|
Abstract:
|
Researchers often assume outliers are statistical noise. What can we learn if we instead view them as significant? In this talk, we show that outliers are central to driving —or delaying— underlying system changes. We propose a volatility metric and demonstrate its utility by studying trends in place of death. Borrowing from the macro-economic concept of volatility within key commodities, we advance this method in 3 concrete ways: 1) applying an agile methodology to manage scope; 2) setting policy-time boundaries, and 3) integrating state geography. To demonstrate utility, we use federal death certificate data (2003–2019). We link volatility scores to ecological health systems data, using the state as the unit of analysis. In this talk, we show that the volatility metric adds substantive insights beyond traditional trend analyses. Outliers were associated with coroner-based death certificate systems and hospital closures. Systemic improvements should focus on modernizing the death certificate systems by allocating additional resources to states with high volatility. This expanded volatility metric can advance trend analyses and improve the utility of policy analyses.
|
