How does iSeeCars (who did the study) know how many miles were driven by each brand's cars? It says they have a database of cars, but do we know whether it's an unbiased sample?