You just don't get it. Without a truly random sample, you can make NO extrapolations from the data. It's that simple. Any estimations based on the currently (unproven) Frizzy data are flawed.