The Social Science of Privacy - Effects on Industry and Government Data Use

The sharing of data is at the core of many technology companies. Data sharing is also increasingly important for government decision-making, as stated by the Commission on Evidenced-Based Policymaking, which led to the Foundations for Evidence-Based Policymaking Act. However, in many instances, data used for decision-making is generated by people, and needs to be explicitly shared by the data subject with those wanting to use the data. The decision-making process behind sharing (private) information needs to be understood to assess (and circumvent) potential biases in the resulting data. When assessing bias in algorithmic decision-making, awareness of biases in the training data is essential. This presentation will review social science theories behind data sharing decision-making, highlight a series of experimental studies designed to affect sharing decisions, and present a framework design to detect sources of bias in various data sources.