May 13, 2014 by Thomas Hauber
I wonder if anybody said that to Presidential candidate Truman when he went to bed early, at 6:30 PM on election Day 1947. Harry was hopelessly behind in every poll including the Gallup Poll, arguably the world’s most famous after Gallup successfully called the 1936 Presidential election for FDR over Alf Landon.
They were all wrong that night even the Chicago Tribune that had pre-printed the Nov 3rd issue with the headline “DEWEY DEFEATS TRUMAN.” George Gallup, founder of the firm according to his son was forced, “…to visit many newspaper clients after the election to lure them back after 30 canceled their poll service.”
Gallup’s record has since been perfect until the 2012 call of Romney 49%, Obama 48%. You can’t blame George for blowing it in 2012, he died in 1988 and the company was sold to an outfit called Selection Research, Inc., yet the company retains the Gallup name.
Americans love to poll each other. Today Gallup is one of over thirty large polling companies that poll personal opinion on everything from politics to personal well-being. There are even polls asking which polls are the most accurate. Pew research published a report basically saying most people don’t trust anybody. It’s no secret academic polls are trusted more than media polls and even more than polls sponsored by political parties.
Most of the larger political polls share about an equal claim to accuracy, though from time to time, they are charged with varying biases. Rasmussen has had to fight a claim of Republican bias while PPP (Public Policy Polling) fought a Democratic bias in 2012 even though its charge of pro-Obama sentiment was vindicated by the election.
Most private company pollsters depend on accuracy because their results are often purchased by other companies and political campaigns. Harper Polling was founded by a past member of the NRCC in order to get a better true reading of the electorate than what was being perceived by GOP party insiders.
True accuracy is based on methodology, sampling and timing. Who to sample: voting age, registered voters, or likely voters. How to stratify the sample: by age, sex, race, etc. How to take the sample: by voice or Interactive voice recording (IVR). Finally how to weight the results and most importantly when the sample is taken.
It’s generally agreed Gallup’s incorrect 1948 presidential survey was completed too early, weeks before the election and without paying attention to sampling constraints. Many voters changed their minds late in the game. It goes without saying the closer to Election Day, the more accurate the poll is likely to be.
In an off-year election it is unusually difficult to screen for likely voters. The two parties as usual approach things differently. It appears Democrats seem to rely on voter history, while Republicans love to count on voter enthusiasm. The eventual turnout is always a bit of a mystery.
Polling becomes an ever-changing numbers game, a voter’s version of a racetrack tote board, a snapshot of what the voters think at any one time right up to the call of the race. The results can surprise even the best Las Vegas bookie. The final tally usually presents a surprise or two, usually provided by the mysterious undecided voter. Many times he is the smart money gambler who waits till the very last moment to put his money down, knowingly or unknowingly playing the odds.
The poll itself can determine both voting behavior and campaign strategy. The bandwagon effect is real, undecided voters often go by default for the likely winner. Conversely, being behind can energize a campaign to increase voter turnout, likewise being too far ahead can discourage continuing campaign efforts.
The trend is always your friend. Being slightly behind but gaining, like the horse perfectly positioned for the stretch run, can invoke the powerful “comeback kid “ scenario.
All things being equal it often comes down to which candidate can avoid the last minute stumble or faux pas that sabotages the campaign. Less one get too enamored of bookmaking, Campaigns are more like horse races, everybody loves a close one. Failing that, winning wire to wire is always preferred, but unlike bookmaking in other professional sports, winning is everything and you don’t make a dime by beating the spread.