My “problem is with” the premise that a phone survey of 1,000 people accurately provides data about the entire population.
They don’t need to call more than a 1,000 people. Calling 10,000 or 100,000 would show much the same result. It’s a math thing. One thousand is a common sample amount.