When analysts use “spread” to determine poll accuracy. It's not only not right…it's not even wrong. It's nonsensical 👇