It’s time to end presidential debates — forever

Share this

Headshots of President Trump and former Vice President Joe Biden during the first presidential debate.

Yeah, I’m not happy about these debates either. | Saul Loeb, Jim Watson/AFP via Getty Images

At least in the general election, they’re a vector for misinformation and don’t seem to meaningfully inform voters.

Last Friday, the Commission on Presidential Debates officially canceled the second presidential debate of the 2020 general election, originally scheduled for Thursday, October 15. The cancellation was caused by President Trump’s refusal to debate remotely (despite his positive test for Covid-19). According to Newton Minow, a member of the debate commission and one of the negotiators behind the first televised debates in 1960, the cancellation’s real victim was American democracy.

“In seven decades of televised presidential debates, this is the first debate to be canceled,” he told the New York Times. “The loser is the American voter.”

Minow is being a bit cute here. In 1964, 1968, and 1972, debates were not canceled — because they were not held at all. And while there is little doubt that the 1960 debates he helped negotiate for John F. Kennedy were a watershed moment for television as a medium and for popular democracy more generally, they happened 60 years ago. Telecommunications has changed dramatically, and so has politics.

Plenty of observers called for canceling the remaining debates after the embarrassing first confrontation between Trump and Joe Biden. (They almost got their wish — after being canceled, the second debate ended up being replaced by dueling Thursday town halls.) But I would argue that we should give up on presidential debates, period. Debates can still provide useful information on Senate or governor’s candidates, or in primary elections. But there is no reason to think, in 2020, that televised debates between major-party presidential nominees provide any real value to voters.

“I don’t have any more information than when I started watching,” undecided voter Ellen Christensen told the New York Times after the first presidential debate. She’s right: She didn’t get new information. Instead, she got a lot of falsehoods, in large part because Trump has learned that it is easy to overwhelm the system by repeating falsehoods too quickly to debunk them, that the costs of spreading misinformation are minimal and the benefits immense.

It is obvious why Trump likes a system where he can claim “there aren’t 100 million people with preexisting conditions” when there are, or that he has “given big incentives” for electric cars when he hasn’t, that Biden wants to “take out the cows” when Biden obviously does not want to outlaw cows, and that “young children” are not vulnerable to Covid-19 when they are — all without getting called or corrected on it by the moderator (Biden sometimes tried, to little avail). It’s less clear why any of the rest of us should tolerate such a system.

Debates don’t appear to change many minds

The argument against debates is simple. Debates are only a good use of time if they provide useful, new information to voters that they either would not or could not obtain through other means.

But there is not much reason to think they provide that kind of information, or at least a substantial amount of it.

The simplest way to see if debates are providing useful information is to test if they change voters’ preferences. This is a tricky matter: Debates take place in a busy fall campaign season, which means political scientists have a hard time untangling the effects of debates from the effects of campaign ads, news coverage, and so forth. Moreover, the vast majority of viewers are already decided — in a recent poll, only 6 percent of likely voters planning to watch the first debate even claimed to be undecided — and measuring the effects on the tiny share who are undecided is difficult.

But with that caveat, the bulk of the political science literature has not found much evidence that debates substantially influence the outcomes of presidential elections.

Robert Erikson and Christopher Wlezien’s 2012 book The Timeline of Presidential Elections compiled evidence on the effect of general election presidential debates from 1960 to 2008, and they indeed found that in general, not many voters are up for grabs by the fall when debates take place; in 2004 and 2008, the most recent elections in their sample, opinions of voters had hardened even earlier, by the summer. That puts a relatively low upper bound on how influential debates could be: Even if they explained all the variation in candidate support across the fall campaign, they wouldn’t be explaining much, because there wasn’t much variation to explain. (Note that this analysis does not cover primary debates, which take place earlier in an election cycle, are much more numerous, and probably have different effects.)

In any case, Erikson and Wlezien find that except for the 1976 election, “vote intentions the week after the debates closely matched those the week before the debates.” In other words, it’s hard to detect much change in overall voter opinion attributable to the debates. (Even the apparent exception of 1976 is a bit ambiguous: Jimmy Carter had a large lead over Gerald Ford going into the fall, which declined steadily until Election Day, with the debates not appearing to speed up or interrupt the decline that began before them.)

