Friday, September 16, 2016

The Tale Of Two Polls

'It All Depends On Who Shows
Up to Play on Election Day!'
Your head must be spinning 360 degrees every day as you watch the news and see the various polls from the various states go up and down and jump all around.

These polls are not 'rigged'. The great ones are not, at least. Any pollster worth his salt will tell you outright they can write a poll question to elicit the answer and result they want for their client anytime they want.

However, they won't be in business long. Mainly because pollsters in political campaigns are like U2 reconnaissance planes over Iraq or J.E.B. Stuart riding around enemy lines for Robert E. Lee to determine the strength and weaknesses and most importantly, the numbers of the opposing forces.

Pollsters give key information to strategists and candidates as to that snapshot in time ONLY. They don't 'predict' or 'divine' who is going to win any more than a football prognosticator can tell you that one team on a particular weekend is going to win that particular football game 'for sure'.

The victor of the game in question depends solely on 'who showed up to play' that day. Most times, the team with overwhelming talent wins easily.

Sometimes, such as when the Appalachian State Mountaineers went into the Big House in Ann Arbor, Michigan on September 1, 2007 before 109,208 totally stunned Michigan fans, a team with lesser talent and resources upsets the favorite, this time defeating the Wolverines, 34-32.

App State showed up to play that day. UM did not. They lost. App State won.

The point of this is not to denigrate political polls. Most of the time they are right. Much as weathermen are 'most of the time' proven right when they say 'It won't rain today' or 'Tomorrow it will be hot'.

They are only 'wrong' if a huge tornado suddenly shows up or a massive thundercloud appears out of nowhere and proceeds to prove them 100% wrong.

It just depends on 'what happens that day'.

Which brings us to 'voter turnout'.

Pollsters are usually pretty good at predicting the presuming the past is prologue, that is. Most of the time, pollsters presume that because the voter turnout was a certain level with particular percentage breakdown of vote by age, sex, race, gender, and socio-economic factors in a previous similar election, say in the 2012 re-election of President Barack Obama,it is more than likely to happen the next time in the next similar election, such as in 2016.

That is why many polls early on seemed to overly favor Hillary Clinton. Many national polls just looked at the turnout and composition of the electorate in 2012, the last available data point for a presidential election in America, and assumed 'it will be just the same in every state in 2016'.

However, voting tendencies tend to change over time, ever so lightly every 2 years and a bit more every 4 years. If they didn't, we would still be voting for the Federalists and Anti-Federalists that erupted in Elections of 1796 and 1800, right?

What really WOULD be simply astounding would be if it was possible to accurately predict the exact percentages of who will actually show up to vote or vote absentee or by mail say 6 months to a year before any election.

It can't be done. Why? Because human beings are mercurial creatures. Sometimes they vote because of common sense and reason. Sometimes they don't.

Sometimes, they just don't like the sound or tone or tenor of a Presidential candidate's voice. Sometimes, people just don't like the way a candidate looks or acts or the way he or she dresses.

Sometimes, a candidate is so boring that millions of people just either 'forget' to show up to vote for him or her or they just plain don't have enough energy or interest to cast their one sole vote for them.

Here's two instances of where polls failed to adequately predict beforehand what the makeup and composition of the actual final voting electorate looked like:

  1. In 2002, we were working on the US Senate campaign for eventual winner, Elizabeth Dole.

    After a very long and very expensive campaign as we gathered together in Salisbury election night, our campaign manager came into a small room and said this to about 30 of us:

    'My advice to to to start drinking heavily. This is going to be a long, long night. We probably will not know the outcome of this Senate race until 4-5 am tomorrow morning. Every poll, both on our side and theirs and nationally predict this will be a race determined a a very few percentage points and maybe less than 25,000 out of over 2.3 million votes to be cast' 

    The polls closed at 8:00 pm.

