Dat Poll Don’t Lie! Unless it’s an NBC/Wall Street Journal Poll
Most of us suspect that political polls are just a bunch of hot air designed to make people vote the way the powers-that-be want.
Politics is a little like sports. We root for our favorite “teams” and in the U.S., political discourse often takes on the reputation (deserved or undeserved) of some other nations and their sports traditions like British soccer fans (bad) or Brazilian Formula 1 fans (better but still extreme).
There’s a whole tradition of poll-watching, poll averaging, and poll prognosticating.
All of these are pretty nasty and sloppy, much like the sandwich depicted as the illustration for this article.
I recently saw the NBC/Wall Street Journal Poll that showed Hillary Clinton is leading Donald Trump by 6 points among likely voters (September 19). This poll was different from other recent polls that showed Trump closer to Clinton, or in some cases, even ahead. It aroused my interest, yet when I clicked on the Scribd document link in the NBC News article, both links said “the owner has set this document to ‘private’.” Aw man.
It turns out Wall Street Journal has all the links and the polls are conducted by Hart Research Associates/Public Opinion Strategies. These polls have been conducted for so long that they have favorable/unfavorable data for President Obama dating back for his entire presidency, as well as historical data on Donald Trump favorability and the famous “country on the right track/wrong track” question going back into the 1980s.
I’m anything but a polling expert, but I recall a friend telling me a few months ago how little I understood about the Democratic party primary process, how I couldn’t read or understand poll data or “toplines” and a variety of other things that came true (Bernie Sanders did not achieve the Democratic Presidential nomination) and that did not come true — Donald Trump did become the Republican Presidential nominee. The friend quoted 538’s Nate Silver, the famous oddsmaker that predicted Barack Obama’s legendary 2008 Presidential campaign victory.
I said, “Didn’t Nate Silver say Trump would never win the Republican …”
So here’s the deal. Polls lie sometimes, but Silver is right on one thing. In the aggregate, over time, they sort of do mostly “predict” and give somewhat of a general idea of where things are going.
So when I did locate the actual NBC/WSJ poll data, I had a look at it, and because of my business and nonprofit background, I saw a few interesting things. This poll sampled 1,000 registered voters. It illustrates how a variety of factors can be used to favor one candidate over another if a favorable result is desired.
All of these differences fall in the category of subtle differences between actual demographics and “desired” ones that would benefit a certain candidate. I am not “unskewing” this poll. I’m illustrating where it diverges from both the U.S. population and voting population.
I wanted to see if this was true over time, so in cases where it was possible, I included numbers from the same organization’s June poll that showed Hillary Clinton 5 points ahead of Donald Trump in the U.S. Presidential race.
Football, legendary Green Bay Packers coach Vince Lombardi said, “is a game of inches.”
Here are the inches.
Sampling more women than men.
Subtle Changes in Political Affiliation
You can see it more clearly when “leaning” independents are combined with their respective main political parties.
Two points here, three points there adds up. The dark red line is the most recent NBC/WSJ poll (Hillary Clinton 6 points ahead), medium red is the June poll (Hillary Clinton 5 points ahead) and lightest red is the Pew Center’s most recent (2015) survey of U.S. political affiliation, a reliable source that does not conduct election polling, only overall trends and affiliations.
Income Level / Voting Participation Disparity
It is true that higher-income voters tend to vote with much greater regularity and frequency than lower-income voters — anywhere from 70 to 80% voting participation for those earning above $100,000 a year to less than 50% for those under 30 and earning less than $20,000 a year, according to the Center for American Progress.
It’s still not quite the level of disparity seen in the NBC/WSJ poll sampling.
The blue line represents the most recent, September 19 NBC/WSJ poll, the red line represents the same poll conducted in June, 2016, and the green line represents the actual income levels from the U.S. Census Bureau/American Community Survey.
As can be seen, the poll diverged in its sampling of mid-low income voters ($10,000-$20,000) and middle income voters ($40,000-$50,000 and $75,000–100,000) while increasing the number of $100,000+ voters.
Union Membership Disparity
NBC/WSJ somehow managed to find a large number of union members to respond to its most recent survey. Union membership in the U.S. has been on the decline for a long time. I was hoping to find this question had been asked on previous NBC/WSJ surveys conducted by Hart Research Associates, but I did not find such a question in this election cycle.