“It is clear that debates do not have major impact to the same degree as party conventions,” Erikson and Wlezien conclude carefully. So allow me to be blunter: Their data appears consistent with studies that came before that showed small effects from presidential debates on voter choice. There’s not much reason to conclude, from the data they present, that debates have a major impact at all.

The 2012 election appeared to be an exception to this finding at first, as Mitt Romney was widely viewed as having won the first debate and then saw a modest bump in polling afterward. But as political scientists John Sides and Lynn Vavreck note in their book on the 2013 election, The Gamble, the overall effect of the debates that year was quite minor.

One reason is that Barack Obama did pretty well in debates two and three, undoing some of the damage from the first debate. But Sides and Vavreck also detail evidence that Romney’s bump after the first debate was illusory, the result of response bias: Because Romney was thought to have done well, Romney supporters were more likely, and Obama supporters less likely, to answer pollsters afterward. Indeed, a YouGov poll that interviewed and then reinterviewed the same 25,000 people before and after the first debate found almost no change in voter preferences. Another large-scale panel survey found the same thing.

“The sum total of the debate season, then, was to create a tighter race but not put Romney in the driver’s seat,” Sides and Vavreck conclude. “This was consistent with history and the academic literature: debates have moved the polls but rarely determined the winner of the election.” Even in our recent, super-close elections, the debate effects have been too muted to turn the outcome. When Erikson and Wlezien updated their book for 2012, they agreed, writing, “The 2012 debates seem to have had some effect on preferences, but it was small and short-term.”

2016 is a much trickier year to unpack: The period of the debates also saw a controversy over Alicia Machado (a former Miss Universe whom Trump bullied for her weight and ethnicity), the release of the notorious Access Hollywood tape wherein Trump bragged about sexual assault, accusations that Trump had in fact forcibly touched or kissed multiple women, the WikiLeaks release of hacked private emails of Clinton campaign chair John Podesta, and the abandonment of Trump by House Speaker Paul Ryan and former nominee John McCain. When all that is happening at the same time, and then the debates are followed up by the FBI director resurrecting an investigation into one of the candidates, it’s next to impossible to isolate the effect of individual factors like debate performance.

Sides, Vavreck, and co-author Michael Tesler find in their book on 2016, Identity Crisis, that the first debate and the Machado controversy led to a “sharp drop in the percentage of voters who rated [Trump] favorably.” They find the second debate had a muted impact, but that after the third debate, “Republican voters who had soured on Trump after the first debate and release of the Access Hollywood tape returned to him.”

They do not credit this return to his debate performance, which was not viewed as exceptionally good at the time, but to the tendency of partisans to return to their candidate, the failure of a critical mass of Republican establishment figures to reject Trump or endorse Clinton, and several other factors. It doesn’t add up to a compelling case that debates were decisive, either.

“Raw polls at 50 days out, on Sept 19, one full week before the first debate, had Clinton at 51.7 percent; on October 26, one week after the last debate, the share was 52.4 percent,” Wlezien notes in an email. “There is even less movement — a 0.4 percent drop in Clinton share — of house-adjusted poll averages.”

“‘Winning’ debates is overrated,” Erikson concluded in an email to me. “Ask John Kerry or Hillary Clinton.”

Will 2020 break this streak? I suppose it’s possible, but the evidence is lacking so far. The Economist’s poll average, which corrects for polls that don’t weight by party and thus can avoid the kind of non-response bias described above, found that Biden was averaging 53.9 percent of the two-party vote on September 29, the day of the first debate, and … 54.2 percent on October 15. That’s a net gain of 0.3 points. And given everything else that has happened, especially Trump’s Covid-19 diagnosis, it’s hard to even credit that minimal shift to the debate.

There’s no reason to think debates generate new, previously unavailable information

The finding that debates don’t really matter is not, on its own, proof that debates are useless. Maybe individual voters’ opinions are moved by them in a durable way that influences their eventual vote, and the movements simply cancel each other out. You could mount an argument that the debates are thus informing voters even if this does not have an ultimate effect on vote shares.

