How public polling has changed
Plus, Americans are buying pot more than chocolate, primary swing voters on board for DeSantis's "war on woke", podcast consumption, and how AI might put polling out of business
No. 263 | April 21st, 2023
📊 Public Opinion
How Public Polling Has Changed in the 21st Century (Pew Research Center)
“The 2016 and 2020 presidential elections left many Americans wondering whether polling was broken and what, if anything, pollsters might do about it. A new Pew Research Center study finds that most national pollsters have changed their approach since 2016, and in some cases dramatically. Most (61%) of the pollsters who conducted and publicly released national surveys in both 2016 and 2022 used methods in 2022 that differed from what they used in 2016. The study also finds the use of multiple methods increasing. Last year 17% of national pollsters used at least three different methods to sample or interview people (sometimes in the same survey), up from 2% in 2016.”
The Intersection is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
AI in Hiring and Evaluating Workers: What Americans Think (Pew Research Center)
“The rapid rise of ChatGPT and other artificial intelligence (AI) systems has prompted widespread debates about the effectiveness of these computer programs and how people would react to them. At times, Americans are watching the general spread of AI with a range of concerns, especially when the use of AI systems raises the prospect of discrimination and bias.
One major arena where AI systems have been widely implemented is workplace operations. Some officials estimate that many employers use AI in some form of their hiring and workplace decision-making.
A new Pew Research Center survey finds crosscurrents in the public’s opinions as they look at the possible uses of AI in workplaces. Americans are wary and sometimes worried.”
“Trump’s near-win in 2020 was significant because it showed the resilience of a Republican coalition based on working-class voters. (I’ve even got a book coming out that focuses heavily on this topic.) Trump hasn’t needed to be a popular politician in the classic sense because of the unique advantages such a coalition brings to the table in the Electoral College, with a 3.8 point pro-Republican bias in 2020.
A big part of Trump’s near-win in 2020 was his surprising strength among Hispanics—the same group he ranted about ‘bringing drugs, bringing crime’ in his 2016 announcement speech. He surged in places like Miami-Dade County, giving him the advantage in Florida, and in the Rio Grande Valley in Texas.
But Trump’s showing, impressive though it was, was not the high-water mark for Republicans and the Hispanic vote in the modern era. George W. Bush earned an eye-popping 44 percent of the Hispanic vote according to the 2004 exit poll, topping Donald Trump’s 36 percent in 2020 in AP VoteCast survey, the more reliable of the two modern exit polls.”
Which 2024 Republican Presidential Candidate Has The Most Endorsements? (FiveThirtyEight)
Amy Walter: Over Before It Began? (Cook Political Report)
“Polling suggests that the pathway to victory for a candidate not named Trump is to win over the significant group of voters who like Trump but are also open to an alternative. A survey done earlier this year by GOP pollster Whit Ayers for the Bulwark found a GOP electorate divided into three categories: 30% or so are ‘Always Trump’; 10-15% are ‘Never Trump’; and the rest fall into the ‘Sometimes Trump’ bucket.
More recently, the GOP firm Echelon Insights divided the GOP electorate into four categories: 25% say they’d vote for Trump but not DeSantis; 22% would support DeSantis but not Trump; 15% said they’d support neither DeSantis nor Trump; and 39% said they could support Trump and DeSantis.
And, when Echelon divided these groups by policy, DeSantis’ attacks on ‘woke’ businesses and schools were more popular with the 39% of ‘both’ voters than with the overall GOP electorate. For example, when asked if “businesses should be held accountable if they are too focused on being ‘woke’ or if, instead, ‘businesses should be able to operate however they like as long as they don’t break the law,’ a bare majority (50%) of GOP voters choose the ‘businesses should do as they like’ approach while 41% pick the ‘too woke’ option.”
“National polling generally suggests that voters trust Republicans more than Democrats when it comes to handling fiscal policy. In the lead-up to the 2022 midterms, pre-election polling from ABC News showed the GOP enjoying a 12-point advantage on the economy, compared to a 12-point deficit on the issue of abortion. More recently, CNN released a poll showing an identical edge for the GOP. These surveys beg an important question: do Republican candidates for treasurer and controller, two positions which involve fiscal policy on the state level, get an electoral bonus compared to their partisan counterparts?
