For the second set of Democrat debates in Detroit, observers were looking to see if both Joe Biden and Bernie Sanders could bounce back from less than enthusiastic reviews from the first round, which led to weakening poll numbers over the past weeks for both. Kamala Harris and Elizabeth Warren were looking to build off strong performances in the first round which gave them both positive polling momentum after the Miami debates.
For those currently occupying the top six positions in the race, the Tuesday debate was positive for Warren, O’Rourke and Buttigieg. Williamson had the strongest online numbers of those candidates outside the top six as she was the second most discussed candidate during the debate, and her mentions were the most positive of any candidate. The fact she was also the most Google-searched candidate in 49/50 states during the debate indicates a high level of curiosity about her. Of the frontrunners, Sanders’ performance had the worst volume and sentiment combination with the highest volume, but the lowest sentiment of any candidate on the stage.
In Wednesday’s debate, Biden put in a performance that is unlikely to generate new support as he was by far the most discussed candidate while earning a slightly negative sentiment. As the current race leader, however, it is reasonable to attribute a larger than normal share of the negative sentiment to his frontrunner position. In his position, a slightly negative sentiment indicates he still has a strong base of support keeping the negativity in check.
Kamala Harris had the most difficult night with the second-most mentions but also the second-lowest sentiment. This could effectively put a halt on the growth in support that she enjoyed after the first debate.
Andrew Yang had the best performance of candidates outside the top six on Wednesday with the third most mentions and the best sentiment score. As Yang was starting to move up in the polls pre-debate, it will be interesting to see if this performance gives him a boost into the top six as O’Rourke and Buttigieg continue to slide.
Over the two nights of debates, we analyzed comments from approximately a quarter million social media accounts. The nature of real-time, short-text comments means that the vast majority of online mentions focused on the candidates, as opposed to the issues they are discussing. However, we sampled our dataset for comments with specific mentions of candidates along with issues to get an idea of how the public reacted to each of the candidates speaking on the key issues.
Same as in the first round of debates, the two candidates that are most aligned with the Medicare for All stance, Sanders, and Warren, were mentioned often throughout the night, but again with lower sentiment scores than many of the other candidates.
However, the issue was problematic for all of the frontrunners as both Biden and Harris earned some of the lowest sentiment scores on the issue. We found in the first debate that the Medicare for All candidates were not as well-received as those with a Public Option stance, but the water has since been muddied reflecting the lack of clarity and details from many of the candidates on what they are proposing. Inslee, a Public Option proponent, had the highest sentiment score on the issue.
Same as in the first debate, Castro and Klobuchar pulled in some of the most positive comments on the immigration issue, while it continued to be a problem area for Joe Biden. Biden’s volume and sentiment combination was the worst of the candidates, and the issue was another trouble spot for Kamala Harris as she received a very negative sentiment score on the issue. Sanders also struggled on immigration with a low sentiment score.
3) Climate Change
Jay Inslee led the charge on climate change with both the highest volume and the second-highest sentiment score. The issue was also strong for Yang with the third-highest volume and positive sentiment.
The issue remains a challenge for many of the candidates with only seven positive sentiment scores. The lowest scores came from Hickenlooper, Delaney, Harris, and Bennet.
There was enough discussion pertaining to the Mueller hearing and impeachment to warrant an analysis of the issue. Sanders generated the most comments but again found himself attracting negative sentiment.
Booker’s positive debate reviews seem to have come primarily from these discussions as he generated strong volumes with one of the highest sentiments.
Castro also had a strong volume and sentiment combination while Yang scored the highest sentiment, albeit on low volume. Ryan and Gillibrand did not perform well on the issue with the two lowest sentiment scores.
5) Racial Equality
The issue of race did not have quite the same level of fireworks in the second round of debates as it had in the first round as both Biden and Harris were unable to generate the same levels of discussion this time around.
Williamson generated the most discussion, but it was not well received online with a negative average sentiment. In fact, the four most discussed candidates on the issue all garnered negative sentiment.
O’Rourke had the best combination of volume and positive sentiment, followed by Bennet and Gillibrand. Both Biden and Harris had low to moderate volumes and negative sentiment indicating that the busing issue has likely run its course.
1) If there will be any movement in the polls based on the July debates, it will likely be a return to the pre-debate status. Biden’s performance improved and may have slowed some of the support declines he was encountering, but it is unlikely he will reverse that trend. Harris will likely lose some of the gains she made over the last month, and other than Williamson, Yang, and to some extent, Booker, no lower-tier candidates caught the attention of online users.
