It would statistically improbable that a candidate would get no votes. That would imply error-free voting and mean that not one person made a decision to the contrary. You would have zombies voting if that was the case.
People are not statistics. Each person makes a decision about who to vote for. Your claim is that it is statistically impossible that every person living in a building could possible each have chosen to vote for Obama; but the nature of the people who live in that building is that they would vote for Obama.
Elections are not statistical. To some degree, with large enough numbers, we can model them as if they are random variations of a scientific process, but they are not.
Error-free voting isn’t all that hard, when you are voting straight party; some places allow a single mark to vote for party, although I don’t know if Philidelphia is one of those.
And it’s not hard for people to vote “Obama”.
So the argument that Obama voters are so stupid that some of them would have to have mistakenly voted for Romney isn’t very persuasive.
On the other hand, it wouldn’t surprise me if someone was committing fraud in the inner city; if they are, it wasn’t Obama in 2012, it is the democratic party for the past two decades.
In every case where I have been able to compare Obama’s numbers in a precinct with the 2008 and 2004 election numbers, they have been similar. There were dozens of precincts in Ohio in 2008 where McCain got 0 votes, and even many in 2004 where Bush got 0 votes (although I think with Obama being black, there was even more of a reason for the black inner-city population to vote democrat, as we can see by the higher total percent of black votes for Obama than for Kerry in 2004).
In fact, it is easy to disprove the notion that voter error would make it probable that a candidate would get at least one vote — simply look at how many of the non-major candidates get zero votes in thousands of precincts in each state.
If it was “impossible” for there not to be voter error, that voter error should equally spread votes among all candidates. So the total votes received by the worst candidate would establish a maximum baseline for the “error rate” of voting.
There is a website that allows you to look up individual philidelphia precincts, and see the results. It’s not searchable; there is a full database but in it’s text form it also isn’t searchable.
Anyway, I poked around for a while, and found Ward 6 Precinct 12, which in the 2012 general election, gave Romney only 2 votes:
In that same precinct (ward 6, precinct 12), in 2008: Obama=276, McCain=0.
In that same precinct in 2004:, Kerry=176, Bush=7;
In that same precinct, other statewide elections:
2002 Governor: 122 to 3
2006 Governor: 98 to 4 (black republican)
2006 Senate: 95 to 3
2010 Governor: 166 to 2
2012 Senate: 167 to 3
So what we see here is a district that in 2012, barely registered a vote for Romney, but in fact has barely registered a vote for republican since the 2004 presidential election, and even then it was very few republican votes.
I am sure that if someone actually provided the list of wards and precincts where Romney got 0 votes, I’d find that republicans were getting no votes in those precincts in prior elections as well.
Which means it wasn’t some special fraud in 2012 to steal the election from Romney.