Do You Believe in Election Polls?

Election Day 2018: a day that holds magic keys to a large part of the immediate future of the United States of America. Today, millions of Americans — and apparently many more than in normal midterm elections — are headed to the polls to decide (among numerous state and local offices) who will represent the American populace in 435 chairs in the House of Representatives along with an assortment of U.S. Senate seats. This will probably go down in history as the most contentious midterm election in the U.S. And many are already saying it is the most important election in our lifetime.

This election just like most is driven by media polls. Media polls are as different from each other as the North is from the South. Each polling entity claims to have the most accurate results, the fairest and most comprehensive samples as the basis for their numbers, and they all poo-pah their competition. But are they accurate?

The Polls

Take a look at the Real Clear Politics list of polls from the day before the 2016 election and on election day:

(right-click to download poll)

I know it’s tough to read because there are so many lines that include the polls from multiple polling companies and their data summaries. The point to make with this is illustrated best by looking at the far right column in which Real Clear Politics shows the final numbers for each poll on each day based on the party color of the candidate that is ahead. Note that in that far right column, almost every poll is blue — Democrat. And that means Hillary Clinton. The few lines that show red final numbers are Trump’s. According to every national polling service but one, Donald Trump was to lose in a landslide to Hillary Clinton.

Leftists explain this away by stating that the polls showed only the popular vote. (Hillary reportedly won the popular vote by over a million votes.) What those same leftists do not want to mention is that we do NOT elect presidents by popular vote. The electoral college determines the ultimate winner in presidential elections.

Polls showed that Hillary Clinton would win by a landslide in the electoral college as well. The Real Clear Politics summary poll of the electoral college projections showed Hillary would win that vote 331-207. Trump won the electoral college AND the White House 306-232. How and why do these guys miss so badly?

Let’s just be completely honest:

  • There are 335 million people in the U.S. How can any polling company put create a fair template that actually models accurately a sample of American voters? It is a virtually impossible task, as their results prove. What’s a sample size that should be considered a “fair” sample? What mix of cell phone and landlines to use in contact is truly representative? What percentage of party split is accurate in the chosen geographic region of the sample used?
  • How do they create their samples? The term we have all become familiar with as a result of the Facebook scandals is “algorithm.” Every polling entity prepares a secret algorithm (or “formula”) that supposedly factors in each variable that is integral in accurately estimating who to poll in the polling process which includes specifics of those polled and the compilation of their polling results. That algorithm must include accurate estimates of likely voters within each of these classifications: male or female, age, ethnicity, income level, employment status, homeownership status, education level, voting history, marital status, and, of course, political party affiliation.
  • What days of the month and/or days of the week should polling calls be made? Who should make those calls: male or female, African-American or Caucasian; English speaking or Spanish or “other;” which area codes and which exchanges to include; the actual questions that are asked; how many questions; how long should each call be; the time of day to place calls. Often those called refuse to answer certain questions asked. How many answered questions must occur in each call to include those  results in polling data?

There are many more questions to be considered.

Polling Companies

Here’s an alphabetical list of the 22 best known national election polling companies. Included are the hyperlinks for you to use. If you want to look at any of these companies’ websites, just click on the highlighted company name to be linked to their site. It is noteworthy that there are dozens of more such companies in existence. Most of those are specialty or “niche” polling services that specialize in certain types of polls. Those have not been included.

The purpose of listing these is to illustrate just how many “reputable” companies there are who extrapolate data from likely voters in each election — local, state, or national — to fulfill specific polling requirements. Why do these companies go to such trouble when everyone knows that poll results are at best sketchy, at worst totally unreliable?

Virtually every campaign organization budgets some amount of money to do their own “internal” polling. Most candidates foam at the mouth for the status of probable votes during each campaign. To get those results, someone must obtain the numbers and compile the data. National campaigns spend millions of dollars in each campaign cycle for polling data.

It is strange to me that while knowing just how unreliable is most polling information given by the polling entities, campaigns continue to beat the drum called “Numbers.” Why is that?

Good polling numbers mean good campaign contributions: “Follow the Money!”

Poll Results Morph into Campaign Bucks

It has always puzzled me why Americans allow this election campaign financial boondoggle to continue. Yet every two years, we all signup to be willing participants in it. How much do these campaigns cost?

The final price tag for the 2016 election is $6.5 billion for the presidential and congressional elections combined, according to campaign finance watchdog The presidential contest — primaries and all — accounts for $2.4 billion of that total. That means that in the 2016 Congressional election process, candidates spent $4.1 billion. The 535 members of Congress each spent an average of $7.663,551.40 in their campaigns. But wait: not every Senate seat was in that election. That means that the actual dollars spent for each seat for which there WAS an election was higher than $7.663,551.40!

Let’s look at Congressional dollars and cents another way: how much in salary and benefits do members of Congress make?

