"How wrong" is something I can't quantify. It’s tough to pin down a single “modified margin of error,” but I’d say it’s bigger than what the poll shows. Multi-t

"How wrong" is something I can't quantify. It’s tough to pin down a single “modified margin of error,” but I’d say it’s bigger than what the poll shows. Multi-tier questions automatically have smaller sample sizes per option, which inflates the margin of error. Throw in social desirability bias (people not wanting to admit an unpopular opinion) and you get systematic skews on top of random error.