Imagine a puzzle that changes its shape and complexity depending on who is solving it. That is what selection bias represents in the world of research. Long a topic of fascination in epidemiology, biostatistics, population health, and social sciences, selection bias stands out as a particularly intricate challenge, often (arguably) more so than other structural biases (i.e., confounding and measurement bias). But what makes selection bias truly unique? It is the way it morphs and shifts across different disciplines. In this presentation, I will explore these disciplinary nuances in defining selection bias, trace its evolution over the past two decades, with a special focus on its evolution within the realm of epidemiology. I will highlight how emerging methodologies have reshaped our understanding of selection bias. It is anticipated that as we embrace new methodologies, our comprehension and interpretation of selection bias will deepen and mature.