  • The current salary (2018) for rank-and-file members of the House and Senate is $174,000 per year.
  • Leaders of the House and Senate are paid a higher salary than rank-and-file members.
    Senate Leadership

    Majority Party Leader – $193,400
    Minority Party Leader – $193,400

    House Leadership

    Speaker of the House – $223,500
    Majority Leader – $193,400
    Minority Leader – $193,400

Members of Congress receive retirement and health benefits under the same plans available to other federal employees. They become vested after five years of full participation. Since all provisions of the Affordable Care Act or “Obamacare” took effect in 2014, members of Congress have been required to purchase health insurance plans offered through one of the Affordable Care Act-approved exchanges in order to receive a government contribution toward their health coverage.

Average Annual Pensions

According to the Congressional Research Service, there were 611 retired members of Congress receiving federal pensions based fully or in part on their congressional service as of October 1, 2016. Of this number, 335 had retired under CSRS (former retirement program) and were receiving an average annual pension of $74,028. A total of 276 Members had retired with service under FERS (current retirement program) and were receiving an average annual pension of $41,076 in 2016.


Each member of Congress receives an amount of money each year — an “Allowance” — from which support staff personnel and all office and travel expenses are to be paid. In 2016, for example, the office personnel allowance for each member was $944,671 to be spent on up to 18 staffers.


No doubt, some will shake their heads when realizing just how much money those 535 members of Congress cost taxpayers — not to mention the additional expenses required to operate Congress and the U.S. Capitol. (See  “The “Bottomless Pit:” The U.S. Congress published here at TNN on October 18, 2018, to get exact costs of everything to taxpayers for Congressional operations) But even knowing that, how can anyone in their right mind obligate themselves every two years to raise an average of $7.66 million to pay for a campaign to obtain a job that at most pays them $200,000 – $250,000 per year? Why would they?

The only logical answer again is “Follow the Money.” The “off-ledger” financial benefits of serving in Congress are staggering. (again see details in that October 18th story) It’s for the “whole banana” they do it. That’s why it is so important to raise all that campaign money — not just to pay for their actual costs of simple campaigns. They must raise and spend those millions  to secure election victories, often with little or no consideration of policy matters. And far too often, the concerns and desires of the voters from the districts for which those candidates seek election are seldom concerns to the candidates, if they are ever that important!

Now you understand just why election polls are so important and therefore attract dozens of companies that charge millions of dollars to turn out results of voter preferences.

You see above just how accurate those polling entities were in the runup to the 2016 elections: they were horrible. What do they think about election results for these midterms? I’ve posted the Real Clear Politics numbers released today. (see below) Save those and Wednesday compare their accuracy to the inaccuracies across the board from 2016.

You’ll see how little accuracy in projections of election results really are. And you’ll understand why Americans should summarily disregard election polling results — PERIOD. They are all “About the Money!”

Real Clear Politics Monday, November 5, 2018

(right-click to download poll)

Wanna Know Who’ll Win the Midterms?


I received a bunch of feedback about the “tease” yesterday about today’s story. I simply thought your weekend news digesting before college football would be piqued by a few minutes away from the political fray we’ve all been immersed in for so long. Even though this story today is about the midterm elections, it’s outside of the hype of polls and polling data.

In our non-stop research I uncovered an interesting entity called FairVote. What is SO intriguing to us about them is that they are NOT pollsters, NOT political hacks, really are “non-partisan,” and a scientific analysis entity with no “dog in the hunt.”

As you go through the data, explanations of how they operate and look at their projection record, remember this: the political landscape of America is changing dramatically. In the last decade, in fact, it has morphed from something that is honestly a mundane practice of democracy to something that closely resembles a Barnum-and-Bailey circus! (At least it is certainly a sideshow) In way of explanation, all one must do is look back a couple of years at the 2016 presidential election and all the hoopla leading up to it to understand this: election projections are apparently NOT a science — even if there is science to enable reasonable and realistic outcomes. Why is that?

Does anybody reading this today not yet realize the zoo we call elections is without question an economic boon for every media company in America, whether that company is an actual news broadcast or reporting entity, a political lobbying firm, political party or Political Action Committee (PAC), or any entity in a support capacity for any of these. “Follow the Money!” Realizing that makes it a bit easier to understand why political polls, analysis, and projections are always all over the place: dollars drive results.

FairVote is a not-for-profit entity that uses scientific analysis methods (detailed below) that have resulted in some mind-boggling accurate results in Congressional elections — only in the House of Representatives. I found it curious they focus primarily on just House races, and not the Senate or presidential elections. But in their computations, they do use all previous federal election results to set a pattern that factors heavily in their calculations.

Enough of my drivel! Meet FairVote.

I’ll see you back here for a wrap-up.

FairVote and How it Works

FairVote’s methodology projects with high confidence only the very safest seats. With 435 seats elected every cycle, you might think that means 50 or 60 incumbents. Think again. This year they are projecting 374 seats with high confidence – that’s nearly 86% of House seats. Almost every incumbent seeking re-election can feel very confident about victory in November 2018, no matter who their opponent is, how much is spent, or what kind of partisan wave there might be.


