Zipf’s law applied to research

What makes excellent research? Here are some ideas: a great research question, reasonable and relevant data, rigorous methods, beautiful exposition, clear tables, well-cited and researched references, and clear implications.

Excellent research is not just about having a great research idea, but also superior research execution.

Most people think that great researchers are bright, maybe even brilliant, and have a set of deep technical knowledge that allows them to come up with an answer to important questions.

While this may indeed be a part of what makes a great researcher, my view, coming from observing some of the best researchers in the social sciences, seeing Ph.D. students succeed and occasionally fail, and observing my own process over 15 years, my view has evolved. Being a successful researcher often boils down to having a rigorous and well-thought-out research system. This is because research consists not just of doing one or two big things right or well, but rather doing potentially dozens if not hundreds of things — little things — well. Of course, having a great idea or a new theory is essential. Still, it probably isn’t the factor that will get your paper published in top journals. It is dotting the I’s and crossing the T’s.

In fact, maybe only 20% of what makes good research are these big things. Of course, they matter. But they are not enough.

The remaining 80% consists of lots of little things. Small things. The small things are learned little by little rather than through our doctoral education. We get advice from our co-authors, pick up little tricks in the process of doing our research work, learn from the papers of our peers, from our reviewers during the review process, and so many other places. For instance, how do you set up a project directory? What is the best way to organize code? How should we respond to reviewer letters? How should the introduction be structured? What does a good table look like?

Even beyond the more conceptual questions, there are essential implementation questions as well. For instance, how do you export a table from Stata or R into a LaTeX document? How do you note the presence or absence of fixed effects inside a regression table? How do you create sideways tables for very long tables? What is the best way to organize a reviewer response document?

This is akin to a furniture maker or an artist. While the work of an artist indeed consists of coming up with a new idea for a painting, the artist also needs the technical skills to execute that idea and produce art that matches the original concept. The carpenter may have an idea for a new chair, but implementing this into an actual piece of furniture will require a significant amount of attention to many, many, many details, including the cutting of each piece of wood, the joining of these pieces into a whole, sanding each corner, staining, and so on. Like the carpenter, our work also consists of tiny details that can separate what looks like an amateurish attempt at a chair from a study in craft.

My goal in this book is to distill down the art of social science research into as many little (often mundane) tangible pieces of craft that I have learned over the years. I hope to help you reduce some of the cost of learning the craft part of research to make it easier for you to produce beautiful research papers.