Idk what was ”wrong” exactly with the other models, but I was following @RacetotheWH at the time, and their model showed Cuomo squeaking out a victory in the 10th(!) round of RCV. A slight tweak of their model could’ve predicted Mamdani’s win, despite the massive polling errors.
Every polling model predicted Cuomo would win because the average of the polls predicted Cuomo would win. In aggregate, the polls were significantly wrong, which is not really something the poll analysts can correct for.
Jun 2, 2026 · 1:13 PM UTC
1
1
44
All the repeat polls showed Mamdani gaining ~2 points per week while Cuomo stood still. But because momentum was calculated from how their polling *average* had moved, Mamdani was given negative momentum, which was obviously not the case. Methodology still matters!!
The main thing they can fix on their own is how they calculate momentum. Instead of apples to oranges, it might be worth looking at patterns within each pollster. Afaik Mamdani has been improving his numbers month-to-month *within* each repeat pollster.
22