The construction of a set of scales is delineated, for evaluating the performance of social agents (e.g. providers of services as hospitals, schools, etc.) condition- ally on “reference states” x := X ∈ {x 1 , . . . , x R } of the governed individuals. Each scale is associated to an index which uses conditional “worthiness increases” ω l|x , between the levels of an ordinal outcome indicator Y := l ∈ (0, 1, .., L). This indi- cator was been defined on a scheduled, by the policy-maker (PM), chain of hier- archically ordered goals. The “worthiness increases” are interpreted by modeling interrelated latent evolutionary processes, on the scheduled goal chain, up to hyper- parameters γ which are driven by conditions x. Then, to standardize the set of scales on a given “reference behavior”, a pseudo-Bayesian (see [1]) method is used which elicits value γ ∗ by minimizing “residual from updating” (see [4]). It norms the model specifications on the “reference data” of the (chosen a priori) “standard agent”. Fi- nally, adhering to general requirements in rational choices from the decision theory, a standardized worthiness-based index can be implemented, which takes into input the agents actual data
Worthiness Based Social Scaling
D'Epifanio Giulio
2018
Abstract
The construction of a set of scales is delineated, for evaluating the performance of social agents (e.g. providers of services as hospitals, schools, etc.) condition- ally on “reference states” x := X ∈ {x 1 , . . . , x R } of the governed individuals. Each scale is associated to an index which uses conditional “worthiness increases” ω l|x , between the levels of an ordinal outcome indicator Y := l ∈ (0, 1, .., L). This indi- cator was been defined on a scheduled, by the policy-maker (PM), chain of hier- archically ordered goals. The “worthiness increases” are interpreted by modeling interrelated latent evolutionary processes, on the scheduled goal chain, up to hyper- parameters γ which are driven by conditions x. Then, to standardize the set of scales on a given “reference behavior”, a pseudo-Bayesian (see [1]) method is used which elicits value γ ∗ by minimizing “residual from updating” (see [4]). It norms the model specifications on the “reference data” of the (chosen a priori) “standard agent”. Fi- nally, adhering to general requirements in rational choices from the decision theory, a standardized worthiness-based index can be implemented, which takes into input the agents actual dataI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
