John Pudner: Tracking the Silent Vote and Guilt Vote

Auburn, AL

If a “guilt vote” propelled President Obama to two election wins, will Donald Trump be elected with a “silent vote” because the national media makes it too embarrassing to tell a pollster you are supporting Trump?

Guilt Vote

Democrats provide intense training for their field operatives, immersing them with one group after another. First you talk with an LGBT group, then a group of young Evangelicals etc., until you learn how to win more of them over. After getting comfortable with various groups, they figure out ways to maximize the vote with each group. For example, many young evangelicals in recent years said they could support a candidate who supported “marriage equality” as long as they also said they were “pro-life,” so the trained operatives use those words to make inroads with young evangelicals while still maximizing the LGBT vote.

In other cases, back in 2008, they found that many moderately conservative voters were embarrassed by their older family members – they remembered their grandparents sounding racist, and they wanted to prove they were not likewise racist – so they were given the golden opportunity to make history by voting for the first African-American president – which voters did in states like Ohio, Virginia, North Carolina, Indiana and Florida.

When some of these moderate conservatives discovered the President they helped elect actually differed from them on many issues and, four years later, Mitt Romney closed the gap and appeared to have a chance to win back Florida, North Carolina and Virginia, the response was ready made. They were told the “Old Confederacy” was falling back in line against America’s first black President – and all but North Carolina were brought back into the fold with the Focus Group-tested wording and huge field staffs.

While Hillary Clinton’s KKK push attempts to recreate this momentum from the last two elections, she faces one obstacle that President Obama did not face but that has defeated other Democrats over the years, the silent vote.

Silent Vote

The one time general election polling can be off by several points is if the media and billions in commercials make it too embarrassing for someone to tell a pollster they are for a particular candidate, thus skewing the results.

Neither John McCain or Mitt Romney benefited from a “silent vote” because both were viewed as “respectable” choices. A polite war hero and successful businessman were legitimate choices.

The election of Doug Wilder in 1989 as Virginia Governor was one of many instances in which the media created a silent vote by making it embarrassing to vote for Wilder’s opponent. On election night the most reliable polls of all – exit polls of people leaving the booth after just having voted – showed Wilder winning the election 55 percent to 45 percent. And yet, when results came in, it turned out Wilder had barely received the 50 percent he needed to win.

My favorite political commentator, Larry Sabato, went on the air to make an incorrect assertion accepted as Gospel by most commentators: that the only explanation for the skewed polls was that the voters simply lied to pollsters.

However, it is actually not necessary for anyone to lie to a pollster to skew a poll.

When pollsters want to run a poll of 1,000 voters, they may get phone numbers for 30,000 because the vast majority will either not answer their phone or refuse to go through a poll. I last ran a polling call center a few years ago, and at that time we ran 60 percent of each poll via a land line, 15 percent with an on-line poll, and 25 percent to cell phones which we had to manually dial.

Lets assume that there were 1,000 people on your list who normally would have answered the phone and taken the poll, except that 50 of those 1,000 decide to tell you they are not going to take your poll for one reason – they secretly like Donald Trump but are scared of getting ridiculed or called a racist. If the 1,000 people would have broken 520 to 480 for Trump, and instead those 50 silent Trump supporters are replaced by 24 Clinton supporters and a new 26 Trump supporters, then the poll would show Clinton winning but Trump would really project to win the race 52 percent – 48 percent on election day.

Measuring the Silent Vote

One poll that is trying to measure this silent vote is the USC Dorsife/LA Times Daybreak poll which comes out at Midnight each night and tries to measure true opinions by letting voters stay anonymous on-line.

As of August 29, the LA Times showed Trump ahead by 0.4 percent, while the Real Clear Politics average of 4-way polls showed Clinton with a bigger 4.4 percent lead. I believe the most accurate is the National/Battleground Daily Tracker (methodology explained here, updated numbers at end of article) which now shows Clinton 2.3 percent ahead and has shown Trump within three percent ever since August 18, after trailing by between 5.6 percent and 5.8 percent the previous several days.

While the LA Times attempt to measure the silent vote gives it accuracy on one front, they way over sample new voters. Trump should do well with new voters – as many were giving up on the system and now feel they have a champion, and President Obama already maximized new voters in many more liberal groups.

The problem is that the LA Times uses a sample, as I understand it, that includes 25 percent of people who voted for Romney, 27 percent who voted for Obama, and 48 percent that were too young to vote or decided not to vote in 2012.