2) The frontrunners are having difficulty resonating with online users on the issues. When they generate the most volume on an issue, the sentiment is usually negative. Conversely, when candidates from outside the top tier are generating large discussion volumes on specific issues, the sentiment tends to be more positive.
3) The most likely candidates to benefit from the Detroit debates in the polls will be those currently in the five to eight positions: Buttigieg, O’Rourke, Yang, and Booker. All performed well not only in candidate mentions, but they were also some of the best performers on the issues giving Buttigieg and O’Rourke a chance to reverse their downward trend, while Booker and Yang can build on their momentum gained since the last debates.
Something interesting we’ve seen over the past month is the online volume trends of the top candidates in the Democrat Presidential Primary. We will often see large daily variations in volume, but the top tier of candidates has generally been maintaining consistent average volumes over weekly or monthly periods.
The consistency we normally witness is why it is interesting to see the volume trend of discussions regarding Joe Biden throughout July. Biden’s average volume during the past two weeks of the month dropped 30% from his average volume during the first two weeks. In comparison, Warren and Harris saw their volumes decrease a more reasonable 11% and 15% respectively, while Sanders saw a huge increase of 49%, although it came with the largest decrease in sentiment, primarily due to the negative comments surrounding his campaign’s minimum wage issues.
The lack of online discussion about Biden starts with a lack of reasons to discuss him, and this could be due to the other candidates pursuing the media’s attention more aggressively, or simply a decision by his team, such as more debate prep over the past two weeks. Whether it is coincidental or by design, the numbers indicate his campaign team has been comfortable allowing others to take the spotlight in recent media cycles. Whether this continues post-debate is a trend we will be looking for in our online data in the days ahead.
The President's recent Baltimore tweets have caused yet more criticism in the media and online, but will it have any impact on his overall support?
The past two weeks have given us a number of examples that demonstrate the pattern of activity when the President says or tweets something controversial. Tweets about the “Squad” that garnered massive attention, a rally where the “send her back” chant took over the media narrative for days, and the Mueller testimony all made for stories that generated huge volumes of discussion.
Through our partnership with Scott Rasmussen (http://bit.ly/SRTrump), we’ve been monitoring the online sentiment expressed toward the President, which gives us real-time insights into the public reaction to his actions.
If we focus on the last two weeks, we can see the large initial drop in sentiment caused by the Squad tweets. As we discussed in a previous post, we found most of this negativity expressed by people who were previously negative towards the President, and it wasn’t impacting his base of support in a meaningful manner.
From the chart (h/t www.scottrasmussen.com), we see that POTUS’s sentiment started to recover within days. It also appears that the House condemnation of his comments and how that event played out may have provided the impetus for the recovery. The initial negativity of the rally chant the next day put the brakes on the recovery, but it then continued through to the end of the week.
The same pattern occurred when Robert Mueller testified at the end of the past week. Initial negativity drove sentiment temporarily lower, but it is already recovering back to the levels witnessed before the Squad tweets.
It is this recovery from the negative opinion that has been a constant characteristic of the President’s sentiment scores over the past months. While events fuel non-supporters comments in the short term, Trump supporters tend to rapidly come to his defense to change the negative momentum to positive.
This pattern makes sense when we look at what polls have been showing for the President’s approval ratings. Just as our average sentiment tends to stay in a range we would characterize as slightly negative, the Scott Rasmussen/Harris X polling finds the same consistency in their numbers as well. As pointed out in the July 25th update, "since January 28, the number approving of the president’s performance has stayed within two points of 46% every single daily release." The consistency found in the polling results mirrors the consistency we see in the President’s sentiment over the medium to long term.
Overall, for all of the ups and downs that appear to be playing out in the Oval Office, the reality is that public opinion of the President tends to be remarkably consistent regardless of his actions.
Understanding this is likely a key factor in the President’s communications strategy, and Eyesover can provide this information to any organization, in real-time, to help inform strategic decisions.
When a negative story breaks in politics, it is always a challenge for the politician or political party to know what their next step should be. Whether the story is breaking in traditional media or online, in those first few hours, it is next to impossible to know how big the story is, who is talking about it, or how it is impacting the politician or party.
