Sex/gender entanglement: A problem of knots and buckets.

When used as variables in biomedical research, sex and gender can be difficult to operationalize and measure. Questions have arisen about whether either category is stable or causally meaningful in a research context. Here, we discuss some of the limitations of using both or even one of these categories in correlational or experimental work. We argue that attempting to draw a distinction between sex and gender can reignite the nature/nurture debate, inadvertently bringing outdated metaphors and assumptions about innateness and causation into our research. Many researchers, including ourselves, have described sex and gender as separate collections of causal factors (which we describe as a "bucket" metaphor) or as entangled (a "knot" metaphor). Because they regard sex and gender as conceptually separable and internally consistent, such metaphors have limited value for understanding the drivers of diversity in our data. Rather than continuing to reify sex and gender as distinct buckets or threads of explanatory variables, we call for deconstruction of these categories by focusing instead on clearly operationalized, instantiating variables that researchers can manipulate or measure. Our proposed approach differs from recent, similar calls in that we are not suggesting the exclusion of a sex/gender category from statistical models; instead, we recommend keeping it-not as a representation of biological reality, but as a tool used under a careful set of assumptions. We provide example datasets to illustrate how a sex/gender category can, when thoughtfully operationalized, be used to improve statistical rigor and inferential precision. In addition, we advocate for attention to variation within sex/gender, which is more informative in investigations of mechanism than comparing means across sex/gender categories.