*Math Wins*

**Update:** I have received searches related to “are the polls rigged for hillary clinton”. Yes. Yes they are — at worst, Trump is *tied* with Clinton. Expect this bias to go away as the election approaches and nobody wants to be the poll that was wrong by 10-15 points.

(Note, rather than link to each poll individually I am going to drop the RCP polling average link right up here.)

A new Reuters poll came out showing Trump losing by ten to Hillary Clinton, with a survey of 1339 registered voters. This seemed out of whack almost as bad as Bloomberg’s recent poll. The Reuters folks don’t supply a true democraphic breakdown but rather show how much support each candidate gets from professed democrats, republicans, and independents. That makes it difficult but not impossible to determine how many people they polled from each group.

And the poll is rigged.

1339 people responded to this poll, and Clinton gets 44%, Trump 34%, and the remaining 22% are total idiots. But in the partisan breakdown Trump wins Independents 32-27, Republicans 76-9, but loses Democrats 72-10. According to Gallup, party affiliation shows the Democrats slightly ahead. If you assume the Gallup numbers for party affiliation are correct then the Reuters poll should look something like this (rounded):

Party | Percent | Respondents |

Republican | 27% | 361 |

Democrat | 30% | 402 |

Independent | 41% | 548 |

That would mean the vote totals, using Reuters’ percentages, should look like:

Candidate | Democrat Voters | Republican Voters | Indepdent Voters | Total | Percent |

Clinton | 289 | 32 | 148 | 469 | 35% |

Trump | 40 | 274 | 175 | 489 | 36.5% |

If you adjust the Reuters numbers for party affiliation then Trump is **winning**. But it leaves the question: who the hell did Reuters talk to?

I’m not going to go into all the math I did in order to figure this out. Basically I used the support of Hillary, Trump, and the combined “other” options to estimate how many respondents overall said they would vote that way. 44% for Hillary means that 589 of 1399 would vote for her (rounded). Given her support by party breakdown this means that Reuters polled around the right number of Republicans (27ish%) but around 30-31% Indepdents and **44%** Democrats. That’s over-sampling Democrats by **fourteen percent**

**FOURTEEN PERCENT**.

The CNN poll has Clinton winning by 5%. But if you click to see their internals they polled the correct number of Democrats (30%), but too few Republicans (23%). This impact is probably much less given Trump’s numbers with independent voters but it increases my suspicion of the poll.

Gravis, which has Clinton winning 51%-49% oversampled Democrats by 10%, Republicans by 6%, and under-sampled “other” by 13%. Given that Clinton has the support of roughly 70-80% of Democrats and Trump has the same support among Republicans that is a net gain of 3% for Mrs. Clinton — more than the margin of difference.

Rasmussen does not release their internals unless you pay for them. Given the controversy surrounding their 2012 polls I’m not willing to do that.

Bloomberg has buried its demographics somewhere far down — probably hiding a lot of cover that 12% lead.

Trump is currently leading Clinton and that is being buried, either out of incompetence or malice. I don’t care which, and no one else should. But these folks are putting their reputations on the line — as the election nears the numbers will get more accurate.

Vote. Be sure it isn’t close enough to steal.

—

If you want to help unskew the polls, check out The Boots Are Red, and understand that it’s more realistic than the mainstream media!