I bet you thought you’d stop hearing about the U.S. election now, didn’t you?
Think again, it’s not over yet.
Now that the election has happened, social media geeks like us can comb back over all the data they’ve collected throughout the past months and look at it in a new light – how did it actually impact things?
Take the debates, for example.
During the presidential and vice presidential debates, people flocked to social media to share their agreement or disagreement with certain talking points. News outlets rushed to declare winners and analyze each response.
But what did all that even amount to? Were social conversations representative of what actually happened?
Now that the election is over and we’ve seen the results, let’s take a look at how debate conversation might have shaped President-Elect Donald Trump’s win.
Presidential Debate #1: September 26, 2016
The first debate set our patterns and expectations, both in terms of what to expect from the debate itself, and the public and media’s reaction to it.
We had seen each candidate debate in the primaries, but given a volatile election season and high tension between parties, the first presidential debate would help to set the tone for the general election season.
Let’s take a look at how social media users and the media reacted to it.
First of all, let’s look at how people discussed the first debate overall, over time. Here is the total conversation the week of the debates:
Obviously, the most conversation happened on the day of the debate – that’s a given.
But what’s interesting, especially once you break it down and compare it to other dates as we will late in this report, is the lead-up and “tail.”
Given that this was the first debate of the general election, and the strong opinions of both major party candidates this year, you’d expect a fairly large amount of conversation leading up to the event – especially the day before.
However, the volume of conversation in the few days before the debate remains fairly flat. Why weren’t social media users and the media talking about the debates more and more leading up to it?
It could be that the first debate was so heavily anticipated that the leadup started way before that week, and therefore had already reached its “pre-debate conversation peak” by the time the week started.
Of course, it could also be that with two candidates frequently embroiled in scandal, the media and its audience were focusing on different areas of election news.
Regardless, people seemed to be more interested after the debate than before it, as there were more conversations on September 29-30 than on the 23-24.
As this was the first debate, it was too soon to tell if this pattern of conversation is typical for this year’s election cycle.
For example, once people have watched a debate or two, and the media has seen which parts of them the audience reacts to, will we see more targeted conversation prior to debates as anticipation builds?
Will future debates prove more controversial and cause people to keep talking about it for longer afterwards?
We’ll look at this later.
Once again, maximum reach does happen on the same day as the debate.
Twitter is still the go-to source for many for real-time conversation, discussing events airing on television, and discussing breaking news. As a presidential debate aired as primetime television hits all three of those points, there was obviously lots of conversation.
This segment also mirrors the overall volume in terms of its decline. With a fast-paced news cycle and the real-time aspect of social media, people stopped talking about the debate pretty soon afterwards and moved on to something else.
Web / News Only
News and web coverage follows a slightly different path. Because of all the post-event coverage, its conversation peaks the day after the debate and then drastically drops on day 3.
Given the 24-hour news cycle, as well as the nature of campaign coverage, it’s not surprising that outlets quickly move on to a different area or story.
Mention reach tells us two very important things about the conversational trends around the debates: both quantity and impact.
As more web publications – and ones with larger reach – begin publishing more and more debate coverage, the debate conversation’s reach will increase in that regard. Additionally, as the coverage is shared across social media, it reaches new people – maybe even reaching viral status.
As you can see, mention reach peaked on September 27 – the day after the debate. This may be because volume contributes to reach. If the debate hashtag was being used by more people and in more tweets, and it was talked about in more news articles from more publications, reach would be impacted. However, the mentions and conversations would take time to pick up steam.
Now, what were people actually talking about? We were interested in looking at this to see if people were speaking to logic or emotions, sharing facts or opinion.
As you can see, it’s a mix of both.
In addition to basic pieces of information about the debates – candidates’ names and Twitter handles, descriptions, and information about the location – there were some more emotional discussions.
For example, the “deplorable” keyword speaks to people talking about an offensive comment made by Hillary Clinton about Trump supporters. “Fact” relates to the live fact checks media organizations do during the debates, speaking to the public’s distrust of both candidates.
And with “drinking” and “game” both trending, a lot of people were obviously about the sardonic coping method that goes viral before each debate.
Vice Presidential Debate: October 4, 2016
I predicted that in terms of conversation patterns, the vice presidential debate would have a few things in common with the first presidential debate, and a few things in common with the last.
After all, as the only VP event, it was both the first and last time Pence and Kaine shared the debate stage.
Let’s take a look at how people actually talked about it online.
In terms of overall volume trends, it’s pretty generic and what you would expect.
A little bit of conversation before and after the event with a huge bump on debate day. Because this graph looks at both web and social mentions, live reactions and news coverage have the opportunity to balance each other.
There was a bit less conversation leading up to the debate than we expected, but I think we can chalk this up to two things:
- The vice presidential debate may not have attracted as many viewers as one with the main candidates.
- The main candidates tend to dominate the regular news cycles, meaning both news coverage and the conversations it can cause focus less on the vice presidential candidates in general.
Let’s look at a closer breakdown.
Another sad graph that shows people may not have cared about the vice presidential debates too much.
Whereas with the first presidential debate, social conversation was still high the next day (decreasing by approximately 50% on day two), there was a much higher drop-off with this event.
This shows that people may have moved on more quickly than with events featuring the presidential candidates. Given that the spotlight on Pence and Kaine was so temporary in nature, this debate may have been a fleeting victim of the quick social media news cycle.