To underscore our level of confidence, they made similar projections going into 2012, 2014 and 2016 in a total of 1,062 House races. They missed only one seat – that means their high confidence projections have an accuracy rate of more than 99.9%. The 2018 report shows the most ossified electoral landscape yet, being the first year they have projected more than 370 seats at this degree of confidence. The following map shows each congressional seat as an equal area. Only the yellow seats are in play; the purple seats are all safe enough to be projected with high confidence.


In addition to the 374 high-confidence projections, FairVote also projects favorites for the other 61 seats, though at a lower level of confidence. There are an additional 40 seats that clearly favor one party over the other, but not enough to warrant a projection. That leaves only 21 true “toss up” seats that only very slightly lean to one party. When they projected all 435 seats in 2016, they were remarkably accurate, even in the lower confidence projections. Of the 56 seats FairVote did not project, but which favored one of the parties, they were right in 50 (89.3% correct). Of the 18 seats that were true “toss-ups” with only a very slight lean toward one of the parties, they were right in 12 (66.7% correct). That means the full projections were correct in 423 out of 435 districts (97.2% correct). Those projections were made more than two years before the 2016 elections.


What makes this all the more disturbing is that FairVote’s remarkable accuracy ignores all polls, all demographic characteristics of the districts, and ignores the incumbent’s voting record and any scandals. They use only the presidential election results (both in the district and nationally) from 2016, and the incumbent’s performance in prior elections. The only updates they make after receiving have that data is to remove incumbents when they announce that they will not seek re-election and to recalculate projections if a state redraws its district lines. FairVote explains their methodology in the report in detail, but the overwhelmingly important factor is a district’s partisanship, measured only by the relative presidential vote in that district.

The two-party preference has varied over the years from 54.3% for Democrats in 2008 to 53.5% for Republicans in 2010; in 2016 it was 50.51% for Republicans. The “incumbency bump” that shows the extra percentage points earned by an incumbent in a given year has ranged from a high of 7.7% in 2000 to a low of 2.8% in 2014; In 2016 it was 3.3% percent.

Their state-by-state analysis drives home the entrenchment of incumbents under the current system, underscoring why it was no surprise that more than 98% won re-election in 2016. In more than half of the states, FairVote project every single House seat. Every incumbent seeking re-election is projected as safe in a total of 27 states. Many of these are large states with multiple seats, including Ohio (all 16 seats safe), Georgia (all 14 seats safe), and North Carolina (all 13 seats safe). They project a majority of the seats in every state except Delaware and New Hampshire, which only have one and two seats respectively.

The way we elect representatives in Congress does not create a fair reflection of the voters who elect them. There are 19 states where FairVote has already called more seats for one political party than that party should earn according to its statewide partisanship. In fact, there are three states where they can already safely project that one party will earn at least three seats more than the state’s partisanship suggests they should win (North Carolina and Ohio for Republicans, and Massachusetts for Democrats). If using projections for every seat, fully 32 of the 50 states would disproportionately favor one party over the other in a 50-50 year.

As a whole, the national landscape tilts in favor of Republicans, with Republicans sitting on 208 safe seats, only 10 away from a majority, and 22 additional unprojected seats favoring Republicans. Looking at the “tipping point” median district, FairVote projects that Democrats would need to earn more than 55% of national two-party preference among voters to earn even a one-seat majority. That imbalance creates a core problem of accountability. Power is exercised most responsibly when those in power believe they might lose if they cannot keep majority support.


Before anyone gets emotional distraught or hillariously happy, please understand that these projections are based on logical facts derived from various scientific and historical events. They do NOT have any basis other than the past. They have no basis drawn from polls, campaign ads, current political events like U.S. foreign policy, the economy, or even on the immigration issues and policies now front-and-center because of the Central American caravan headed to the U.S. southern border.

Will there be a blue wave as Democrats have declared? Will there be a red wave that shocks most of America? Will Nancy Pelosi as she has famously declared in the last 48 hours take back possession of the House leadership gavel? NO ONE knows for sure — not Pelosi, Schumer, McConnell, Paul Ryan, FOX News, The New York Times, or any candidate, not even President Trump.

So what can you do about election results? There’s only one thing that matters, and one thing only: VOTE. Americans — both those Americans who vote and those who do not — will determine the outcome of this election as they have in every election.

The constitution of the House of Representatives is not the only political institution or office that is stake in these midterms. The U.S. Senate, governorships, and many local and state elections are at stake. The best shot for every American to have a say-so in the outcomes of each of these races is to VOTE.

And if you don’t vote, you have no right to complain about the results of these House and Senate races or any other race in your state or municipality.

Will the House race results prove to be another successful analysis by FairVote? Tune in Tuesday evening, November 6. We’ll all know for sure!