The Trump voters pollsters didn't see

Trump won with good margins in states where pollsters predicted he will lose by big margins. The pollsters seem to have got their community-targeting assumptions wrong, too

At the time of writing, Joe Biden is set to become the next US president though we’re days, perhaps weeks, away from this being official. The implications are positive for India for an elemental reason: Mr Biden is by political standards, an honest man. It is possible to deal with him in good faith. It is delusional to imagine a serial liar will deal with anyone honestly, whether at the personal level, or in realpolitik.

Donald Trump logged over 20,000 verifiable, public falsehoods between inauguration on January 2017 and July 2020 — The Washington Post kept count. The lies got so bad this week that TV networks repeatedly cut him off for lying, while Twitter and Facebook labelled everything he said “false”, or “misleading”. That’s quite apart from being at the centre of many accusations of financial irregularity. He may face multiple criminal charges once he exits.  

It was fascinating what the pollsters got wrong.  Most US pollsters assume a margin of error (MoE) of 3.5-4 per cent in projections. Averaging out polls should hold the MoE to within 3.5 per cent. We won’t have the full data for weeks with millions of votes still being counted. But it looks as though Mr Biden will win the popular vote with a spread of 4.5-5 per cent over Mr Trump.  That’s almost outside the MoE, considering polls averaged out at almost 8 per cent more for Mr Biden.  

There were huge errors at the state-level polls, where the Electoral College is decisive. Both parties strategise presidential campaigning according to the polls. Maybe they both got it wrong. The two candidates focus on so-called battleground states where they think there is a good chance of winning, or losing.

Florida was outside projections, which averaged 2.5 per cent in favour of Mr Biden. Mr Trump has won by a 3 per cent margin. Texas was supposed to be competitive, with the Democrats within 2 per cent of challenging the Republicans. Mr Trump won by 6 per cent. Iowa and Ohio were supposedly toss-ups. Mr Trump won both by 8 per cent. These margins may eventually reduce when all votes are counted. But the pollsters clearly got it wrong by much more than the MoE.

They seem to have got their community-targeting assumptions wrong, too. It is a given that African Americans and Hispanics will vote Democratic. This was really taken for granted in 2020, what with the Black Lives Matter movement and the Hispanic refugees sitting in internment camps, separated from their children.

Well, Mr Trump clearly made serious inroads into the large Cuban-emigre community in Florida, which was already known to have some Republican bias. As importantly, or more so, a chunk of Mexicans in Texas voted Republican. So did some in Arizona. Republican support in the African-American community may also have grown, which is counter-intuitive, given the love affair between white supremacists and Mr Trump.

This trend of Hispanics leaning Republican should be a red flag. Perhaps these Hispanic-Republican voters were the legendary “shy” Republicans, unwilling to disclose preferences because they were outliers in their communities? But it should have been obvious to strategists that Puerto Ricans, Mexicans, Cubans, Venezuelans, et al, can’t be lumped into one bucket because they share a language.  

There was a split along gender lines. More African American men voted Republican than African American women; more Hispanic men voted Republican than women; more white men voted Republican than white women.

In 2016, pollsters realised post-election they had failed to account for education as a variable. There is a divide between white college graduates (who vote Democrat) and white non-graduates who vote Republican. That continued. But both parties have made inroads into the other’s turf.

Most of all, I think pollsters simply got voter turnout wrong. There was a massive turnout expansion. All forecasts involving “likely voters” were gross under-estimates. Many “unlikely voters” voted — who knows if this trend will continue?

Last but most encouraging, there were split votes.  Many voters who opted for Republican Senators and House Representatives, did not vote for Mr Trump. This disappoints those who hoped for a Democratic sweep. But it also indicates there are conservative Americans who are repulsed by criminality and racism.