This does not look super-likely when you dig into particular cases. Indeed, debates instead can be a tool for entrenching polarization and bringing voters in line with their preferred candidates’ views — seems to be the case.

Political scientist Gabriel Lenz has analyzed the effects of Social Security-centric debates in the 2000 general election and found that they “did not cause voters to switch their support to whichever candidate agreed with their own position on that issue.” Instead, Lenz found that the debates led voters — specifically voters who learned the candidates’ issue positions for the first time — to adopt their preferred candidates’ positions. Gore supporters came to love Gore’s idea of keeping excess Social Security/Medicare revenue in a “lockbox,” and Bush supporters came to love Bush’s idea of investing some Social Security taxes in the stock market.

That’s not necessarily a bad thing — some of these voters presumably changed their minds because they trusted Bush or Gore, and changing your opinion to account for learning the views of someone you trust is very rational behavior.

But that phenomenon is clearly not the rationale offered for presidential debates. They’re supposed to help voters make up their minds — advocates and researchers Diana Carlin and Mitchell McKinney argue “the debates could be crucial shaping new voters’ choices and deciding the outcome” — not help parties convince their bases to adopt their candidates’ positions on important issues.

And insofar as the latter is a valuable service, it does not seem like the job of news organizations, but of the parties themselves. ABC News shouldn’t have to invest its money and resources to help the Republican Party rally supporters around Trump’s ideas, or to help the Democratic Party rally supporters around Biden’s.

This still leaves the theoretical possibility that there are some voters who change their minds due to debates, but whose votes cancel each other out so there’s little net effect. This, I would argue, is only valuable insofar as these voters are gaining new, accurate information about the candidates.

There’s just no reason to think that happens. General election candidates have many, many avenues through which to distribute their platforms to the public, including many more than existed at the advent of television debates in 1960. The campaigns post their platforms online, they explain them in depth to a wide number of news outlets (including online outlets without the word limits that newspapers and magazines labored under), and if voters miss an article, they can simply Google it. Primary debates, where issue disagreements between candidates are minimal and most voters are undecided, probably give voters useful information. The same likely can’t be said of general election debates.

Debates might offer some value in creating a single, focusing event that lets low-information voters get back up to speed. But for them to serve that purpose, the information would need to be accurate, and the 2020 debates to date are excellent illustrations of candidates’ abilities to lie without being interrupted by moderators.

Vice President Mike Pence, for instance, was able to claim without correction or interruption that Biden and Kamala Harris want to “abolish fossil fuels and ban fracking” (they do not) and that the White House event that produced a cluster of Covid-19 cases was outdoors (the reception was indoors). The co-chair of the Commission on Presidential Debates has gone so far as to say that fact-checking isn’t the moderator’s job. Someone tuning in to the vice presidential debate and hearing that Biden wants to ban fracking is not actually gaining useful, true information about the candidates.

If we want another kind of central, focusing event, something like single-candidate town halls with the individual candidates and — this is key — a moderator who is equipped to fact-check, or perhaps a simulation where the candidates both attempt to complete a modeled national security crisis, would be far likelier to produce useful information for voters.

We do not have as rigorous social science for the proposition that debates do not lead viewers to have more accurate information about the candidates as we do for the proposition that debates don’t swing many votes. But the content of the debates themselves, and the existence of numerous accurate sources of information on the candidates, strongly suggest to me that debates, even ones between relatively responsible candidates like Romney and Obama, do not improve public understanding of the issues discussed.

So the question then is: If debates do not surface information voters cannot get elsewhere, and they cost a ton of money and effort to put on, and they can be a means to propagate misinformation, and they do not influence the ultimate vote totals … why do we insist on continuing to hold them?


Will you help keep Vox free for all?

The United States is in the middle of one of the most consequential presidential elections of our lifetimes. It’s essential that all Americans are able to access clear, concise information on what the outcome of the election could mean for their lives, and the lives of their families and communities. That is our mission at Vox. But our distinctive brand of explanatory journalism takes resources. Even when the economy and the news advertising market recovers, your support will be a critical part of sustaining our resource-intensive work. If you have already contributed, thank you. If you haven’t, please consider helping everyone understand this presidential election: Contribute today from as little as $3.

