What We Learned from the 2020 Election Exit Polls

Thank you for joining us for our first webinar of 2021, What We Learned from the 2020 Election and Exit Polls.

Click here to download a copy of the presentation

In this webinar, EVP Joe Lenski and VP Randy Brown gave an overview of the challenges presented by exit polling during a pandemic, the processes that Edison Research developed to conduct the exit poll of record for the National Election Pool, and some of the findings about the U.S. electorate.

Click here for more on Edison Research and our role in the U.S. elections.


Podcast Brand Lift Studies

To gauge the efficacy of podcast advertising, Edison adopts three different strategies, depending on the most applicable circumstances. We have extensive experience with all three and all of our work is custom—a true partnership between Edison, the advertiser, and (where relevant) the agency. We work with all stakeholders to craft a questionnaire to achieve the brand’s measurement goals and align our work with existing brand measures.

Edison employs the following methodologies, depending on the size of the campaign and audience to be reached. All three methodologies carry roughly the same costs–we select the best tool for the job based on methodology.


  1. Traditional Pre/Post Campaign Research

For this style of research, which works best with existing, well-performing shows and networks, we will run audio solicitations to take a survey in the podcast(s) selected for the campaign. Ideally these are host-read, and do NOT mention the brand in any way. The solicitation will drive respondents to a link to take an online survey hosted on our enterprise survey system. We run this survey twice—once before a campaign begins, and a second time after the flight or some other mutually agreed-to timeline (we typically prefer at least 6 weeks of advertising).

This type of research is a convenience sample, but it is at least an apples-to-apples look at the perceptions of the correct, exposed audience. Significance testing is used to determine if the pre and post populations gave different answers at a level that is determined to not be a chance result. In addition, to account for sample variance between the pre and post studies, we weight both studies to the average of the two. So, if (for example) the Pre were 60% female while the Post were 40% female, we would weight BOTH studies to 50% female, to provide a very simplistic example.

Depending on the popularity of show(s) selected as part of the campaign, we may recommend an incentive to ensure an adequate response rate. This can either be a guaranteed incentive (e.g., an Amazon gift card) for a maximum number of respondents or a sweepstakes where Edison will collect name and email address of each respondent and provide the information of the randomly selected winner. The costs of any incentives are the responsibility of the client.

We do not guarantee a sample size for these studies since they are not “sampled,” per se. We work with the podcast network and/or media buyer to ensure a sufficient sample for both studies to provide significance testing on the results. We provide a complete presentation with all relevant charts and graphs, all hosting and data processing, and questionnaire(s) creation. Edison Research has conducted hundreds of similar studies and the results have been presented to leading agencies and media buyers around the world.


  1. Panel Research

For large campaigns, stretching across multiple podcasts or podcast networks, we are able to employ our large national panel of podcast listeners (which we also tap into for our syndicated Podcast Consumer Research quarterly report.) If an advertising buy is significant enough for it to be feasible, a panel study approach allows us to project the findings across the entire audience for the podcast or network in question, and is the most representative form of podcast research.

Also, in some cases, a “pristine” pre and post process is not possible (e.g. when the brand advertising is already running prior to engaging with Edison, or when the brand is running across multiple networks, which muddies attribution) which makes a panel approach cleaner than a pre/post survey methodology.

For panel-driven studies, we employ a national online survey of Americans 18+ (or whatever demographic is appropriate) who regularly listen to podcasts, and use the same brand questionnaire that we might use for a pre/post study methodology. In a panel study, we will guarantee a sample of podcast listeners, podcast listeners who have likely heard the advertising in question, and non-podcast listeners. We will then be able to gauge the relative impact of the specific podcast ads by comparing brand measures with non-podcast listeners, podcast listeners who do NOT listen to the specific shows/networks that featured the advertising, and an audience of the targeted podcasts who would certainly have been exposed to the ads. Where appropriate, we can also fold in an audio call-out methodology in relevant podcasts, as described above.

For this style of study, all other deliverables are as mentioned above, including presentations to the relevant agency/brand team.


  1. Forced Listening and Controlled Exposure

Finally, in the case of a newly-created branded podcast, advertising on a new show, or other projects with potentially small samples, we engage a “forced listening” approach. For this approach, we recruit a sample of podcast listeners who would at least have an affinity for the programming being tested (e.g. parents of small children for a parenting podcast). The sample is instructed to come to a facility and listen to a half-hour of programming and instructed that they will later be asked what they thought of the show.

In reality, they are also given questions about the advertising and sponsorship messages in the show. This approach, unlike the first two, is strictly qualitative, which means that traditional measures of margin of error or statistical significance really don’t apply. However, this approach does provide directional research where a traditional survey would not be feasible either due to the size of the show’s audience or for a completely new program.