    75 minutes later at 9:15 pm in the Salisbury depot that was serving as the election night headquarters for the Dole campaign, Candy Crowley or some other CNN reporter turned on her microphone, the camera lights were lit and said this to the world:

    'CNN is now declaring that Elizabeth Dole will win the US Senate seat in North Carolina 56-44%. Back to you, Wolf!' 

    She wound up winning by over 200,000 votes. In a race where EVERY SINGLE REPUTABLE POLL had the race almost exactly even just 3 hours previously.

    Had it been a presidential year, the margin of victory might have approached 500,000 since voting turnout roughly doubles when presidential races head the ticket.

    What happened? More people showed up than expected. Many more of them showed up to vote for Elizabeth Dole instead of her opponent.

    Next question.
  2. In 2014, NC House Speaker Thom Tillis was running against incumbent US Senator Kay Hagan who had knocked off Senator Elizabeth Dole 6 years earlier in 2008. (What goes around, comes around in politics)***

    Most polls, probably including his own internal polls, showed a tight race with Hagan ahead slightly...ASSUMING voter turnout in certain counties such as Wake and Mecklenburg would hit certain target levels, especially among the usually dependable African-American vote for Democrats.

    Senator Hagan came out of Mecklenburg and Wake Counties with over 100,000 more votes than Thom Tillis. However, the vote total margin of victory coming out of Mecklenburg County was 'only' about 56,000 instead of the 70,00+ target the Hagan campaign needed to combine with other margins of victory in the most populous cities and counties in North Carolina  to win re-election.

    Thom Tillis won the rural areas by comfortable margins across the state and won the election by 45,608 votes.
By most evaluations, Elizabeth Dole 'over-performed' with many sectors of the population across the state in 2002 based on election models and Kay Hagan 'under-performed' in 2014, especially with a depressed or 'much lower than hoped for or expected' turnout of the African-American voters in Charlotte/Mecklenburg and other large cities in North Carolina.

What is going to happen in 7 weeks here in North Carolina?

If you can tell us right now how many people will show up and what percent of the vote each voting category will break down into, we could tell you right now who was going to win each election without paying anyone to do an expensive poll for you.

If you think the voter turnout in absolute numbers and percent is going to be exactly what it was in 2008 when Barack Obama was running for President the first time, we would have to say Hillary Clinton would win the state and she might sweep in Roy Cooper for Governor and Deborah Ross for the US Senate.

However, the energy and enthusiasm that was on full display in 2008 (and even in 2012 when Obama lost NC to Mitt Romney by 80,000 votes) is not evident this time around for Hillary Clinton.

If there is a less-than-optimum turnout among African-American voters in the major cities in North Carolina this fall for Hillary Clinton, it is hard to see how she can win the state and become only the second Democratic candidate for President since 1976 (Carter) to win North Carolina's 15 electoral votes.

And if there is any kind of surprising 'surge' perhaps east of I-95 for Donald Trump a la the 2002 Elizabeth Dole landslide victory, then it is hard to see how Hillary Clinton could win the state's 15 electoral votes either.

One rule of thumb that has held true for decades if not most of the last century in North Carolina electoral politics is that if the conservative candidate can somehow make it out of the big cities and counties close to even west of where I-95 cuts south today, then they will roll up big margins of victory in the eastern part of the state and win statewide elections.

No one knows for sure if there will be a less-than-enthusiastic turnout among African-American voters for Hillary Clinton in North Carolina or not yet. Nor does anyone know if a 'surge' is building for Donald Trump in the more rural, agricultural and military-friendly base of eastern North Carolina.

It all depends on who shows up to play on Election Day. Just like when App State beat Michigan.

***(Side note and little known fact: NC has had 22 US Senators in Washington since WWII whereas other states such as South Carolina and Mississippi have had close to 1/3rd as many US Senators during the same time span)

Do You Want Better People to Run for Public Office?
Support the Institute for the Public Trust Today

Visit The Institute for the Public Trust to contribute today

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.