Ordinal Utility represents preferences through an ordering. Numbers assigned to alternatives can encode higher, lower or equal positions in a model; their differences and ratios do not by themselves measure the intensity of satisfaction.
A before B before C states how three alternatives are ordered. It does not say moving from C to B brings the same gain in satisfaction as moving from B to A. Distinguishing preference from measurable intensity is central even when a numerical table is used. [Mises — Human Action, IV.2–4] [Oregon State — Utility, 2.1]
For A, B and C, both 3, 2, 1 and 100, 20, 10 can represent the same order. A strictly increasing transformation of an ordinal function preserves rankings and ties. A number’s size, its double or the difference between two numbers therefore supplies no extra information about personal benefit. [Oregon State — Utility, 2.1]
Choosing A does not provide a complete ranking of everything someone might ever want. Interpretation requires knowing what was available and under which conditions; unchosen B might have been unavailable. Mises ties value rankings to actual action, not an independent unchanging table hidden in someone’s head. [Mises — Human Action, IV.2–4]
A standard model often assumes comparable alternatives and transitivity: if A precedes B and B precedes C, then A precedes C. A function’s form or indifference curves add structure. State those assumptions; one choice does not establish stable, complete preferences in every future situation. [Oregon State — Utility, 2.1] [Yale — From Classical to Neoclassical Utilitarianism]
A value of 10 in one person’s model need not mean the same as 10 in another’s. Each ordinal representation can be renumbered while preserving its ordering. Adding or subtracting such numbers alone therefore cannot justify an aggregate-welfare conclusion; interpersonal comparisons require additional rules and justification. [Mises — Human Action, IV.2–4] [Mises — Human Action, XI]
Transformation can change numerical differences and derivatives while preserving the ordering of alternatives. The shape of one ordinal function alone cannot reveal an amount of feeling. Choice analysis remains possible, but must use preference properties and explicit constraints rather than an arbitrary numerical scale. [Oregon State — Utility, 2.1]
In an Expected Utility model, outcome utilities cannot be renumbered by just any increasing transformation and then averaged over probabilities without potentially changing conclusions. Preserving this form permits a positive affine transformation a·u+b, where a>0. These are further model conditions, not proof of comparable happiness across people. [MIT — Expected Utility Theory]
Ranking wallets by security, convenience and price depends on the user’s goals and constraints. A score of 80 against 40 does not mean double the personal utility; a money budget is not a scale of feelings either. Explain criteria, feasible options and who assigned weights instead of presenting numbers as objective benefit. [Mises — Human Action, IV.2–4] [Mises — Human Action, XI]
For the clearest picture, read this entry together with Marginal Utility, Subjective Value Theory, Human Action, Opportunity Cost, Time Preference. The reverse links also lead from Subjective Value Theory, Marginal Utility, Opportunity Cost.