A Look Back at Debates and Polling

Charles Burgess, Jordan Moskowitz, Jack Ward, and Westly Garcia

 

The first presidential debate of the 2016 election cycle has come and gone, reaching record numbers for viewership across a wider number of platforms than ever before. Thought pieces now cover print and screen with questions of how the candidates did, and, more importantly, how their performances will affect voters. But maybe it is better to ask: do debates have much of an effect on polling? A comprehensive, hundred-year study is outside the scope of this post, but we attempt to provide a shallow look at how debates have affected polling in the last few election cycles: Jordan covering this year’s first debate between Clinton and Trump, Charles with debates from the 2012 Obama-Romney race, and Jack with 2008’s Obama-McCain race.

 

The First Debate of 2016

 

The debate was pretty standard. It was Trump being Trump and Clinton being Clinton. I don’t think the candidates disenfranchised any voters, but not many were won either. Trump dominated the first third of the debate as the questions were on issues such as jobs, taxes and regulations. Then the debate was able to take a turn as Lester Holt asked some very biased questions against Trump such as on his tax returns, the whole birther movement, and beauty pageants while avoiding questions directed towards Clinton on her email scandal, Benghazi, and her Clinton Foundation. I think that this election will be decided on what issues the American people decide that they care about whether it be trade, jobs and regulations, or tax returns, beauty pageants and birther questions.

 

After the debate there were many sources that took online polls to determine who won the election.  A majority of them determined that Trump was able to win the debate.  Now these are not exactly scientific, and it might be a little while until more polls come out about the debate, but I think this shows two things: Trump did a lot better than expected and he definitely didn’t lose any voters.

 

Now there are two polls that I want to talk about specifically.  The CNN poll and the Drudge Poll.   The Drudge report is the leader in news aggregation, has a strong following, and been determined to be one of the media outlets that will have an impact on this election.  On his poll of the Debate, he had 500,000 votes and said that Trump won with 82%. Now a lot of visitors on his site tend to lean to the right, it still shows tremendous support for Trump.  Now the same thing can be said for CNN’s poll, which leans left and determined that Clinton won with 62%, but only asked 521 people.  47% said it would not affect their vote, 34% said it moved them towards Clinton and 18% towards Trump.  I think that the first debate, because of its biased questioning that helped Clinton put Trump on the defensive, didn’t anything to lose voters, but didn’t really sway any voters wither.

 

Now a very interesting determinant for the election has been the Mexican peso.  Investors clearly thought Clinton won and the peso shot up 1.5% against the dollar on Tuesday.  This will be a good forecasting tool to look at for the election in the coming months.

 

The 2012 Cycle

 

In October 2012, there were three key Presidential debates between Barack Obama and Mitt Romney. These debates fell on the days of October 3rd, 16th, and 22nd. By analyzing several different polls, I hope to illustrate that the 2012 Presidential debates had an influence over the 2012 Presidential election.

 

The image below comes from a site called Real Clear Politics. The data is a weighted average of various different poll results. The highest margin of error is 4% and the lowest sample size is roughly 800 people. Therefore, they are fairly strong sample sizes from the shallowest analysis. This data draws from sites such as CBS, FOX News, Wall St. Journal, Politico, and various other noteworthy sources. As a side comment, I am unsure how the weight is applied to specific sources, so there might be some bias in that sense.

 

From the line graph below, one can see that the spread between Obama and Romney support before the first Presidential debate is quite high at almost 4%. However, as the first debate came to a close. One can clearly notice a distinct increase in Romney and decrease in Obama’s voter breakdown. If you recall, in this debate there was a clear and decisive winner. That winner was Mitt Romney, the polls reacted accordingly. In their second debate, there was not a clear winner, this also can be seen in the polls as very little change occurs. With the debate results and the change in polling results in mind, I believe it is possible to make an educated prediction of the Presidential election outcome based off of debates.

picture1

The 2008 Cycle

 

During the 2008 election cycle, Barack Obama and John McCain faced off in three televised presidential debates, with the first being held on September 26th at the University of Mississippi by PBS. While the original plan was for more of a foreign policy debate, half was dedicated to the economy in light of the recent financial crisis. Ultimately, most polls ended up pointing to an Obama win: Talking Points Memo’s poll (via CNN) gave Obama a 51-38 victory over McCain, and a CBS poll gave it to Obama with a 39-24 spread (with the rest being undecided). Nate Silver attributed the first night’s victory to Obama’s speaking to middle class voters, something the transcript could not find for his opponent.

 

But did this create an actual difference in the horse race? Specific causes can be hard to tease out of noisy polling data, as presidential elections are chaotic systems. However, if a clear gap were to quickly emerge between the candidates after the late September debate, it would be easy to point to the debate as the cause, as long as there weren’t any other major events around that time that would interfere.

 

As in Charles’ coverage of 2012, Real Clear Politics provides good historical data to this end. Their information from August until the election is reproduced below, with a line at the date of the first debate:

 

picture2

The data proves unconvincing. There is a small McCain drop around September 26th, but one that he regains just days later. Additionally, FiveThirtyEight’s polling experts enjoy emphasizing that polling lags behind current events by around five days. Even after that wait, there is no spike in either candidate’s numbers. Obama does slowly gain on the +4.2 spread he held on the night of the debate, but not quickly enough to attribute it to his debate success alone.

 

Ultimately, it is unclear how linked debate performance and polling success are. The two seem to move together in 2012, but the correlation is lost in 2008. Debate performance may be a useful tool in predicting elections, but more research with more debates going farther back is needed to make any solid conclusions.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s