Here is your sampling data.

They sampled a little over 1,500 people for a margin of error of 3%.

First, tell me why a sample of 1,545 is representative of the 340,000,000 people in this country?

They sampled 0.00045% of the country.

Second, the 3% margin of error means the results could differ by at least 6% outright.

The margin of error is related to the sample size. You could bring that down, but you would have to sample more people.

Nowhere tells you HOW they collected the sample, which was likely done by calling people on the phone. In some cases, these surveys use landline numbers.

How many people still have a landline?

They sample the same people over and over again, so nothing ever changes.