Analytical decisions pose moral questions.

How often do we reflect on the potential moral or value implications - what is right, wrong, has value and is (in)appropriate - of seemingly trivial analytical decisions, such as how to dichotomise a variable? I argue that analytical choices relate to multifaceted and oftentimes challenging moral issues that scientists should take into deeper consideration. Here, I illustrate a variety of potential considerations about moral values, including issues like exclusion, marginalisation, autonomy, responsibility, non-maleficence in relation to various common analytical choices and practices, such as the use of thresholds for disease diagnosis or population definition, the use of composite measures in the context of clarifying effects, classification practices, decisions on variable selection, as well as decisions relating to (dis)aggregation of data. I discuss these examples in the context of reasonable theoretical or statistical reservations. I advocate for deeper engagement with the difficult moral implications of analytical decisions, and for a principled and pluralistic science, that is also a more robust science. Such a science can include diverse moral views through a coupled ethical-epistemic approach, sensitivity tests, multiverse analysis, as well as stronger commitments to participatory and mutual learning practices.