Hawaii has long had a reputation as a difficult state in which to conduct an accurate poll. This year, our pre-primary and pre-election polls, conducted in partnership with Civil Beat and Hawaii News Now, proved to be very accurate.

In particular, our October poll was, top to bottom, one of the most accurate state-level polls in the country.

While we are always pleased to get it right, we have had our occasional misses, too. When we’ve missed in Hawaii, we have written about it and learned from it, adjusted our methodology accordingly, and invited conversation about it after the election.

You learn more from your mistakes than from your successes. This year, with significant errors in many other state polls throughout the country, we wanted to offer our experience from polling in 2020; the things that went right and the challenges we and our colleagues face going forward.

And, as our sister company Merriman River Group focuses on election administration, we want to start a conversation about the increasingly diverse ways people are able to cast a vote and what that means. While we don’t want to speculate too much about polls we did not conduct, we did notice response rates dropping, especially in battleground states.

Our conclusion is that part of it was an “over-fishing” phenomenon, where the sheer volume of campaigning and polling may have promoted a form of political fatigue where it became increasingly difficult to get people to participate in polls.

Very frequent voters are always more likely to take polls, making lower-proclivity voters (those who vote sometimes but not always) that much harder to reach. And data has suggested that President Donald Trump has been especially good at motivating those lower-proclivity voters to come out and support him.

While participation rates in polling dropped in many states, voter participation at the polls actually soared.

Despite his second defeat in the popular vote, President Trump actually exceeded his 2016 national vote total by over eight million votes, despite a perceived erosion among more frequent voters (indeed, President-elect Joe Biden received over ten million more votes than Hillary Clinton’s 2016 total).

It is likely that Trump’s delivery of these lower proclivity voters had a large impact on Senate races, where many Republicans over-performed public polls by large margins. And of course, expanded early voting and mail-in voting that was allowed in response to the pandemic further drove turnout. (Hawaii approved a full vote-by-mail program before the pandemic.)

In Hawaii, we thankfully didn’t experience a drop in response rates to our polls, and our new, upgraded method of reaching cellphone-only voters actually improved our response rate with this growing sector of the electorate. Using our sample of registered voters, we now send a text message with a link to our online survey to a random selection of voters.

To reach landlines, we use IVR calling, in which a recorded voice asks questions and survey participants respond by pushing buttons on their phones. And once the cellphone and text data are collected, we weight the results by age, gender, race, location and educational level.

This double refinement of contacting a representative sample of voters and further refining the results by demographic weighting has worked well for us in the past and will remain our methodologic approach in the future, even as we always seek to improve the way we reach voters.

Some polling aggregators believe that polls conducted by live-agent callers are superior, but we disagree, especially in places like Hawaii. We believe a good IVR poll greatly reduces bias for a number of reasons.

First, using one professional recorded voice ensures that everyone has the same experience. It dramatically reduces mispronunciations, it eliminates data entry errors, and we believe people are more truthful and honest talking to a friendly and neutral voice than to another person who may be judging their responses.

Some short-cut IVR polls use a computer-generated voice. We strongly believe our use of an actual voice makes for a more motivating experience. We’re proud that our polling in Hawaii has consistently been the most accurate polling in Hawaii for several years now and hope to build on this in the future.

Finally, while voting in a pandemic raised obvious challenges, our democracy largely rose to the occasion. New ways to vote were offered to voters, and voters responded with the highest total turnout ever and the highest turnout by percentage in 120 years, two decades before women were allowed to vote.

Obviously, this was not universal. Some states did as little as possible to make voting accessible (hello, Texas). Other states, Pennsylvania most notably, expanded mail-in voting but forced election officials to wait until Election Day before touching the crush of incoming mail.

Hawaii is a clear success story. By expanding mail-in voting, turnout soared, and by allowing election officials to process that mail as it came in, Hawaii results were available very quickly. Ironically the slight delay in Hawaii reporting was related to last minute in-person voters. Hopefully, more states will follow Hawaii’s lead in the future.

Access to voting should not be a political issue. While mail-in voting greatly favored Democrats this year, it does not necessarily follow that it will do so in the future. Indeed, in many states with a history of mail-in voting, results typically are fairly evenly split.

This year, one party urged voters to vote by mail and the other party urged voters to vote in person. Mail-in balloting and early voting are reliable partners to in-person voting and should be expanded nationally in the years ahead.