Share this

Leave a Reply

It’s time to end presidential debates — forever

Share this

Headshots of President Trump and former Vice President Joe Biden during the first presidential debate.

Yeah, I’m not happy about these debates either. | Saul Loeb, Jim Watson/AFP via Getty Images

At least in the general election, they’re a vector for misinformation and don’t seem to meaningfully inform voters.

Last Friday, the Commission on Presidential Debates officially canceled the second presidential debate of the 2020 general election, originally scheduled for Thursday, October 15. The cancellation was caused by President Trump’s refusal to debate remotely (despite his positive test for Covid-19). According to Newton Minow, a member of the debate commission and one of the negotiators behind the first televised debates in 1960, the cancellation’s real victim was American democracy.

“In seven decades of televised presidential debates, this is the first debate to be canceled,” he told the New York Times. “The loser is the American voter.”

Minow is being a bit cute here. In 1964, 1968, and 1972, debates were not canceled — because they were not held at all. And while there is little doubt that the 1960 debates he helped negotiate for John F. Kennedy were a watershed moment for television as a medium and for popular democracy more generally, they happened 60 years ago. Telecommunications has changed dramatically, and so has politics.

Plenty of observers called for canceling the remaining debates after the embarrassing first confrontation between Trump and Joe Biden. (They almost got their wish — after being canceled, the second debate ended up being replaced by dueling Thursday town halls.) But I would argue that we should give up on presidential debates, period. Debates can still provide useful information on Senate or governor’s candidates, or in primary elections. But there is no reason to think, in 2020, that televised debates between major-party presidential nominees provide any real value to voters.

“I don’t have any more information than when I started watching,” undecided voter Ellen Christensen told the New York Times after the first presidential debate. She’s right: She didn’t get new information. Instead, she got a lot of falsehoods, in large part because Trump has learned that it is easy to overwhelm the system by repeating falsehoods too quickly to debunk them, that the costs of spreading misinformation are minimal and the benefits immense.

It is obvious why Trump likes a system where he can claim “there aren’t 100 million people with preexisting conditions” when there are, or that he has “given big incentives” for electric cars when he hasn’t, that Biden wants to “take out the cows” when Biden obviously does not want to outlaw cows, and that “young children” are not vulnerable to Covid-19 when they are — all without getting called or corrected on it by the moderator (Biden sometimes tried, to little avail). It’s less clear why any of the rest of us should tolerate such a system.

Debates don’t appear to change many minds

The argument against debates is simple. Debates are only a good use of time if they provide useful, new information to voters that they either would not or could not obtain through other means.

But there is not much reason to think they provide that kind of information, or at least a substantial amount of it.

The simplest way to see if debates are providing useful information is to test if they change voters’ preferences. This is a tricky matter: Debates take place in a busy fall campaign season, which means political scientists have a hard time untangling the effects of debates from the effects of campaign ads, news coverage, and so forth. Moreover, the vast majority of viewers are already decided — in a recent poll, only 6 percent of likely voters planning to watch the first debate even claimed to be undecided — and measuring the effects on the tiny share who are undecided is difficult.

But with that caveat, the bulk of the political science literature has not found much evidence that debates substantially influence the outcomes of presidential elections.

Robert Erikson and Christopher Wlezien’s 2012 book The Timeline of Presidential Elections compiled evidence on the effect of general election presidential debates from 1960 to 2008, and they indeed found that in general, not many voters are up for grabs by the fall when debates take place; in 2004 and 2008, the most recent elections in their sample, opinions of voters had hardened even earlier, by the summer. That puts a relatively low upper bound on how influential debates could be: Even if they explained all the variation in candidate support across the fall campaign, they wouldn’t be explaining much, because there wasn’t much variation to explain. (Note that this analysis does not cover primary debates, which take place earlier in an election cycle, are much more numerous, and probably have different effects.)

In any case, Erikson and Wlezien find that except for the 1976 election, “vote intentions the week after the debates closely matched those the week before the debates.” In other words, it’s hard to detect much change in overall voter opinion attributable to the debates. (Even the apparent exception of 1976 is a bit ambiguous: Jimmy Carter had a large lead over Gerald Ford going into the fall, which declined steadily until Election Day, with the debates not appearing to speed up or interrupt the decline that began before them.)

