As we consider the flood of polls proclaiming that President Trump’s approval numbers are lower than those of any other president at this point in his presidency, a former Clinton pollster and current co-director of the  Harvard-Harris Poll takes issue with the majority of political polls’ methodology.

Mark Penn, writing at the Hill, refers to the problem as a “polling bubble” that involves both polling entities and the media.  He pinpoints three reasons for the unreliability of polls about Trump:  polls directed at “all adults,” polls focused on sensationalized “stories,” and participants’ unwillingness to be honest about their views.

In his article, “Why the polls are wrong about Trump.  Again,” Penn writes:

Today we live in a polling bubble – surveys taken from the perches in New York, Washington and Los Angeles may be obscuring rather than illuminating many of the underlying views and trends of the American electorate.

How else can one explain that although many polls showed a close race last November, almost no one (myself included) predicted a lopsided victory for Donald Trump in the Electoral College. Most media analysts and modelers concluded a Hillary Clinton victory was in the bag. One Princeton professor even agreed to eat a bug if Trump won.

As President Trump enters his 100th day, several of the same organizations are using their polls to proclaim that he has had the worst start of any modern president and the worst ratings of a president at this time in his presidency. While Trump is no FDR when it comes to forming a political coalition, a fairer reading of the polls and the election results shows his performance is probably 5 or 6 points better than is being touted and that his base of support with which he won the election remains intact.

Penn goes on to explain that most polls have moved away from polling likely or even registered voters, that the questions are focused on sensationalized narratives from the often anti-Trump media, and that voters are simply unwilling to share their actual views about the president with pollsters they regard with (well-earned) intense distrust and skepticism.

There are several reasons for this mismatch between likely reality and the interpretations we are seeing. Most polls have moved away from voters or likely voters to U.S. adults with no screen for registration or even citizenship. And the questions often focus on storylines and narratives critical of Trump. Rarely are they written from the perspective of having missed the major swings and economic discontent that upended the election.

The current crop of stories also sets Trump ratings expectations, as though America went through the typical process of coming together around the winner. Instead we had recounts, Russian conspiracies, investigations and rallies unlike any seen after any election. The country was sharply politically divided on Election Day and remains that way today. That is the backdrop of any realistic assessment of what is happening in America.

. . . . [T]he media echo chamber has, I think, made it more difficult for people to express their political views, especially to live interviewers. With the growing gender gap, I’m not sure most men are even telling their spouses or partners what their real views are on the president. In a recent Harvard Harris poll we did, only about 60 percent in the country now feel free enough to express their views to friends and family.

Ultimately, Penn concludes that Trump’s approval is probably at about 48% among those who voted in November.  (As an aside, this conclusion undermines Obama’s catty trolling of Trump by “pointing out that Obamacare is now more popular than his successor trying to repeal it.”)

All of this rings true.  Indeed Penn is not the first to argue along these lines.

Nate Silver at FiveThirtyEight has also called polls regarding Trump into question.

Back in February, he wrote:

The differences between the polls aren’t random, or at least they don’t appear to be based on the relatively limited amount of data we have so far. Instead, Trump’s approval ratings are systematically higher in polls of voters — either registered voters or likely voters — than they are in polls of all adults. And they’re systematically higher in polls conducted online or by automated script than they are in polls conducted by live-telephone interviewers.

I’ve seen at least as much cherry-picking from liberal and mainstream reporters. In my Twitter feed last week, for instance, a Pew poll that had Trump at 39 percent approval got a lot more attention than a Fox News survey which had him at 48 percent instead.

In some ways, the pattern reflected the one before November’s election, when reporters and pundits selectively interpreted the evidence and assumed that Hillary Clinton was a much heavier favorite than she really was based on the polls. Trump is not very popular, but he’s also no more unpopular than Barack Obama was for much of his presidency. If his numbers hold where they they are right now — especially among registered voters — Republicans would probably hold their own in 2018, and 2020 would be another highly competitive election.

It seems that the same outlets that erred so spectacularly in November haven’t learned a thing; indeed, they appear to be doubling-down on the very methods that led to their failure to predict Trump’s win.