This paper examines the implicit links between models containing ordinal variables and their underlying unquantified counterparts that are necessary to make the former viable theoretical constructions. It is argued that when the underlying unquantified structure is unknown, the permissible transformations of scale applicable to the ordinal variables have to be restricted beyond that which is permitted by dint of the ordinality itself. The possibility of an underlying structure being known but unspecified is also considered. In the case of the efficiency wage model, the only usable transformations of the ordinal effort scale are those which are multiples of each other.