What Matters and How Much? Assessing the Relative Influence of Physical and Social Forces on Tornado Warning System Performance

In recent research, scientists have explored the individual impact of meteorological, environmental, climatological, and technological variables on tornado warning system performance (e.g., Brotzge, Erickson, and Brooks 2011 and Brotzge et al. 2013). In this project, we build upon this research in two ways: first, we assess the collective and conditional impact of these “physical” variables on warning system performance by estimating a set of multivariate regression models that simultaneously account for all of these variables at once. This will allow us to detect the relative effect of one variable or group of variables (i.e., convective mode or population density) on performance while statistically controlling for the effect of other variables know to influence system performance (i.e., time of day or distance from radar). Second, and more importantly, we expand upon previous research by attempting to isolate the influence of “social” forces on warning system performance by including a variety of organization-level variables in the models we estimate. This will allow us to explore the oft discussed but seldom measured effect of human and organizational judgment on warning decision-making. In so doing, this research will advance our understanding of warning system performance, which, as demonstrated in a separate line of recent research, systematically influences public trust in the National Weather Service and, ultimately, warning response (Ripberger et al. 2014).