“On to the next debate…” may have been the feeling on social media. Once it was over, it was over.
Now onto web / news conversations…
Web / News Only
It seems that news coverage patterns for the vice presidential debate were a little more predictable and in line with the other debates.
Instead of one large spike in conversations on the day of the debate, web mentions started building up a day or two before the debate. They then peaked the day after it, as publications started posting online content about the actual event.
After that, you can see it drops pretty drastically as Pence and Kaine probably wished they had as much charisma and meme-ability as Biden. 😉
The reach evolution for the VP debate was a little strange, and seems to further enforce that people weren’t that interested in talking about this debate.
Even though reach peaked the day after the debate, as you might expect, the huge drop in conversation reach on October 6 is a little unexpected. Given that the reach would increase as people shared and discussed the coverage published after the debate, the big drop suggests people weren’t interested.
To wrap up the debate of predictability, most of the popular keywords being used alongside the debate were straightforward and informational. Taking a look at the word cloud above, you can see they were mostly neutral/fact-based posts being shared.
But was the vice presidential debate an outlier this debate season, or was the first presidential debate? Looking at the other events will help us decide.
Presidential Debate #2: October 9, 2016
The second presidential debate was our chance to confirm or deny the patterns we noticed in analyzing the first debate.
We also got to observe how both mentions and tensions increase as election day looms closer. Between your average online hostility around discussing politics (thanks, trolls) and the leaked Entertainment Tonight taped conversations of Donald Trump, stakes and emotions were high in this debate.
Let’s see how that impacted online conversation.
Overall volume for the third debate of the election cycle, the second featuring presidential candidates, was a little bit lower than conversation the first time around.
The three reasons I saw for this were:
- This debate occurred on a weekend, when primetime TV viewership tends to be lower anyway. Less people watching, less people talking.
- As this election season continues, people may be getting tired of watching so much live coverage of it and discussing in real-time.
- A main talking point of this debate in the media (the leaked tapes) was its own separate story for news outlets to talk about on its own.
But let’s take a closer look.
The Twitter conversation about this debate was pretty unusual when you look at the graph and compare it to others of Twitter debate buzz.
It’s extremely unusual in this set of data for there to be almost as much conversation the day after the debate as there was during the actual event.
So either way less people live tweeted the debate than usual (either because it was a Sunday or due to debate burnout) on Sunday night, or way more people kept talking about it the next day.
After day two, the steady decline we would expect from a live tweeting graph makes an appearance, with the big drop just happening later than in other news cycles.
Web and news mentions for this debate also broke pretty far away from the normal patterns we saw for others.
For example, pre-debate discussion was much higher with previous debates, particularly the day before the debate.
Again, I think this can be attributed to the other political news stories going around that weekend. Since Donald Trump’s Entertainment Tonight tapes leaked so close to the debate, a lot of coverage about it brought up the debates and how it might play into them.
So while there may not have been as much web coverage of the debates specifically, it was brought up in more related political stories before the debate.
Afterwards, we saw the expected drop, somewhat more drastic than we may have predicted given next-day recaps and coverage.
Overall reach may have been the one statistic for this debate’s conversations that actually panned out as we predicted, as expected based on previous data.
Because of the large volume of conversation the day before and day of the debate, the momentum kept going for overall reach to peak the day after the debate.
Again, a lot of the top keywords were fact- or information-based in nature, like each of the candidates’ Twitter handles, names, and such showing up in the tag cloud.
However, talking points related to the debate issues popped up as well, such as treatment of Muslims and women in the US, in addition to “Bill” as Mr. Trump brought Secretary Clinton’s marriage into the discussion.
Presidential Debate #3: Wednesday, October 19
The final and fourth presidential debate came during a kind of quiet week for the election. It was further away from the other debates and instead surrounded by more hardcore campaigning on behalf of the candidates as the election drew closer. Let’s see what that meant for online conversations.
The conversation here was pretty standard, with a slow build-up in the days before and after the debate, segmented by a huge increase on the debate day itself.
You do see in the graph that there was a bit more conversation than usual in the days leading up to the debate. This could be because the press wasn’t still talking about a previous debate that had just happened a few days ago, as it was with debates happening once a week earlier in the fall.
Twitter had a more “normal” pattern, with conversation staying relatively quiet except on debate day. Again, I’m sure election fatigue was a factor here!
The web and news conversations, on the other hand, were a little bit different. Instead of peaking or dropping the day after the debate, conversation remained about equal. Once again, this is something that could be reasonably explained by timing and what else it was competing against in the 24/7 news cycle.
The reach for this debate is where we saw a little more predictability – finally! As you can see on the graph, as different content about the debate conversations pick up steam, culminating in a peak right after the debate. After that, people kind of stop talking.
This debate seemed to have more issue-driven conversation than fact-based statements. In the above topic cloud, you’ll see things like “women,” “nasty,” and “#draintheswamp,” which were all likely discussions more about emotions and opinions and facts.
What does this all mean?
This data shows how people react to politics in an election season. Now that the election’s over, we can look back at these conversations and see with more clarity whether or not they changed the results.
Did people talk more about the debates that impacted their choice? Does conversation equal action? Now that we’ve seen the results, we can decide for ourselves.