The original unweighted poll showed Trump beating Hillary in Ohio 41-39, but when the poll was later amended to include more Democrats, Hillary was handed victory by a margin of 43-39. The unweighted sampling of voters was split 33.3R/29.3D/35.6I, but was subsequently changed (pages 6-7) to 29R/33D/37I in order to ensure Hillary came out on top.

When Twitter users called out Monmouth University over the controversy, pollster Patrick Murray denied the poll was weighted by Party ID, even though it clearly was.

“Just as a reminder “weighting” is the term used for actually changing the raw data to reflect the “assumptions” the pollster puts into the anticipated turnout of the election,” reports the Conservative Treehouse. “Depending on the “assumptions” (guesses about who will vote), and the “weighting” therein, the entire poll outcome can be severely manipulated and even reversed.”

As radio host Bill Mitchell explains in the video below, the vast majority of polls oversample Democrats, which is why Hillary is seemingly doing so well. Americans identify as Republicans or Democrats at a roughly even rate, so overall polling samples should be evenly distributed between Republicans and Democrats. But they’re not. Most polls ask more Democrats than Republicans who they are voting for, which is why Hillary always seems to come out on top. When the sampling is evenly distributed, Donald Trump wins the poll.

In addition, polls are being amended to make Hillary’s victory seem inevitable by adding in her favor the votes of respondents who said they wouldn’t be voting at all. Reuters also eliminated the choice of voting for neither candidate in its poll in order to artificially boost Hillary’s numbers.