We’ve voted in the primaries, the nominations are completed and now the general election looms ahead. Responsible citizen-voters are seriously studying the candidates and their positions on issues of interest to them. Hopefully, they are looking at the positions and voting records of those who have previously held office.
What I hope they are not doing is hanging a lot of credibility on the numerous polls that are reported.
We all hear much of the results of polls as the elections approach. Partisans breathlessly report those that agree with their hopes and play down those that have disappointing results. The media blasts the results of various polls as though they were handed down to Moses carved in stone.
We all hear much of the results of polls as the elections approach. Partisans breathlessly report those that agree with their hopes and play down those that have disappointing results. The media blasts the results of various polls as though they were handed down to Moses carved in stone.
Those who have never actively worked in the trenches of politics or in the heart of a major campaign are likely to have no idea of what goes on behind the “polling scenes.” Based on experience in campaigns, politics and polls, I caution people to exercise caution when accepting the results of any poll as gospel.
The public audience for poll results probably believes that the intent of a poll is to honestly gauge public opinion at a moment in time or track trends. In the best of all worlds, it would be true. Unfortunately, not all engaged in political campaigns have pure intentions.
Some operatives use polls to try to affect an election. Have you ever heard the term “push poll?” Do you understand the weighting of sample? Do you know to question how the poll was conducted before believing its results? Do you understand what the caveat “margin of error” really means? Do you believe that polls are based on scientific methodology?
If the term push poll is unfamiliar, let me give you an example: If you knew that Joe Edmund cheated on his dying wife and fathered an illegitimate child with his lover, would you be more likely to vote for him or for challenger Bill Downey, a fine family man?
Unscrupulous pollsters have been known to frame questions to get the results that a candidate or party uses to demonstrate strength in the race. You have to know how the questions were framed to know if the poll results are remotely reliable.
A reliable poll not only asks unbiased questions but includes a truly representative sample of those most likely to vote.
Obviously, a pollster can’t talk to every person of voting age. A professional polling organization determines a realistic sample based on demographics and other data. All adults, registered voters or those most likely to vote based on past performance? Party registration or no party affiliation? Age group? Gender? Ethnic affiliation?
Constructing the sample has to be done correctly if the intent of the poll is to reveal credible (as much as possible) results.
If you poll a greater percentage of Republicans in a district heavily Democrat-registered, how reliable is the result? If you ignore Independents (a growing demographic) who tend to swing back and forth from one election to the next, how reliable is the outcome? If you skew the sample towards younger adults, who tend to vote less frequently than older voters, is this an accurate sample?
Polling has become even trickier in this world of the Internet and shrinking use of telephone land-lines.
In the past, pollsters relied on people who walked malls and asked questions of individuals who were willing to self- identify in terms of age, party registration and so forth. More recently, land-line phone calls to a supposedly correctly identified and reliable sample were made.
It’s tougher today because of the increased exclusive use of cell phones and people who won’t take a call from an 800 number or one that they don’t recognize.
Although some polling is conducted through the Internet or text messaging, there is no way to correctly balance the representative sample when the opportunity is wide-open and the person responding to the poll questions is essentially anonymous.
Then there is that “margin of error.” Most polls are reported with a tiny print caveat at the bottom: the margin of error. It’s usually between 1 and 3 points, but polls have been publicized lately with MOE as high as 4.5%. This means that the poll can swing by that percentage in either direction.
Then there is that “margin of error.” Most polls are reported with a tiny print caveat at the bottom: the margin of error. It’s usually between 1 and 3 points, but polls have been publicized lately with MOE as high as 4.5%. This means that the poll can swing by that percentage in either direction.
Example: Romney 48 to Obama 45. Sounds like Romney is ahead? Maybe yes and maybe no. If the margin of error is claimed as plus or minus 3%, the actual results could be Romney 51 to Obama 42 (a big gap) or Romney 45 to Obama 48.
When the media reports a 1 or 2 point lead for a candidate in a poll with a 3 point margin of error, that is not something to get you horrified or complacent. It’s really a statistical tie.
All polls conducted whether conducted by independent polling firms or commissioned by news organizations, political parties and other interested groups need to be believed with a grain of salt. Again, who was asked? How likely are they to really vote? How were the questions framed? What is the margin of error?
As much as those conducting and reporting polls would like you to believe otherwise, and as hard as they may try to get honest results, their methods cannot be considered truly scientific. On top of everything else that can be wrong about a poll, some people don’t tell the truth.
There is no way even the best pollster can test for that.
Case in point: the exit polls conducted in the 2004 general election showed John Kerry in the lead, but President Bush won handily. The explanation is likely a combination of the following: some interviewed as they exited polling places lied, a disproportionate number refused to participate or the selection of voting sites at which to poll was poor.
And in our modern age that stretches out “election day” with early voting and extensive use of absentee ballots, there are many votes cast outside traditional precincts on a single day.
Perhaps the most important “take-away” is that if you see a poll showing an issue or a candidate you support either behind or ahead, don’t be disheartened or smug. Work harder. Although the most honestly and carefully conducted polls can be dead wrong, some will be correct.
Perhaps the most important “take-away” is that if you see a poll showing an issue or a candidate you support either behind or ahead, don’t be disheartened or smug. Work harder. Although the most honestly and carefully conducted polls can be dead wrong, some will be correct.
Yes, it’s all a bit technical and maybe confusing. But, if you are a voter who wants to understand what you are hearing and reading, you need this information.
Headline Surfer is Florida's first 24/7 Internet newspaper launched April 7, 2008, initially as NSBNews.net. It is based in New Smyrna Beach, covering Volusia County and throughout the Orlando Metropolitan Area via HeadlineSurfer.com, NSBNews.net and VolusiaNews.net. "Headline Surfer" is a registered trademark owned by NSB News LLC, for editorial, marketing and advertising purposes. All rights reserved.
