It must have been a long night for Team Tony following the release of the 2015 mayoral primary's one and only independent poll.
The results rebuffed widely-held views of the ongoing race as a close contest between state Sen. Anthony H. Williams, former Councilman Jim Kenney and ex-District Attorney Lynne Abraham.
When tallied, the voters surveyed suggest a runaway in the race to succeed Mayor Michael Nutter — with Kenney leading by an unbelievable 27 points over both Williams and Abraham, who tied at 15 percent. Candidates Nelson Diaz, Doug Oliver and Milton Street trailed with single digit support.
The poll commissioned by The Inquirer, Philadelphia Daily News, Philly.com, and NBC10 is the only independent survey of the primary.
That differs from the last contested primary, in 2007, which featured polls — conducted by Franklin & Marshall College for the Daily News — at the beginning, middle, and end of the race.
Why was there only one this time around?
Blame cell phones.
"Part of it is so many voters don't use land lines and, for complicated reasons, it's just easier to call land lines versus cell phones," said Kyle Kondik, a political analyst at the University of Virginia.
Mobile phone records can be harder to obtain and make voters, well, more mobile and harder to track down. Someone with a 215 area code might keep that exchange when they move to, say, Juneau, Alaska, or vis versa. In general, it's about twice as costly to survey cell phone users.
These difficulties are doubled in big cities like Philadelphia.
"The proportion of the population that only owns cell phones is higher in urban places. Young people and minority groups are more likely to have cells," said Scott Clement, survey research analyst for the Washington Post.
And in an age of built-in caller ID and text-prone millennials who tend to avoid phone calls altogether, fewer people bother to even pick up. It all adds up to more calls by surveyors, which means polls are more expensive to conduct.
"The response rates for polls are probably as bad as they've ever been," said Kondik. "If you're making all these calls and no one is picking up, you have to make more calls and there's money involved in that too."
Response rates for pollsters across the country have dipped from 36 to nine percent over the last 18 years. To get 596 responses for the Inquirer/NBC10 poll, survey company National Research Inc. had to make between 20,000 and 25,000 calls.
It's a tough scenario for the groups that have historically pushed for political polling.
"News organizations have long been the main sponsors of local surveys," said Clement. "Costs of surveying have increased and newsroom budgets have not."
Still, polling experts said increased difficulty has not generally decreased reliability.
"Six hundred voters in a random sampling of a population is enough to give you a accurate sense of where voters are today. It's a common sample size, it's not exceedingly low or high for a low-turnout election," said Clement.
Kondik noted that primaries sometimes decrease polling reliability because candidates often have overlapping political philosophies.
"The electorate is probably having a hard time distinguishing between the candidates," Kondik said. "Maybe the poll you take the week before [the primary] happens and maybe it was accurate at the time, but it's not like a presidential election where you don't have a lot of movable voters. A primary is just a lot more unstable."
Clement, however, argued the double digit lead attributed to Kenney, from a poll with a four percent margin of error, was "statistically significant."
"A good rule of thumb is that if a candidate has a lead of one and a half times the margin of sampling error, then you can call that a statistically significant lead," he said. "Two times the margin of sampling error and you're looking at real lead."