IER8

ITS 832 CHAPTER 10 VALUES IN COMPUTATIONAL MODELS REVALUED

INFORMATION TECHNOLOGY IN A GLOBAL ECONOMY

DR. JORDON SHAW

INTRODUCTION

• Technology perceptions

• Technology and public decision making

• Methodology

• Case studies

• Analysis

• Summary and conclusions

TECHNOLOGY PERCEPTIONS

• Debate on underlying assumptions of models

• Are models biased? • Is technology biased?

• Are model builders biased?

• Are model users biased?

• Technological determinism • Technology is not neutral of value-free

• Social construction of technology • Technology is designed with bias, or values

• Technological instrumentalism • Technology is neutral and value-free

TECHNOLOGY AND PUBLIC DECISION MAKING

• Policy making involves complex systems

• Model bias must be understood to evaluate results

• Bias, or value can be categorized • Values of the data

• Values of the model

• Values of the decision-making process

METHODOLOGY

• Select six case studies

• Carry out secondary analysis of results

• Identify cases with three basic characteristics • New model designed for case

• Relate to policy issues with the natural or built world

• Highly complex and controversial issues

CASE STUDIES

• Morphological Predictions in the Westerschele (Belgium and the Netherlands)

• Morphological Predictions in the Unterlbe (Germany)

• Flood-Risk Prediction (Germany and the Netherlands)

• Determining the Implementation of Congestion Charging in London (UK)

• Predicting and Containing the Outbreak of Livestock Diseases (Germany)

• Predicting Particular Matter Concentrations (the Netherlands)

ANALYSIS

• Analyzing empirical data resulted in several findings

• Values in data • Cases 1-4 exhibited higher trustworthiness of data

• Margin of error high in all cases

• Values in the model • Similar to values in data findings

• Values in the decision-making process • Clear lines of authority in cases 1, 4, and 5

• Lack of clear authority (cases 2, 3, and 6) leads to conflict

SUMMARY AND CONCLUSIONS

• Model effectiveness is impacted by bias

• Values can originate from multiple sources

• Data

• Model design

• Model use

• Outcome validity requires a clear understanding of values put forth by model use