“It is clear that debates do not have major impact to the same degree as party conventions,” Erikson and Wlezien conclude carefully. So allow me to be blunter: Their data appears consistent with studies that came before that showed small effects from presidential debates on voter choice. There’s not much reason to conclude, from the data they present, that debates have a major impact at all.

The 2012 election appeared to be an exception to this finding at first, as Mitt Romney was widely viewed as having won the first debate and then saw a modest bump in polling afterward. But as political scientists John Sides and Lynn Vavreck note in their book on the 2013 election, The Gamble, the overall effect of the debates that year was quite minor.

One reason is that Barack Obama did pretty well in debates two and three, undoing some of the damage from the first debate. But Sides and Vavreck also detail evidence that Romney’s bump after the first debate was illusory, the result of response bias: Because Romney was thought to have done well, Romney supporters were more likely, and Obama supporters less likely, to answer pollsters afterward. Indeed, a YouGov poll that interviewed and then reinterviewed the same 25,000 people before and after the first debate found almost no change in voter preferences. Another large-scale panel survey found the same thing.

“The sum total of the debate season, then, was to create a tighter race but not put Romney in the driver’s seat,” Sides and Vavreck conclude. “This was consistent with history and the academic literature: debates have moved the polls but rarely determined the winner of the election.” Even in our recent, super-close elections, the debate effects have been too muted to turn the outcome. When Erikson and Wlezien updated their book for 2012, they agreed, writing, “The 2012 debates seem to have had some effect on preferences, but it was small and short-term.”

2016 is a much trickier year to unpack: The period of the debates also saw a controversy over Alicia Machado (a former Miss Universe whom Trump bullied for her weight and ethnicity), the release of the notorious Access Hollywood tape wherein Trump bragged about sexual assault, accusations that Trump had in fact forcibly touched or kissed multiple women, the WikiLeaks release of hacked private emails of Clinton campaign chair John Podesta, and the abandonment of Trump by House Speaker Paul Ryan and former nominee John McCain. When all that is happening at the same time, and then the debates are followed up by the FBI director resurrecting an investigation into one of the candidates, it’s next to impossible to isolate the effect of individual factors like debate performance.

Sides, Vavreck, and co-author Michael Tesler find in their book on 2016, Identity Crisis, that the first debate and the Machado controversy led to a “sharp drop in the percentage of voters who rated [Trump] favorably.” They find the second debate had a muted impact, but that after the third debate, “Republican voters who had soured on Trump after the first debate and release of the Access Hollywood tape returned to him.”

They do not credit this return to his debate performance, which was not viewed as exceptionally good at the time, but to the tendency of partisans to return to their candidate, the failure of a critical mass of Republican establishment figures to reject Trump or endorse Clinton, and several other factors. It doesn’t add up to a compelling case that debates were decisive, either.

“Raw polls at 50 days out, on Sept 19, one full week before the first debate, had Clinton at 51.7 percent; on October 26, one week after the last debate, the share was 52.4 percent,” Wlezien notes in an email. “There is even less movement — a 0.4 percent drop in Clinton share — of house-adjusted poll averages.”

“‘Winning’ debates is overrated,” Erikson concluded in an email to me. “Ask John Kerry or Hillary Clinton.”

Will 2020 break this streak? I suppose it’s possible, but the evidence is lacking so far. The Economist’s poll average, which corrects for polls that don’t weight by party and thus can avoid the kind of non-response bias described above, found that Biden was averaging 53.9 percent of the two-party vote on September 29, the day of the first debate, and … 54.2 percent on October 15. That’s a net gain of 0.3 points. And given everything else that has happened, especially Trump’s Covid-19 diagnosis, it’s hard to even credit that minimal shift to the debate.

There’s no reason to think debates generate new, previously unavailable information

The finding that debates don’t really matter is not, on its own, proof that debates are useless. Maybe individual voters’ opinions are moved by them in a durable way that influences their eventual vote, and the movements simply cancel each other out. You could mount an argument that the debates are thus informing voters even if this does not have an ultimate effect on vote shares.

