A Power Analysis of Two Surveillance Methods in Terms of Average Run Lengths
Gerald Shoultz, (Department of Statistics, Grand Valley State University), email@example.com,
Paul Stephenson, (Department of Statistics, Grand Valley State University), firstname.lastname@example.org, and
J. Wanzer Drane, (Department of Epidemiology and Biostatistics, University of South Carolina), email@example.com
This talk compares two methods for testing hypothesis usable in disease surveillance and process control to determine which is more powerful: TEXAS (Hardy et al 1980) and CUSUM (Hawkins and Olwell 1998). TEXAS, a modification of the procedures of Shewhart (1931), uses a two-step decision rule to determine when a process is out-of-control. CUSUM finds a process to be out-of-control when the sum of a set of measurements exceeds a given threshold. While there are many reasons for monitoring disease incidence in a community or region, one likely application of such methodology is identifying if a government agency should investigate whether or not a terrorist event has occurred. First, the authors will discuss how these process control procedures can be used to monitor disease surveillance. Then the authors will present a simulation that compares the performance of the TEXAS and CUSUM methods to determine which method is more powerful for a variety of hypotheses.