1.4.2 Simplicity over complexity

Complex research can never convince anyone of something surprising. If you care about changing minds, then your research should be simple.

Social research in the digital age is often associated with complexity, such as fancy algorithms and sophisticated computing. This is unfortunate because the most convincing social research is often the simplest. To be clear, simple research is not the same as easy research. In fact, it is often much harder to create simple research.

The most important reason to prefer simple research is that it is the only way to create believable, unexpected results. For example, imagine that you have just conducted some research using an incredibly complex methodology. If your results match your expectation, then you will probably accept them. But, if your results are different from what you expected, you have two options: accept the unexpected results or doubt the complex methodology. My guess is that you are much more likely to doubt the complex methodology. This makes perfect sense, but it means that the more complex the methodology, the less likely it is to produce unexpected results that you will actually believe. At some point, methods can become so complex that the only results that you can believe are those that match you expectations. At that point, the research has lost something very important: research should be able to change your mind.

The problem that I’ve just described is ever more severe once you start trying to change someone else’s mind. Imagine presenting an incredibly complex piece of research that has an unexpected result to someone else. That other person has not spent months writing your code and working through your data so when they are faced with the choice of accepting the unexpected result or doubting the complex methodology, they are almost certainly going to doubt the complex methodology. If you care about convincing someone else to change their mind, then your research needs to be simple.

Simple research comes from a natural fit between question and data; in other words, good research design. Poor research design, however, leads to the ugly complexity that come from stretching your data to a question for which they are not well suited. This book focuses on two approaches to create a natural fit between question and data. First, this book will help you ask realistic questions of your data. Second, this book will help you collect the right data to answer your question.