All recruiting, execution, incentives, and facility fees are included in this type of project, in addition to reporting of the results. We have successfully executed and presented these for dozens of agencies and media partners.


Finally, Edison is a custom market research provider with 25 years of service tackling the most challenging research projects, and we have developed many of the current approaches to podcast research in use today. Should none of these approaches completely address a measurement challenge, we’ll develop a new one as needed.

Questions? Email Edison Research SVP Tom Webster at twebster@edisonresearch.com.


January 5 Election Day Coverage

Edison Research continues coverage of the 2020 election with exit polls of early voters and election day voters for the January 5, 2021 Senate runoff election in the state of Georgia. Edison Research conducted the exit poll of record for the November 3 general election for The National Election Pool (NEP), comprised of ABC News, CBS News, CNN and NBC News.

The NEP’s exit poll is the only survey that will be released on election night that represents the views and opinions of actual voters interviewed after they cast their ballots at early voting locations, and after they cast their ballots on election day in Georgia.

Click here for more on Edison Research and U.S. elections

Click on the links below for coverage of the Senate runoff election in Georgia, including exit poll data:





Edison Research Published Studies 2020

It is customary at year’s end to reflect a bit on the work that was accomplished in the prior twelve months. This year we are grateful to be able to look back at the studies that we conducted, and recognize that through the dedication of our Edison Research team and our research partners Ad Results Media, Adonde Media, Lantigua Williams and Co., Libsyn, Marketplace, NPR, Pandora, PodcastOne, and Triton Digital, we were able to produce many valuable research insights in spite of the challenges of 2020.

This year’s findings shed light on audio, podcasts, economic anxiety in America, and of course the ubiquitous impact of COVID-19. Our biggest project of the year was by far the 2020 election, as we conducted the exit poll of record for the National Election Pool (NEP) comprised of ABC, CBS, CNN and NBC, which meant we interviewed over 100,000 voters for the general election. As of this writing, exit poll work continues with the runoff Senate elections in the state of Georgia.

So thank you all for your support during the past year, and we hope you enjoy perusing these research projects from 2020.

Latino Podcast Listener Report in English:

Latino Podcast Listener Report in Spanish:

Edison Research’s 10 for ’20

Welcome to Edison Research’s 10 for ’20. It’s like a Top 10 List for 2020, but better — because we know 2020 doesn’t warrant a standard Top 10 List.

Despite the circumstances of quarantine, we were fortunate enough to continue to do market research projects and talk to thousands of people about their audio habits, their views on the economy, and their voting choices as well. We had to revise some of our research methods, particularly when it came to the work of exit polls, and we had to embrace working remotely, but we have 2021 in the sights, so we are thrilled to bring you these findings.

The three defining news stories of 2020 were the global pandemic from Covid-19, the U.S. presidential election, and a time of racial awakening and reckoning across the country. Those three wound their way through all of our research this year, which is one of the reasons our 10 for ‘20 list is impossible to present in rank order.

Here are our biggest findings from 2020, presented by various members of our Edison Research team, in an order that tells the story of this momentous year. Below each finding you will find links to the cited studies that you can explore for more information.

U.S. audio consumption changed as a result of COVID-19
Source: Share of Ear Listening Location, Share of Ear Audio Day Starts Later
Listening location changed dramatically in second quarter with 70% of total listening taking place at home, down from 49% pre-quarantine.Third quarter saw some listening shift away from home, with 59% of total listening taking place at home. Americans’ audio day started later as quarantine restrictions began. Before the pandemic, half of all respondents in Share of Ear had recorded some type of audio consumption by 7:15. During the quarantine restrictions of second quarter, that time shifted forward by 75 minutes so that half of respondents had recorded some type of audio consumption by 8:30. In third quarter, that time had only bounced back to 8:15am.

100 Million Americans listen to Podcasts Monthly
Source: The Infinite Dial 2020 with Triton Digital
While we still haven’t hit ‘peak’, this is the year where podcasting hit over 100 million monthly listeners. After years of single-digit growth, that’s more than a 40% increase in a two year period. Compared to the explosive growth of monthly podcast listening in the last five years from 21% of the US population in 2016 to 37% in 2020, monthly Twitter usage has essentially stayed the same 17% of the US population in 2016 and 17% in 2020. That means today, well over twice as many people listen to a podcast every month than use Twitter monthly.

