The Quality of Social Simulation: An Example from Research Policy Modelling

profiletestwork
ITS832_Chapter_3.pdf

ITS 832 Chapter 3

The Quality of Social Simulation: AnExample from Research Policy Modelling

Information Technology in a Global Economy

Professor Miguel Buleje

Introduction

• The Quality of Social Simulation: • An Example from Research Policy Modelling

• A simulation is good • “… when we get from it what we originally would have liked to get

from the target”

• Different views • Standard • Constructionist • User community

• Chapter focus • Different approaches to assessing the quality of a simulation

Simulation comparison

Standard View • Verification

• Does the code do what it is supposed to do?

• Refers to well know method of verification.

• In SW, does the code is working as expected or has defects?

• Validation • Do the outputs resemble observations of the target?

• Relies on the observability of reality • Must be able to compare simulation output to reality

• Compares the REAL with artificial data produced by simulation

• Standard view may suffer from under-determination • Multiple incompatible theories may result from the same data

Constructionist View

• Compares • What you observe in the real world with,

• What you observe as simulation output

• Seems similar to Standard view, right? • Constructionists view all observations as constructions

• Constructionists reject the possibility of evaluation because there is no common reality to refer to.

• Evaluation is not possible • Even observations of reality lack the ability to pass validation

User Community View

• Evaluation is carried out • Using the observations of the affected user community

• Not just based on prior knowledge

• Closer to “real” results

• Often, results are influenced by multiple related factors

• The evaluations of any simulation would be impacted by: • Expectations

• Anticipations

• Experience of the Community that uses it.

Policy Modelling for Ex-ante Evaluation of EU Funding Programs

Horizon 2020 Study Workflow

Summary

• Simulation quality depends on simulation process

• Three different simulation views • Standard

• Constructionist

• User community

• User community view • Most promising

• Most work-intensive