A ~1% popular advantage is not anywhere near what the committee's own polling data suggested (30-40% spread), nor that of any mainstream media outlet. They were closer to the edge of the scale than the actual number. That is to say, they were almost as wrong as numerically possible, without going the opposite direction.