In 2016, Ohio was among the worst-polled states in the country. An RCP average leading up to the election showed Trump with just a 2.2% lead in the state the day before the election. Trump ended up winning that state by 8.1%, a larger margin than any poll had given Trump in the days leading up to the election and well outside any margin of error factored by any polling firm. The closest, Emerson, listed the Trump lead at 7 points, three days before the election, with a 3.2 point margin of error.
As previously pointed out, despite the 2016 polls being off by huge margins, the media and their polling firms have not indicated that they have identified the issues they had with their methodology nor have they suggested any changes they have made to their 2020 methodology that would give them more accurate results. They continue to release poll after poll this season that shows Biden with massive leads, while ignoring their questionable means of conducting the polls. Even in places where they do admit issues from 2016, their 2020 methodology has not reflected any corrections.
In early October 2016, CBS/YouGov released a poll that showed that Clinton was leading Trump by 4 in the Buckeye state. That poll was conducted October 5-7, 2016 of 997 likely voters. While some of the methodology information is missing from these results, much of the data appears self-explanatory. The first thing noticed by looking at this poll is the partisan sampling, with 36.6% Democrat, 27.3% Independent, and 34.9% Republican. I certainly understand the problem with identifying voters in Ohio as the state does not register voters according to party. That being said, some organizations identify Ohio voters according to party, which in 2016 had voters at 16.9% Democrat, 56.4% Independent, and 26.6% Republican.
If you were to adjust the October 2016 poll for what the registration should be, it shifts from a 4-point Clinton advantage to a 6.8% Trump advantage. Remember, Trump won the state by 8.1%. The previously mentioned Emerson poll was closest, with two Republican firms rounding out the next closest, Remington and Trafalgar both at Trump +5.
In Ohio this weekend, a new poll was released by CBS/YouGov, for the 2020 season. This poll has Biden and Trump at a dead tie in the state, but again, some glaring sampling issues come to light. First is that the poll has sampling at 35% Democrat, 27.1% Independent, and 36% Republican. Sampling Democrats at that rate is almost laughable as it is twice as high as the state’s registration suggests. Even if that’s correct (it isn’t) Democrats are only showing a 6% bump over their 2016 support from Clinton to Biden, while Independents are a 4% bump in favor of Trump (from 42 to 46) and Republicans show a 10 point bump in favor of Trump over 2016, 81 to 91. In other words, while Dems have increased the share of their own vote by 4 points, support for Trump has increased for both Independents and Republicans over 2016 numbers. Independents and Republicans make up more than 82% of the total Ohio electorate.
If we were to adjust the poll results to factor for the poor sampling, we end up with a Trump lead, 49.403 to 39.82, or a 9.58% lead.
Again, if we look at the most accurate polls from 2016, Trafalgar just released a poll this morning showing Trump with a 3.7% lead in Ohio, with a 2.97% margin of error.
If anything, YouGov’s current results show that Biden has indeed slipped in support compared to 2016, which would do more to suggest a larger Trump margin of victory in 2020 than his 8.1% margin in 2016. As I have shown, the registration statistics in 2016 do line up with the polling results, just with bad sampling. If the registration statistics hold, Trump is on his way to a 10-point victory in Ohio in 2020.
