Polls 6 month before an election are not very accurate!

With the mainstream media continuing to hype recent polls suggest that Vice President Biden is likely to defeat President Trump in November, we decided to look back at the historical record of how accurate political polls are 6 months before a national election. We’ll cut to the chase. They are not very accurate. Let’s take a look at a few Presidential elections.

Let’s start with 2016. Given their long track record of polls, we will use the Gallup polling data. At the beginning of the Summer in 2016, Hillary Clinton had a comfortable 5% advantage over Donald Trump. She held this throughout the Summer and the lead even widened during August to 6% with the final Gallup poll showing her with a 4% lead on the eve of the election. Just days later, Trump would go onto win a decisive 304 to 227 electoral victory.

In 2012, Gallup conducted six polls in April and May of Governor Romney against President Obama. Romney led in 4 of the polls, with Obama leading in 2. During the summer of 2012, Obama gained ground taking the lead in most of the Summer polls, but by October Gallup said it was a dead heat with the two running neck and neck. Gallup’s final poll in November had Obama leading by 3%. The final result had a 4% advantage for Obama. The same thing happened in 2008. Gallup conducted three polls in May 2008 and McCain led Obama in two of the three polls, but in November, Obama would defeat McCain by nearly 8% of the popular vote.

So in the last three Presidential elections, the candidate that led in April / May actually lost in the general election. Of course, there have been times where the candidate ahead at that time held on to win. George W. Bush held slim leads over both Al Gore and John Kerry in 2000 and 2004 six months before the election and managed to win both in very close elections. Likewise, Bill Clinton held a commanding lead over Bob Dole six months before and won easily on election day.

Then there are elections where the results change dramatically. There is perhaps no better election to make this case than the Presidential election of 1980. In Gallup’s first polls of 1976 conducted in February, President Carter had a commanding 60% to 31% lead over Ronald Reagan. After John Anderson entered the race as an Independent, Carter’s lead fell to 8% in May 1976. As late as October of 1976, the race had narrowed but appeared to favor Carter as he led in two of the three Gallup polls conducted that month. On election day, Reagan won by almost 10%, winning 489 electoral votes. Talk about closing well!

So with 6 months to go before the general election, it’s clear that this race is wide open. The current polling favors Biden, but Biden has a relatively weak record of exciting the electorate and drawing voters to the polls. With Trump, we only have his record in 2016 and the 2020 primaries, but in both instances Trump has tended to outperform predictions, particularly in terms of getting the Republican base to vote. For this reason, we think Trump is more likely to be like Reagan, Obama, and Bill Clinton: a candidate that outperforms their pre-election polls. Accordingly, we put very little weight in the current polls suggesting that Biden is likely to defeat Trump by 7% or 8%. Six months is an eternity in politics and with Trump’s record of outperforming polls, we continue to believe this election will be very close but would continue to give Trump a slight advantage.