Donald Trump speaking on the fourth day of the Republican National Convention
Donald Trump speaking on the fourth day of the Republican National ConventionJoe Raedle/Getty Images

After he briefly led former Secretary of State Hillary Clinton after the Republican National Convention in mid-July, polls have increasingly shown Donald Trump falling behind his Democratic opponent.

The latest RealClearPolitics average of polls shows Mrs. Clinton leading Trump by 7.7 points, 48.0% to 40.3%, a margin that’s only grown over the past few weeks.

Not surprisingly, the Trump campaign has brushed off suggestions the Republican is currently trailing Clinton by a significant margin.

Lies, Damn Lies, and Polls

Yet even many voters have expressed skepticism over pre-election polls, with claims of bias, inaccuracy in past polls, and skewed demographic samples.

So just how reliable are polls, and what should we make of polling giving Clinton a decisive lead over Trump?

Presidential election polling, which celebrates its 100th anniversary this year, has had its ups and downs over the past century.

After the Literary Digest correctly predicted the results of the 1916 election with its groundbreaking mail-in poll, as well as the next four elections from 1920 to 1932, it infamously predicted Alf Landon would defeat FDR in 1936, despite what turned out to be a landslide victory for Roosevelt.

But with the advent of scientific polling models, such whoppers have been avoided. While close elections, like 1976 and 2000, have been difficult to predict, polls have usually reflected the general state of the race, indicating when races were very tight, and when one candidate had a substantial lead.

Polls versus Reality: How do they fare?

So why is there so much skepticism regarding polls this year?

To some extent, the aversion by some to polling has to do with a misunderstanding of the purpose and nature of polls.

Public surveys do not – and cannot – predict what will be on election day. They instead capture a moment in time, gauging support for the respective candidates at this particular moment. But campaigns can and do alter those levels of support – that’s the purpose of campaigning.

At this point, voters still have three months to consider the two candidates, watch three presidential debates and one vice-presidential debate, and be inundated with election ads (particularly in swing states). And of course major events – terror attacks, changes in the economy, scandals, etc. – could radically alter the trajectory of the race.

But as far as how the race stands at this ponit,, polls give a reasonably accurate picture, particularly when taken as an aggregate, thus reducing the effects of outliers.

This can be seen with the relative success of past polling – taken in the aggregate – in predicting the outcomes of elections.

An aggregate of the final polls by the different polling agencies in 1996 showed Clinton beating Dole 51% to 40%, close to the actual results of 49% to 41% in favor of Clinton.

Polling in 2000 accurately indicated a dead-heat, while in 2004 polls showed Bush beating Kerry by an average of 1.5 points (48.9% to 47.4%), close to the election result of a 2.4 point margin (50.7% to 48.3%). In 2008 the polls were even closer, showing Obama defeating McCain by 7.6 points (52.1% to 44.5%) compared to a 7.3 point margin (52.9% to 45.6%) in the election itself.

Are Biases Skewing the Polls?

Despite this, in 2012 many conservative pundits aired sharp criticisms of polls that year, suggesting the polls were skewed in President Obama’s favor, underestimating Romney’s support.

Believers in the “skewed polls” theory argued that the demographic models used in many polls overrepresented traditionally Democratic groups – like African Americans and Hispanics – and underrepresented non-Hispanic whites, who leaned towards the Republican Party.

The argument was a legitimate one – but proved to be incorrect. Pundits assumed that the spike in turnout among minority voters caused by Obama’s election as the first African American president in 2008 would subside – at least somewhat – in 2012, or at the very least, remain the same.

A significant segment on the American right become convinced the polls were simply wrong, that the lead Obama maintained over Romney was the result of inaccurate or biased polls, and that Romney was either tied with or leading the president.

The reality was that the polls were indeed skewed – in favor of Romney and against Obama.

While the aggregate of final pre-election polls showed Obama beating Romney by just 0.7 points, he won the election by 3.9 points, 51.1% to 47.2%, a wider margin than any of polls taken in the final week showed.

The ‘Silent Majority’ Theory

This election some again have suggested polls are inflating support for the Democratic nominee, arguing that just as polls ahead of Britain’s “Brexit” referendum missed the anti-establishment angst among working class voters, so too pollsters in the US are undercounting working class whites who may be turning out to vote for the first time.

While this argument is plausible, it is also unlikely.

Firstly, Brexit polls did not fumble the vote as some have claimed. Polling ahead of the referendum showed “remain” leading for much of the campaign, followed by a sudden surge for “leave” in the final weeks.

While the murder of a Labour MP by a British nationalist boosted “remain”, polls in the final stretch showed a deadlocked race. It was not the polls that failed, rather the conventional wisdom of mainstream pundits and polling analysts, who even while “leave” led in the polls, insisted the “fundamentals” of the race indicated otherwise.

Secondly, returning to the United States, the trend towards Clinton in the polls has been consistent. Every poll conducted over the past two and a half weeks shows Clinton leading nationwide, and her margins over Trump have only widened.

State-level polling is also consistent with a significant nationwide lead for Clinton. The Democrat now leads Trump in all of the states won by Obama in 2012, as well as North Carolina and Georgia, giving her a total of 362 electoral votes to Trump’s 176.

Is a Trump Comeback in the Offing?

While a dose of skepticism is always useful, it is fairly clear, given the degree of Clinton’s current lead and the consistency of that lead across polls, that at this point in time, she is ahead of Donald Trump.

With that being said, the election is decided in November, not August, and Trump still has time to make up lost ground.

Earlier in the election season he was boosted by terror attacks at home and abroad as well as mediocre economic growth. Current events could revive his campaign once again, as could a scandal revolving around something discovered in one of Mrs. Clinton’s emails.

The three presidential debates could also alter the direction of the race. Conservative pundits have also rightly noted that Trump hitherto has run virtually no advertising, compared to a barrage of ads run by the Clinton campaign to the tune of tens of millions of dollars.

The election has not yet been decided, and polling is not a crystal ball into the future. But one thing is clear: for the time being, Trump is most definitely the underdog in this race.