This survey markedly oversamples union members. As I am, myself, a union member, I know that our “rank and file” members voted to endorse Senator Bernie Sanders for President last year; the leadership selected Mrs. Clinton for the official endorsement. I don’t belong to a small union; in fact, our members make up about 45% of the total union membership in the U.S. so I bet you can guess which union it is.
The blue column indicates the NBC/WSJ percentage of union membership among respondents, while the red column indicates the official numbers as per the U.S. Bureau of Labor Statistics. As you can see, the poll found 24% of respondents were union members, while according to the BLS, only 11% of working-age adults currently belong to a union.
Occupational Disparities
Because of the politicization of our government and journalistic institutions, it’s almost impossible to tell who does what or what the genuine unemployment rate is.
One might think lending institutions might want to know these facts reliably, but since Wells Fargo was just found to have opened more than 2 million fraudulent credit or loan accounts: I guess not.
So, the very significant disparities between actual occupations according to the Bureau of Labor Statistics, and those found in the NBC/WSJ survey may either be real — or not!
The blue column depicts the numbers reflected in the NBC/WSJ poll, while the red column reflects the Bureau of Labor Statistics (BLS) most recent data. As can be seen, professionals and managers are much more greatly-sampled by the poll, while blue collar workers, stay-at-home parents and even the officially unemployed are undersampled. Retirees and professional or managerial employees are oversampled. These groups do vote more than others — but not by as much as 50 to 400% more.
Ethnic, Cultural and “Racial” Disparities
The NBC/WSJ Poll shows significant ethnic/cultural and racial disparities between the polled population and the overall population. While not all questions have been tracked over time by Hart Research Associates, the polling organization, these have.
The “Hispanic” sampling is markedly divergent from the overall adult population. “Hispanics” are markedly undersampled.
Red represents the general population; blue, the poll population. 7% fewer
Hispanics were polled than are in the general population.
The overall ethnic and racial sampling of the poll is again, a game of inches. “Other” racial and ethnic identities, such as Native American, Pacific Islander, Asian or mixed race, are undersampled by the poll, while whites are oversampled.
Voter Age Disparities
The NBC/WSJ poll is undersampling younger voters and oversampling older voters, with a few differences. First, voters in their 30s do not fit the pattern; second — voters in their early 60s, i.e. pre-retirement age, are also undersampled. While people tend to vote more as they age, particularly seniors, the trends in the poll simply do not reflect overall demographic trends.
2012 Voters in 2016
Ignoring the fact that there are millions more voters who will be voting the first time in 2016, as well as a number of 2012 voters who are no longer with us, the NBC/WSJ poll asked for whom respondents voted in the 2012 election. The results are markedly different from the actual results, with a convenient “Did not vote” line that does not agree with the number of respondents under age 25. In other words: it’s discrepant with both new voters who will be voting in their first election, and those who are deceased/no longer able to vote.
The verdict
Over the course of decades, the overall conclusions of this, and other polls that offer historical data, may somewhat be trusted.
In this case, the NBC/WSJ poll managed to obtain almost the same results three months apart: the end June NBC/WSJ poll found Hillary Clinton leading Donald Trump by 5 points nationally. The more recent September 19 poll found that Hillary Clinton led Trump by six points.
Yet the September poll had underlying demographics that differed from not only national demographics, but in subtle ways from general voting and nationwide employment and other statistics. The changes in each poll’s underlying demographics varied only by a few percentage points each way. The results were similar but the ingredients: different.
However these organizations figure the “margin of error,” there are notable disparities between this poll’s population, the general population, and the overall voting population and trends. In addition, this poll is discrepant within its own trendlines and track record over the past three months.
It’s fair to say that the underlying numbers look an awful lot like this poll is being conducted in the following manner:
1,000 registered voters are surveyed — the surveyors call until their sheet is filled up with the desired checkmarks and results. In June the checkboxes filled up one way; more recently, another.
Otherwise known as (picture).
Amy Sterling Casil is the author of 30 nonfiction books, 3 novels, 3 short fiction & poetry collections, hundreds of articles, and also is a former high-level nonprofit fundraiser and executive, and a business planner and developer, primarily for women and non-traditional owners. You can buy her most recent book (sci fi short stories all about women) via Amazon or the largest, oldest author publishing cooperative, of which she is a co-founder.
If you are *amazed!* at the statistics, analysis & charts, this is what Amy did for businesses before financialism, the clampdown, and other related (see pic above) occurred. So if you have a buck and want to try to get your business off the ground you can hire her to research for you.
