Does someone have a good model for estimating these kinds of figures given widely divergent reports in press sources and otherwise? How might we approach such a problem with a greater degree of statistical/mathematical rigor? What information DO we have at our disposal that we might devise a model approach?