Statistics Colloquium

Speaker:

Title:

Abstract:

Statistical tests assessing agreement between two comparable measurement methods regarding certain features of a population have drawn increased attention in recent years.  Such tests are typically based on paired data (x, y) collected from two measurement methods under comparison from several units of the population.  In this presentation, we discuss appropriate statistical methods of establishing agreement when a non-standard paired data of the type (x, y1, y2, . . . ym) are recorded.  This corresponds to the situation when one measurement method (x) is much more expensive than the second method (having multiple observations).  We illustrate our methods with an application from USEPA and simulations.