Tuesday, August 19, 2014

Big Theory

"Essentially, all models are wrong, but some are useful." --George Box*

I need to be honest with you, I'm not all that good at generating novel ideas: Some of my most well-cited papers involve theories that sociologists came up with decades ago; Reviewers frequently accuse me of running post-hoc analyses (asking the data for ideas, rather than generating apriori predictions); When media cover my research, the most common initial comment is something like: "This is so obvious....blah, blah....you suck." You get the idea. 

I don't view this particular characteristic of my research as a flaw. Rather, I'm acknowledging that not all scientists can be ground-breaking theorists/game changers: Some people come up with great ideas and some people test them. For the most part, I test theories and I do it in (what I hope are) convincing ways. Given this characteristic of my research, you might be surprised then, to learn that I love theory!** 

You read that right.

Some researchers might argue that theory creation is an easy step in any scientific enterprise and in principle this is not an incorrect statement. Literally anyone could generate a theory in social psychology by connecting two variables together with an explanation: For instance, messy environments increase stereotyping because they force people to use automatic information processing (a now retracted Science paper). It is this sort of theorizing that leads people to suggest that theory-making is easy--whereas data collection/interpretation is hard.***

I don't buy this argument for a moment. True theory--that which generates new ideas, argues them in compelling ways, and charts new directions for future research with precise predictions--though difficult to generate, can change a scientific discipline in fundamental ways. New theories can launch researchers into new areas of inquiry, lead to the testing of ideas in previously uncharted territories, and motivate a reexamination of old ideas.

When I think of good theory in social psychology, I think of the explosion of interest in cultural psychology following the work of psychologists like Hazel Markus, Harry Triandis, Shinobu Kitayama, and others. Up until the 1980s, researchers in social psychology were convinced that there were fundamental universals in human social interaction (i.e., the Fundamental Attribution Error): The realization that culture influences how we think about the self, what we value, and what it means to relate to others was not something the field acknowledged until after these theories were published. Generating theories of cultural psychology meant standing in opposition to universals that were held up by the field.

Perhaps the best thing that good theories accomplish is that they draw out good data. A compelling theory can inspire talented methodologists, publishing factories (what I call highly productive researchers), and new graduate students to test the theory across a number of different laboratories and methods. Without theorists (thanks Marx and Bourdieu!), I would be toiling along on less interesting topics of far less consequence.

Which brings me back to George Box's quote at the top of the page: "Essentially, all models are wrong, but some are useful." The quote reflects what must be a truth about even our best theories: They're wrongheaded but useful. Theories can't possibly be even mostly correct because if they were, social life would be much less complicated than it actually is (and we'd all be jobless). Social life is complex with individuals and situations contributing large amounts of random variation to all social interactions. Even the most compelling, eloquent, and sensible theories won't likely explain all that nuance without getting it at least partially wrong. But good theories will galvanize us to test them, and in that fashion, theories are truly useful!****

*I first learned of this quote via a twitter conversation with an esteemed colleague. Score one for social media!

**As a social psychologist, I couldn't resist the counter-intuitive opening to this blog post. It's the resurrection stone of the deathly hallows of psychological science.

***Data is hard.

****Simine Vazire's blog has the best footnotes!


  1. http://strangebeautiful.com/other-texts/popper-logic-scientific-discovery.pdf

    1. Thanks for posting the Popper piece. It's timeless and brilliant!