As crossover voting continues to decrease, the effects of down-ballot lag that have historically enabled strong candidates to overperform have become less pronounced. Split Ticket has found that this polarization has permeated into state-level elections, helping Republican treasurers and controllers more consistently and clearly overperform their counterparts for other state row offices.”
J. Miles Coleman: Leaning Into State Trends: The Northeast and Greater South (Sabato’s Crystal Ball)
“As most of our readers already probably appreciate, the 2020 election was an unprecedented contest. More votes were cast in 2020 than in any previous election. The then-incumbent, Donald Trump, became the only sitting president to stand for reelection after an impeachment trial. Then there was the age of the candidates: if successful, Trump, at 74, would have been the oldest president to secure reelection — but he was replaced by now-President Joe Biden, who was even older.
Getting down to a more granular level, 2020 was unique, at least compared to presidential elections since 2000, in that no single state tracked especially closely with the national popular vote. In the 2000 election alone, for instance, 4 states — Oregon, Iowa, Wisconsin, and New Mexico — voted within half a percentage point of the national popular vote (Florida, infamously close that year, was actually just outside that range).”
🗺️ Data Visualization
Harry Stevens: America needs clean electricity. These states show how to do it. (The Washington Post)
“If humans escape climate scientists’ gloomiest projections, if we buy ourselves time to adapt to higher seas and fiercer heat waves, we will likely use more electricity than we do now, and we will make it without emitting greenhouse gases.
Today, the United States is running a natural experiment in electricity generation, with a patchwork of policies and power grids. To eliminate electricity’s greenhouse gas emissions, it makes sense to ask: What can we learn from the states that make cleanest power?
The chart below shows how the United States has made electricity for the past twenty years, represented as the percentage of power generated from each fuel source. To show how their relative usage has shifted, the fuels are stacked each year from top to bottom in order of percentage.”
📰 Media Habits
Podcasts as a Source of News and Information (Pew Research Center)
“Following a steady increase in podcast listening over the past decade, podcasts have become a big part of the normal routine – and news diet – of many Americans, especially younger adults.
Roughly half of Americans say they have listened to a podcast in the past year
Roughly half of U.S. adults say they have listened to a podcast in the past year, according to a new Pew Research Center survey, including one-in-five who report listening to podcasts at least a few times a week. Among adults under 30, about a third listen to podcasts with such frequency.
Today’s podcast landscape is a sprawling one, featuring topics from sports and religion to politics and entertainment, and it attracts large numbers of listeners who turn to podcasts for a variety of reasons.”
Jacob Liedke: Americans name a long, diverse list of podcasts they listen to most (Pew Research Center)
“Podcasts have become a big part of many Americans’ normal routine – and their news diet. But a new Pew Research Center survey reveals that podcast listening is highly fragmented, and no one show dominates.
The survey asked adult podcast listeners (those who said they listened to a podcast in the past 12 months) in the United States to name the podcast they listen to most. Overall, 61% of listeners volunteered the name of a podcast, but there was relatively little common ground in these answers.”
“The distribution of people across our planet is changing pretty dramatically, with populations booming in sub-Saharan Africa and shrinking in parts of Europe and East Asia, including China.
Driving the news: According to a new UN report, India will surpass China as the world’s most populous nation by the middle of this year, if it hasn't already.”
Language Models Trained on Media Diets Can Predict Public Opinion (Cornell University)
“Public opinion reflects and shapes societal behavior, but the traditional survey-based tools to measure it are limited. We introduce a novel approach to probe media diet models -- language models adapted to online news, TV broadcast, or radio show content -- that can emulate the opinions of subpopulations that have consumed a set of media. To validate this method, we use as ground truth the opinions expressed in U.S. nationally representative surveys on COVID-19 and consumer confidence. Our studies indicate that this approach is predictive of human judgements found in survey response distributions and robust to phrasing and channels of media exposure, more accurate at modeling people who follow media more closely, and aligned with literature on which types of opinions are affected by media consumption. Probing language models provides a powerful new method for investigating media effects, has practical applications in supplementing polls and forecasting public opinion, and suggests a need for further study of the surprising fidelity with which neural language models can predict human responses.”