Commentary
French Newspaper Dumps Polls
- by Erik Sass , Staff Writer @eriksass1, January 3, 2017
Now the crisis of credibility is beginning to have a real impact. This weekend, a leading French newspaper revealed it will not be publishing poll results for the country’s upcoming elections.
Stephane Albouy, the editor-in-chief of Le Parisien, told France Inter radio the newspaper will stop commissioning its own polls, although it may still reference the results of other polls in the lead-up to France’s 2017 presidential election, according to Agence France Press, which republished parts of the interview.
Albouy portrayed the move as an attempt to re-position the newspaper with more first-hand reporting: “We have decided, and it was the subject of a lot of debate, to return to the heart of our profession which is working on the ground.”
However, it’s not hard to interpret the subtext of the statement. In addition to being relatively abstract, political polls are increasingly wrong, and publishing poll results merely serves to mislead readers.
Even more damning is the implication that journalists, while largely untrained in statistical science, can still do a better job getting a sense for the overall political mood— even if, in large part, it is through anecdotal reporting.
In addition to the high-profile failures noted above, political polls also missed the outcome of the conservative Republican presidential primary in France, which gave the candidacy to Francois Fillon in a surprise result in November.
As noted previously, in the final days leading up to the 2016 U.S. presidential election, multiple polls had Hillary Clinton leading Donald Trump in key swing states, including Florida, Ohio, North Carolina and Pennsylvania, not to mention supposedly safe Democratic states, like Michigan and Wisconsin.
Despite razor-thin margins, all voted for Trump.
Erik Sass is editor of Publishers Daily. You can reach Erik at eriksass@gmail.com.
I'm politely asking, ... no, begging, ... someone to research, write and publish an article on the actual mechanics of poll-taking, since I'm convinced that the methods used are the primary reason for the inaccurate results.
Cold-calls, to land-lines, using questionable call lists, often lasting ten minutes or more, in the internet age, seems a certain recipe for questionable numbers. And yet that out-dated polling methodology, and others like it, seem to still be the norm.
The only conclusion seems to be that either a whole lot of people have somehow decided spontaneously to lie to pollsters, or the process needs to be changed.
Chuck, it's a well known belief that a certain percentage of respondents---the amount varies by polling methodology as well as subject matter---either lie deliberately or, more commonly, give misleading or inaccurate answers. So an analysis of methodologies, alone, might not provide a clear answer as to why we were provided such bad results regarding Trump's chances in the last election. I happen to believe that a fairly high proportion of the white, blue collar respondents in states like Michigan, Ohio, Iowa, Pennsylvania, Wisconsin, etc. as solid union people and long time Democrats, were reluctant to affirm their leanings towards Trump, despite his reprehensible campaign behavior and offensive history. They probably were concerned least others in their circle of family and friends---mostly dedicated Democrats---would think less of them for defecting as a "protest" vote. Sadly, I also believe that voting for a woman president also played a role in the false answers that the pollsters got from many male respondents, again, mainly from the blue colllar precincts. Here, too, this was not something one wanted anyone to know. Of course, these are just theories, but I believe that they have some merit---pending some study that gets at the real reasons---if such is possible.