Trump will likely get millions of voters that did not vote for Romney, which is why he and Marco Rubio were the only two candidates with a chance to reset the map that was skewing heavily toward Democrats in Presidential elections.

But 48% new voters?

In 2008, Obama’s incredibly effective effort to reach new voters resulted in 15 million of 131 million voters being first time voters. There may have been a few million more who were not first time voters but did vote for Obama in 2008 after skipping 2004. However, I doubt there were more than 18 percent of voters in 2008 who had not voted in 2004, so the LA Times appears to be counting new voters about three times as much in their sample.

New Voters and Silent Voters Should Add to Trump’s Total

Typically a candidate with an expanding base, which Trump clearly has and none of the other potential Republican nominees besides Rubio would have had, beats a candidate with a contracting base, which Clinton would seem likely to have just because it is so hard to match the turnout enthusiasm Obama had.

If Trump stays within his current -4.4 percent in the Real Clear 4-way, -2.3 percent in our Battleground Tracker, and +0.4 percent lead in the LA Times poll then he is much closer than when the silent vote turned Wilder’s 10-point lead into a 0.4 percent win.

And all that can be true without a single “Silent Voter” lying to a pollster.

Update on Battleground Tracker: Trump Within 3 Points Ever Since August 18

Donald Trump’s 4-point improvement in Michigan last week helped him stay within three points in the overall National/Battleground Daily Tracker – where he has been ever since August 18. August 14-17 he trailed by between -5.6 and -5.8 every day. He then improved two more points in Michigan on Monday.

As outlined in the explanation of the National/Battleground Daily Tracker, Trump likely needs to get within 3.0% of Clinton in Michigan to become President, and his improvement from 11 points down in the last two Michigan polls on Real Clear Politics (Aug. 5 and 17) to just seven points Thursday and five points Monday indicates he is getting closer as part of the weighted vote shown below.

Previous daily totals going back to Clinton’s 7.7 point lead at the beginning of the month are posted along with the explanation of how to weight each state poll to gauge who is leading. As noted, the tracker is updated every time Real Clear Politics posts at least two national or battleground state polls to average with the daily LA Times poll. A 7-point deficit in Michigan is the equivalent of Trump’s one-point deficit in North Carolina Sunday.

In modern polling, results are generally released a day or two after the poll is completed and polls take a few days to complete – so all completed surveys occur within a week of the release of the results.

The one poll that did not meet this criteria is the Roanoke College poll released August 23 showing Hillary Clinton with a staggering 16 point lead over Donald Trump in Virginia. The problem with this poll is that it was “old news” by the time it was released. The poll started on August 7, more than two weeks before the release date, and was not finished until August 17.

The middle of the polling was therefore August 12, when the National/Battleground Daily Tracker showed Trump near his low point at 7.2, so the Roanoke poll would indicate Trump was even further behind in early August. It was not averaged, as we will only include polls run completely within a week of their release date.

Here are the day-by-day totals:

August 29 Average Trump minus 2.3

Real Clear National -7 add 0.5 = -6.5

Michigan -5 plus 3 = -2.0

Ohio tie minus 1.5 = -1.5

Pennsylvania -3 plus 0.5 = -2.5

LA Times +0.4 plus .5 = +0.9


August 27 Average Trump minus 1.4

Real Clear National -1 add 0.5 = –0,5

Florida -2 subtract 1 = -3

LA Times -0.8 plus .5 = -0.3


August 25 Average Trump minus 2.8

Real Clear National (3 polls) -4.7 add 0.5 = -4.2

Michigan -7 add 3 = -4

LA Times -0.3 plus .5 = +0.2


August 24 Average Trump minus 1.8

Real Clear National -4 add 0.5 = -3.5

Florida +2 subtract 1 = +1

North Carolina Tie subtract 3 = -3

LA Times -2.1 plus .5 = -1.6


August 21 Average Trump minus 2.3

Ohio -6 subtract 1.5 = -7.5

Iowa Tie subtract 2.5 = -2.5

North Carolina +1 subtract 3 = -2

LA Times +2 plus 0.5 = 2.5


See previous post for earlier August National/Battlefield Tracker. On August 18 Trump trailed by 2.0, on August 16 by 5.6, August 14 by 5.7 and August 12 by 7.2 though including the Roanoke poll based on when it was run would have pulled the Trump to just over seven points behind on August 14 or 16 result.