Siena College bombed last week's poll on the Rochester Democratic mayoral primary. The poll predicted incumbent Mayor Tom Richards had 63 percent of likely voter support versus 27 percent for Lovely Warren. On Primary Day, Warren captured 59 percent of the vote.
"We got it wrong. I feel bad," said pollster Don Levy.
Levy said Siena College polled voters in Albany and Buffalo during the same time frame as the Rochester poll. They asked the same exact questions. The Albany and Buffalo races were accurately predicted.
In Rochester, Siena College's polled 60 percent white people and 37 percent black people. The results showed a divide along racial lines. But Levy said even if huge numbers of black people came out to vote, that would not have changed the poll results significantly.
Levy does not believe Richards supporters stayed home because they thought Richards would win in a landslide. Rochester results were predicted for the Buffalo and Albany races and those races had similarly low turnouts. In other words, landslide predictions didn't keep Buffalo and Albany voters from voting for the frontrunners.
So what went wrong?
Levy said the Rochester poll had a bad sample. he said in polls this happens statistically one of every 20 times.
"I think that really the unfortunate truth is once in a while, one time out of 20, you're going to get a sample that is just simply not exactly like the people you want to report on and this happened to be that case," Levy said.
Copyright 2013 Nexstar Broadcasting, Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.