
The Innocence Project. (2014). Mission Statement. Retrieved March 27, 2014, from http://www.innocenceproject.org/about/Mission-Statement.php