Theories and models: What they are, what they are for, and what they are about

In Fried (this issue), I argue that a lot of work in the social sciences generally—and in psychology specifically—reads like an exercise in statistical model fitting, and falls short of theory building and testing in three ways. First, theories are absent, which fosters conflating statistical models with theoretical models. Second, theories are latent, i.e. implied but not explicated. Third, theories are weak, i.e. ambiguous and impossible to test or reject because they fit any data. I focus on psychometric factor and network models and their applications to cognitive, personality, and clinical psychology, showing that selecting statistical models that impose assumptions consistent with the theories they are supposed to corroborate is necessary for bringing data to bear on these theories. Seven commentaries agree with some of the core challenges the field faces. They raise some important criticisms of the target article, and provide extensions by identifying further problems and potential solutions. Here, I aim to integrate some of the core points and criticism raised, and provide a brief primer on theory formation, structured into three sections: 1) what are theories; 2) what are theories for; 3) and what are theories about. This is followed by a section dedicated to the question 4) how to develop theories. I conclude with 5) specific obstacles to theory formation psychological scientists face, and how they can be overcome.