25% of U.S. Latinos Listen to Podcasts Monthly; many are new to the medium
Source:The Latino Podcast Listener Report with Adonde Media, Lantigua Williams and Company, Libsyn, NPR, and Pandora
Listenership among Latinos is poised to grow as their awareness of the medium increases, and this growth may have already started. Over half of Latino podcast listeners started listening just within the last year. There are big differences between Language Dominant groups, with Spanish-language dominant listeners being more likely to have come to the medium within the last six months. Latinos have significantly been contributing to the expansion of podcasting already. The industry will have a better chance at maintaining this momentum if they take time to understand this group.

Spoken Word’s share of audio listening increased 30% over the past six years, 8% in the last year; growth driven by women, African-Americans, Latinos and 13-34-year-olds
Source: The Spoken Word Audio Report from NPR and Edison Research
2020 was a good year for Spoken Word audio, including sports talk and play-by-play, talk and personality audio programming, and audiobooks. The share of time spent listening to spoken word audio in the U.S. increased by 30% in the past six years, and 8% in the just the last year. The highest levels of growth are coming from women, African-Americans, Latinos and 13-34-year-olds. In addition to the convenience and multitasking benefits of spoken word audio, listeners tuned in for personal growth, better content, and the positive effects on mental health.

55% of Gen Z listeners are reached by AM/FM radio every day
Source: Radio’s Roadmap to Gen Z Listenership with Front Row Insights and Strategy
Our study shows that 55% of all Gen Zs listen to AM/FM radio each day, so they recorded at least some radio listening into their Share of Ear diary. This just surpasses our estimate for streaming. In an environment where radio people are constantly hearing that no young people listen to the radio anymore, this is powerful proof that it is not the case. More than half said they listen every day. Of the time spent listening to all audio among 13-24 year-olds, 22% of it goes to AM/FM radio.

60 million Americans now own a smart speaker, and usage increased during quarantine
Source: The Smart Audio Report from NPR and Edison Research
Twenty-four percent of people in the U.S own at least one smart speaker – around 60 million people. Over one-third of U.S. adult smart speaker owners say they are using their device more to listen to music and entertainment since the outbreak – and younger Americans, those 18-34-year-olds, are even more likely to turn to the device for a diversion – a little over half – or 52% — of young Americans said they are using smart speaker for music and entertainment since the pandemic started.

Over half of U.S. Moms agree they are spending a lot time helping their children with distance learning; Moms much more likely to be the primary person responsible for helping
Source: Source: Moms and Media 2020, Marketplace-Edison Research Poll

Fifty-four percent of moms agree they are spending a lot of time helping their kids with distance learning during COVID-19). Moms with school age children who are remote learning are much more likely to be responsible for helping their children with school. Sixty-three percent of moms said they are primarily responsible for helping their children with online learning, compared to 29% of dads who said the same. Nearly half of those moms (48%) agreed that online school was overwhelming. (Marketplace-Edison Research Poll) 30% of moms reported cutting back on work hours in order to help their children with schooling.

More than half of Americans don’t see America as the land of equal opportunity
Source: Marketplace-Edison Research Poll
Forty-eight percent of Americans said that generally speaking, Black workers are typically paid less than white workers doing the same job.” Breaking this down by race – more than three-quarters of Black Americans say this is the case. What is also really striking is that 41% of white Americans and about 48% of Hispanic or Latino Americans say this is the case.
In the same vein, six in 10 Americans think that race has at least some impact on an individual’s long-term financial situation.

COVID-19 changed how we vote
Source: Edison Research Election Polling
During this 2020 General election more people voted prior to election day either by mail or early in person than ever before (at least 100 million). The exit polls and raw vote count showed clear differences in candidate preference according to the method by which an individual chose to vote. Among those who voted by mail (about 43% of voters), Biden won the group by 31 points – 65-34. Those who voted early in person (about 24% voters), split their vote evenly between Biden and Trump. Finally,voters who voted in person on election day, (about 33% of voters), voted for Trump 59-39.

Voter turnout was up by more than 21.5 million people, many of whom voted for the first time
Source: Edison Research Election Polling
The biggest story of this election is the enormous increase in turnout. In 2016 the US had about 137 million voters which was a turnout record itself. For the 2020 General election, turnout will end up at about 158.5 million voters, an increase of about 21.5 million voters. A significant portion of this turnout increase were the 14% of voters who said they voted for the first time this year (whether it be during the primary of general). Of those 2020 first time voters, the exit poll shows they voted for Biden 64 to 32.A final aspect of this high turnout electorate to acknowledge is that it featured a noticeably lower percentage of third party presidential votes than in 2016. In 2016 about 6% of voters voted third party, while this year less than 2% went to a non-Democrat or Republican choice for president.