This does not look super-likely when you dig into particular cases. Indeed, debates instead can be a tool for entrenching polarization and bringing voters in line with their preferred candidates’ views — seems to be the case.

Political scientist Gabriel Lenz has analyzed the effects of Social Security-centric debates in the 2000 general election and found that they “did not cause voters to switch their support to whichever candidate agreed with their own position on that issue.” Instead, Lenz found that the debates led voters — specifically voters who learned the candidates’ issue positions for the first time — to adopt their preferred candidates’ positions. Gore supporters came to love Gore’s idea of keeping excess Social Security/Medicare revenue in a “lockbox,” and Bush supporters came to love Bush’s idea of investing some Social Security taxes in the stock market.

That’s not necessarily a bad thing — some of these voters presumably changed their minds because they trusted Bush or Gore, and changing your opinion to account for learning the views of someone you trust is very rational behavior.

But that phenomenon is clearly not the rationale offered for presidential debates. They’re supposed to help voters make up their minds — advocates and researchers Diana Carlin and Mitchell McKinney argue “the debates could be crucial shaping new voters’ choices and deciding the outcome” — not help parties convince their bases to adopt their candidates’ positions on important issues.

And insofar as the latter is a valuable service, it does not seem like the job of news organizations, but of the parties themselves. ABC News shouldn’t have to invest its money and resources to help the Republican Party rally supporters around Trump’s ideas, or to help the Democratic Party rally supporters around Biden’s.

This still leaves the theoretical possibility that there are some voters who change their minds due to debates, but whose votes cancel each other out so there’s little net effect. This, I would argue, is only valuable insofar as these voters are gaining new, accurate information about the candidates.

There’s just no reason to think that happens. General election candidates have many, many avenues through which to distribute their platforms to the public, including many more than existed at the advent of television debates in 1960. The campaigns post their platforms online, they explain them in depth to a wide number of news outlets (including online outlets without the word limits that newspapers and magazines labored under), and if voters miss an article, they can simply Google it. Primary debates, where issue disagreements between candidates are minimal and most voters are undecided, probably give voters useful information. The same likely can’t be said of general election debates.

Debates might offer some value in creating a single, focusing event that lets low-information voters get back up to speed. But for them to serve that purpose, the information would need to be accurate, and the 2020 debates to date are excellent illustrations of candidates’ abilities to lie without being interrupted by moderators.

Vice President Mike Pence, for instance, was able to claim without correction or interruption that Biden and Kamala Harris want to “abolish fossil fuels and ban fracking” (they do not) and that the White House event that produced a cluster of Covid-19 cases was outdoors (the reception was indoors). The co-chair of the Commission on Presidential Debates has gone so far as to say that fact-checking isn’t the moderator’s job. Someone tuning in to the vice presidential debate and hearing that Biden wants to ban fracking is not actually gaining useful, true information about the candidates.

If we want another kind of central, focusing event, something like single-candidate town halls with the individual candidates and — this is key — a moderator who is equipped to fact-check, or perhaps a simulation where the candidates both attempt to complete a modeled national security crisis, would be far likelier to produce useful information for voters.

We do not have as rigorous social science for the proposition that debates do not lead viewers to have more accurate information about the candidates as we do for the proposition that debates don’t swing many votes. But the content of the debates themselves, and the existence of numerous accurate sources of information on the candidates, strongly suggest to me that debates, even ones between relatively responsible candidates like Romney and Obama, do not improve public understanding of the issues discussed.

So the question then is: If debates do not surface information voters cannot get elsewhere, and they cost a ton of money and effort to put on, and they can be a means to propagate misinformation, and they do not influence the ultimate vote totals … why do we insist on continuing to hold them?


Will you help keep Vox free for all?

The United States is in the middle of one of the most consequential presidential elections of our lifetimes. It’s essential that all Americans are able to access clear, concise information on what the outcome of the election could mean for their lives, and the lives of their families and communities. That is our mission at Vox. But our distinctive brand of explanatory journalism takes resources. Even when the economy and the news advertising market recovers, your support will be a critical part of sustaining our resource-intensive work. If you have already contributed, thank you. If you haven’t, please consider helping everyone understand this presidential election: Contribute today from as little as $3.

Share this

Leave a Reply