Polling experts have criticized the findings of a Pennsylvania election survey which under represented likely voters from Philadelphia.
An American Greatness/TIPP survey of 1,079 registered voters in the key battleground state showed Kamala Harris with a 4-point lead over Donald Trump (49 percent to 45) in a head-to-head. Among a smaller sample of 803 likely voters, Trump leads Harris by 49 percent to 48.
The winner of Pennsylvania in November could determine who wins the 2024 election overall, with surveys frequently indicating the state is essentially tied.
Harris’ clearest path to victory in November would be to win the three blue wall battleground states of Pennsylvania, Michigan and Wisconsin, barring any shock results elsewhere. Trump’s most efficient path to 270 Electoral College votes would be to win the swing states of North Carolina, Georgia and flipping Pennsylvania.
While the American Greatness/TIPP survey suggests Trump is narrowly ahead of Harris in Pennsylvania among likely voters, pollsters have noted the results have largely excluded respondents from the state’s most populous city of Philadelphia.
The poll’s cross-tabs show it asked 124 registered voters in Philadelphia who they would vote for in November’s election, amounting to around 11 percent of the total sample. In comparison, only 12 of the 807 likely voters sampled in the survey were from Philadelphia, or around 1.5 percent.
Alex Wong/Getty Images
Newsweek has contacted American Greatness/TIPP for comment via email.
Pollsters and forecasters have now criticized the Pennsylvania poll results after seeing that only 12 likely voters from Philadelphia were included, seemingly helping to eradicate Harris’ lead in the state.
Lakshya Jain, cofounder of Split Ticket, an election analysis website, posted on X, formerly Twitter: “99% of the time, I tell you all that cross-tab diving is stupid, and that weighting decisions change things more than anyone can expect.
“This time, it’s actually justified to look at it. If you drop Philadelphia respondents from a poll of Pennsylvania, that’s just dishonest.”
Jain added that TIPP has a history of screening its likely voters data, but the Pennsylvania/Philadelphia survey was the “most egregious case” he has seen.
Former pollster Adam Carlson said: “A 5-point difference between registered voters and likely voters isn’t a thing.
“This is what happens when you have n=124 for Philadelphia in your registered voter sample and n=12 (!) in your likely voter sample. Chicanery? Honest mistake? You decide.”
Garrett Herrin, publisher at polling aggregator VoteHub, said it will not be including TIPP’s Pennsylvania survey to their polling average scores for Harris and Trump.
“It seems highly likely there was an error in the pollster’s tabulation or reporting,” Herrin wrote.
Philadelphia is a Democratic stronghold and has a population of more than 1.5 million people. President Joe Biden beat Trump in Philadelphia at the 2020 election with 81 percent of the vote.
Daniel Nichanian, editor-in-chief of digital magazine Bolts, said he had spoken with someone at TIPP about the poll.
“He said he reviewed it, & there’s no error; says the poll’s likely voter screen has a half a dozen variables, and it ‘just so happens that the likelihood to vote of the people who took the survey in that region’ was low,'” Nichanian posted.
According to polling aggregator and forecaster FiveThirtyEight’s average, Harris has a 0.6 points lead over Trump in Pennsylvania (48 percent to 47.4).
The forecast model from Decision Desk HQ/The Hill currently gives Harris a 52 percent chance of winning the toss-up state of Pennsylvania.
