The Tribune

With CHARLIE HARPER

Public opinion polling here in The Bahamas is neither high art nor exact science. In fact, if we weren’t bombarded with references on American television stations to the latest “breaking news” from some poll or another in the US, we might not pay any attention at all to this often influential, lucrative but decreasingly reliable method of determining public attitudes on social, political and economic issues.

Some observers in the US still blame polls for Hillary Clinton’s stunning presidential election loss to Donald Trump five years ago. According to this narrative, Clinton was lulled to sleep by inaccurate last-minute polls in the key Upper Midwest states of Wisconsin and Michigan.

Believing the polls there foretold a comfortable victory for her, Clinton skipped any late pre-election appearances in these traditionally Democratic states. But she lost them, and Pennsylvania where the polling was also dicey, and the US got four years of Trump as its President.

Despite essentially blowing the call on the 2016 election, American pollsters remained mostly confident in their methods. They said the reason for the errors five years ago was the polls had not measured for the educational level of potential voters.

This meant support among white, less educated male voters for Trump was not accurately assessed and thus the poll predictions of election results turned out to be incorrect.

Pollsters figured out a way not to make that same mistake again. And they did avoid sampling errors on education in last year’s election.

But they still goofed. In fact, as pollsters’ performances in 2020 are being assessed in the aftermath of another disappointing performance, the polls at least forecast Joe Biden’s victory.

In polls in individual states, however, the results were disastrous, and have compelled the industry to review their methods once again.

According to a study conducted by the American Association for Public Opinion Research (AAPOR), public opinion polls in the 2020 presidential election were damaged by errors of “unusual magnitude”, the highest in 40 years for surveys estimating the national popular vote and in at least two decades for state-level polls.

In a press release, the AAPOR task force reported it had examined 2,858 polls, including 529 national presidential race polls and 1,572 state-level presidential polls. They found the surveys overstated the margin between President Biden and former President Trump by 3.9 points in the national popular vote and 4.3 percentage points in state polls.

In an election when nearly 160 million votes were cast, that amounts to missing the results by 6.5 million votes or more. Biden’s final margin was not too much larger than that.

It’s hard to conclude on this evidence that polling did a very good overall job in 2020 of correcting the inaccuracies and outright errors from 2016.

Speaking at a private conference recently in New York, well-respected polling historian and analyst Karlyn Bowman agreed. She added that one major American newspaper had concluded “polling is broken and no one knows how to fix it.”

Bowman, who has analyzed polling for more than 40 years, has worked at the conservative think tank American Enterprise Institute for most of that time, and has gained wide respect from her peers, even those many who disagree with her right-leaning politics. Most experts regard her as professionally non-partisan.

In addition to pollsters’ failing to sort for voter education level prior to the 2016 elections, Bowman said, the polls did not – and could not, given the pressure of deadlines – count the large late vote that year.

“Far more voters than usual made up their minds very late in the election cycle,” she said. “They made their decisions and voted after the pollsters had wrapped up their work. And on available evidence, these late-deciding voters appear to have cast their ballots overwhelmingly for Trump.”

Last year, the polls missed again. Bowman thinks a big reason for this is simply many Trump voters have become inured to regarding government officials, pollsters and other official and quasi-official individuals as basically untrustworthy.

Hearing Trump consistently lambast polls may well have hardened his supporters’ attitudes and led them to either decline to participate in opinion sampling or deliberately misrepresent their views, “if only to mess with the polls that they see as biased against conservatives in general and Trump in particular.”

While all 66 polls Bowman personally studied correctly forecast Biden’s win, only two-thirds got the Senate races right. This, she said, “is a huge problem for pollsters”.

A huge impediment to accurate polling in the internet and COVID age is the stark decline of in-person polling, Bowman said. During the pandemic, door-to-door traditional sampling disappeared. Polling by phone broke down as 70 percent via cellular and 30 percent via landline.

“When poll takers used landlines,” Bowman found, “it took over 25,000 calls to harvest 200 valid responses. When they called on peoples’ cell phones, it was also daunting. “People use caller-ID. The estimates were that 40,000 calls were required for 800 good surveys.”

That’s an unsustainable business model. As in-person polling becomes prohibitively expensive, Bowman said “polling needs to become innovative in ways that have not been previously imaginable. They literally have to reinvent their professional methodology”.

“Now,” she continued, “pollsters too often follow the lead of the national media. While polls in the ‘golden era’ of such legends as George Gallup were able to pinpoint attitudes and opinions, the polls now tend to get swept up in the general hysteria and polarisation of American politics.”

Someone asked how Joe Biden is doing so far in the polls. “I think he has very solid approval ratings, with especially high marks on his handling of the COVID pandemic. If the disparity in hospitalisations and deaths from the virus continues to widen among vaccinated and unvaccinated, the Republicans’ embrace of anti-vaxism may hurt them badly.”

Bowman said Biden’s lowest ratings come on matters such as immigration, crime and, increasingly, concern about rising inflation.

That makes sense. The confusing mess at the southern American border with Mexico, regional concerns about defunding police and episodic rises in crime rates and the natural inflationary pressure of massive federal spending on COVID relief are all reasonable concerns in voters’ minds.

Biden’s ratings, Bowman said, are steady at five to ten points above Trump’s high-water mark of 47 percent approval. And she said there is some polling evidence Trump’s preponderant influence on the national scene is receding: more Republicans now identify as members of the GOP than as supporters of the former President.

Still, the spectre of big government and big spending loom over the political landscape. Bowman said “since 2013, a majority of people see government as a large problem, especially the federal government. State and local governments are often considerably more popular”.

Asked about the dampening effect of polls on voters’ willingness to actually go vote, Bowman said this is much more apparent in Britain and other relatively smaller democratic populations. “I see little evidence that there’s much of an effect of polls on American voters,” she said.

Someone asked if pollsters would do better in 2022 than in 2020 in election forecasting and voter assessments. Bowman just smiled.

It’s clear political parties and the news media will both have to hope the pollsters can solve some of their most vexing problems in the next year. If they don’t, the political business in the United States will have to seriously examine how to figure out what voters want.

Have they no shame?

The American House of Representatives is finally convening its own Commission of Inquiry on the hot topic of the January 6 riot at the US Capitol. As is often the case these days in American political life, attempts to make the commission bipartisan and objective have run aground on Republican attempts to wish away a profoundly disturbing and embarrassingly sordid episode in the arc of US history.

As House Republican leaders continue to debase themselves in the fading hope that by aligning with former President Trump they can recapture control of their chamber next year, Democrats will have an opportunity to tar them anew with the sticky sleaze of the former administration.

There are bound to be numerous good theatrical moments during the hearings, and perhaps the Republicans can be made to pay a price for their cravenness.