You’re just plain wrong. A sample of 7,000 against the whole US adult population only provides a margin of error of just over 1%. Hell, you need the representative sample to drop to 350 panel members to get the MOE up to 5%. If you’re going to post comments about experimental design, maybe understand basic statistics and statistical modeling before type something dumb.