A random sample of 2100 people is often sufficient to get a statistical significant result for a population of around 50 million with a 5% margin of error.
I’d maybe add a few thousand more for a study that represents the entire USA, but 2100 isn’t a small sample size.