["Journal for the Theory of Social Behaviour, Volume 56, Issue 3, September 2026. ", "\nABSTRACT\nThis paper offers a constructivist approach to the analysis of continuous variables in macroscopic social research. Scientific constructivists have well developed ideas about how categories are situated in minds as bounded regions in conceptual space. But they have largely ignored continuous variables, which cannot be treated as bounded regions in conceptual space. I address this omission and argue that continuous variables exist as one‐dimensional lines within conceptual space. Covariation between two continuous variables refers to an association between two lines in conceptual space, not a statement about worldly magnitudes that exist independent of human minds. One major implication is that, under most circumstances, continuous variables cannot support interval‐level assumptions or the use of standard causal inference techniques in macro research. Instead, scholars must pursue a research agenda centred on explaining patterns of covariation as products of shared meanings and leveraging continuous variables for probabilistic prediction rather than causal inference.\n"]