You don't think you might be seeing some selection bias in your sample of "people who paid to be somewhere quieter"?