Ironically, polling and association studies like this would be something I'd classify as low precision.