- The Washington Times - Monday, September 25, 2017

ANALYSIS/OPINION:

How soon they forget. Heartened by a stream of poll data suggesting that the public is less than enamored with his performance as president, Donald Trump’s critics who’ve been taken in by polls before seem to think they have the man on the ropes.

The polls they are celebrating today are the same ones who a year ago said there was no way the brash New Yorker could beat Hillary Clinton. That experience and the cancellation of her election-night party should have been enough to warn them off a reliance on data such as they are gleaning from today’s samples. They should remember that those who fail to learn from history and their own mistakes are often doomed to repeat both.

It is both a cliche and true that any poll represents but a snapshot in time which, if accurate, can only tell an analyst how respondents feel at the time the poll is taken. Mrs. Clinton’s lead over Mr. Trump proved illusory for two reasons: Her numbers collapsed just before Election Day and the optimistic numbers on which her campaign advisers were relying were drawn from samples based on a misreading of just who was actually going to vote.

Whether Mr. Trump is in real trouble is a question that simply cannot be answered by a simple look at his approval rating. Consider Ronald Reagan, who at this stage of his first term, was far more popular than Mr. Trump, but whose approval numbers collapsed going into early 1984 as he faced re-election.

At this stage in his first term, President Reagan enjoyed an approval rating of 59 percent, but those numbers headed south, and by the next year he was at 40 percent on his way to 35 percent in January 1983. The talking heads of his day were predicting disaster in 1984 and Republicans on the hill were beginning to go wobbly, as his friend Margaret Thatcher might have put it. But the Gipper was re-elected in 1984 in a landslide and left office four years later with an approval rating above 63 percent.

Pollsters rarely consciously rig their polls to produce inaccurate results, but it is in drawing their samples and guessing who to query that sloppiness and bias make their appearance. This becomes more apparent as individual pollsters close to an election are forced to guess who will or won’t vote on Election Day, but it affects the results they produce even earlier.

Right now few pollsters are screening for “likely” voters. Most of them are polling a much larger universe often without reference to whether the people they are questioning vote at all, and this can give a skewed impression of relevant attitudes among the voting population.

In the last election, 32 percent of those who voted were Republicans, 34 percent Democrats and the rest unaffiliated or independent. Republican pollster John McLaughlin claims that any poll that fails to approximate this breakdown skews the results and should be taken with a grain or two of salt. Of the major polls Mr. McLaughlin analyzes, only the Politico pollsters get things about right. The partisan breakdown of their sample gives Democrats a 4-point edge, which is pretty close to the two-point edge they enjoyed among voters on Election Day 2016. Other polls, however, rely on heavily skewed samples. The Economist relied on a sample that gave Democrats a 14-point edge, while Reuters and Gallup gave Democrats an 11-point and 7-point edge in their samples.

Mr. McLaughlin points out that Mr. Trump’s approval rating is higher in the Politico polls because the paper’s pollsters work from a more accurate sample. He says that in a recent poll his firm conducted, Mr. Trump enjoyed a 90 percent approval rating among Republicans while only 17 percent of Democrats approved his performance, which means that skewing the sample toward Democrats has an incredible impact on the results.

Independent pollsters seem to agree that many public polls are skewed toward Democrats; Raghavan Mayur, president of TechnoMetrica, an independent polling organization, recently told the Daily Caller, for example, that there is an inherently Democratic bias in the polls that most pollsters simply deny exists.

So the polls we are seeing today amount to little more than very fuzzy pictures of public attitudes at a moment in time, pictures that may have been photoshopped by those taking them and which history demonstrates are terrible at predicting what will happen tomorrow or in the next election anyway.

Trump supporters shouldn’t be spooked by them and his detractors would be wise not to rely on them.

• David A. Keene is editor at large at The Washington Times.

LOAD COMMENTS ()

 

Click to Read More

Click to Hide