This is a nifty tool, but it doesn't offer statistically independent samples. For many people, they don't care, but calculating from just a single track is really a measurement of how much your track "drifts" as satellite move across the sky, and not so much your location error.

A better (but still not perfect) way to do this is to take measurements at several known location benchmarks, and then repeat this over a few days. Your local city government probably has a listing of known benchmarks, in Las Vegas the website is: http://www.lasvegasnevada.gov/Find/survey_benchmark.htm

Doing this over the course of several days is important because you get data from several different GPS satellites from several different positions in the sky. This is much more labor intensive, but you will get a much better spec of how much you can trust your GPS.