e管家還是e管區？數位身分識別證下的隱私計算

民眾授權個人資料的意願，是個人資料自主管理（MyData）政策成敗的關鍵。本研究以數位身分識別證政策為個案，應用隱私計算模型分析個人資料授權意願背後的隱私因素，採用偏最小平方法結構方程式（PLS-SEM）來探索數位身分識別證議題上，民眾在利益與風險間的權衡，並經由政治大學選舉研究中心建置的「線上調查實驗室」（PollcracyLab）進行資料蒐集。本研究發現財務報償、個性化服務與服務兼容性都有助於提高民眾的隱私利益認知，而隱私利益認知則進一步會提高授權個人資料的意願；然而，隱私風險認知並不會影響民眾透過數位身分識別證授權個人資料的意願，代表民眾對隱私所帶來的風險有過多的忽視。本研究援引了行為經濟學、隱私悖論與遲滯性風險的觀點，探討了導致此認知缺口的可能邏輯。本研究側面印證了臺灣是一個「遲滯型高科技風險社會」的推論，在具高度不確定性的科技議題上，民眾可能低估了潛在的隱私危害，並高估了預期效益。未來推動數位身分識別證的決策者，應跳脫僅以民意調查作為決策參考的思維，嘗試納入多元的決策機制於政策過程之中。最後，本研究討論了納入調節變項與高階構念等模型修正策略，以及建議未來研究者可以透過入選機率調整法（Propensity Score Adjustments）與實驗設計（experimental design）等方法修正調查方法上的偏誤。

關鍵字

數位治理；數位身分識別證；隱私計算；科技風險；結構方程式

並列摘要

The success or failure of the MyData policy depends on citizens' willingness to authorize their personal data. This study applied privacy calculus model on the case of national electronic identification card (New eID) policy to analyze privacy factors which affect personal data disclosure intention. Partial Least Squares Structural Equation Modeling (PLS-SEM) was used to explore how citizens balance benefits and risks associated with the New eID issue. Research data were collected from PollcracyLab affiliated by Election Study Center of National Chengchi University (NCCU). The study concludes that financial compensation, personalized services, and service compatibility can enhance cognition of privacy-related benefits of citizens, while this cognition will further promote their willingness to authorize their personal data. Citizens' cognition of privacy-related risks, however, has no statistical effect within the model, and shows that citizens excessively neglect privacy-related risks. The perspective of behavioral economics, privacy paradox and delayed risk are cited in this study, to demonstrate the abovementioned cognitive gap. This study verifies the inference that Taiwan consists of a “Delayed High-tech Risk Society”. In technology issues with high uncertainties, citizens may underestimate the potential privacy-related risks, and overrate expected benefits. Therefore, policy makers in charge of implementing the New eID policy should incorporate diversified decision-making mechanisms into their policy process, and avoid taking opinion polls as the only reference. Finally, this study discusses model revising strategies such as adopting moderator variables and higher-order components. Future researchers are also recommended to correct survey method bias by such approaches as Propensity Score adjustments and experimental design.