Let's take a look at Donald Trump's recent tweets about the "Squad." There is no question the overall impression of the tweets and the related backlash took a decidedly negative tone. Trump's overall average sentiment dropped 65% over the three days after he tweeted his comments (-0.0982 to -0.1618 on a -1 to +1 range via www.scottrasmussen.com). While our analysis usually finds the President’s sentiment in a slightly negative territory, it takes an enormous number of negative comments to drive the average sentiment below the -0.15 level.
But from where is the negativity coming? Is Trump losing supporters due to his comments, or is the backlash primarily from people who would never vote for him regardless of his actions? Many feel that much of Trump’s motivation for his approach to social media is to reinforce his base, so the importance of knowing the reaction of those who have previously indicated support for him can't be overstated.
As it happens, this is exactly the kind of analysis you can perform with Eyesover. From a sample (n=126,962) of accounts commenting on the President’s tweets over the past few days, we can determine how many people have changed their opinions towards POTUS. To do this, we compare the sentiment of previous comments to the sentiment currently being expressed from the same account.
The numbers show that while some previously positive individuals have turned negative, Trump has been gaining new supporters at roughly the same pace over the past few days. The lack of a significant change in supporters indicates the negativity shown in his overall sentiment scores is largely caused by individuals who have always spoken negatively about the President and are outside of the base of support he has built and is trying to maintain.
In any campaign, it is critical to know if changing issues require a change in strategy to mitigate risk or take advantage of opportunities. Eyesover provides the data campaigns need to determine when to change strategy or when to maintain the status quo.
While recent polls have shown a tight race between the Conservatives and the Liberals for the lead in the Canadian political landscape, at the same time, there has been some interesting movement among the NDP, Greens, and PPC as well.
When we look at the share of online mentions for each party during June, we see the majority are discussing the Liberals (61%), which is understandable due to the fact they are the governing party. The Conservatives follow at 28% while the New Democrats, People's Party, and Greens are 5%, 3%, and 3% respectively.
The fact that our tracking has found a sustained downward trend in sentiment over the past few months for the Liberals aligns with their recent fall in the polls. The trend continues to indicate that much of the online activity for the Liberals has been increased criticism of their recent actions.
But while the focus tends to be on the LPC and to a lesser extent, the CPC, we can't forget that there is an increasing likelihood that the NDP, Greens, and PPC will have a say in who will be able to form a government this fall.
When we look at the third, fourth, and fifth party mentions online, what catches the eye is the volume of PPC mentions. The party was discussed more than the Green Party in June and had just over 50% the number of mentions of the NDP. For a one-MP party that is less than a year old, there is considerable online traction.
Geographically, comparing where the mentions are coming from compared to provincial populations, the PPC mentions are over-weighted in Ontario, Alberta and BC despite the fact their sole MP, Maxime Bernier, is from Quebec.
The share of online mentions we show above aggregates daily mentions from the same account so it does not skew the one person, one vote concept, but we can also dive into the raw data to get an idea of how active people are when it comes to mentioning a party. To make sure our analysis does not get skewed by fake accounts or bots, we apply a thorough vetting to ensure accounts represent real individuals or organizations, and we ignore non-original comments such as retweets.
Of particular interest with the PPC is the fact that the average number of mentions from an account is 2.46. This level of activity is far higher than any of the other parties (CPC: 2.04, LPC: 1.89, Green: 1.72, NDP: 1.56) possibly indicating the PPC has a higher level of support intensity than other parties. The low number for the NDP could be an indication of a lack of enthusiasm within their base.
Something we will be watching over the next few months is to see how concentrated the PPC support is, and if they can turn their base's enthusiasm into a larger vote count in some ridings.
With the completion of the first set of Democratic debates, it appears that there is a much tighter race ahead, with Joe Biden dealing with the fallout from poor reviews of his performance, and positive reviews for several other candidates—Elizabeth Warren and Kamala Harris in particular.
As we reported post-debates, online activity indicated that among the top six candidates in the polls, Warren, Harris, and Buttigieg served themselves well in their debates while Biden, Sanders, and O’Rourke did not.
Looking at the candidates that have been on the outside of the top six, Wednesday saw Julian Castro with the best combination of mention volume and positive sentiment. Along with Castro, reactions to Tulsi Gabbard, Cory Booker, and Amy Klobuchar were positive as they all either generated significant discussion online, positive sentiment, or a reasonable combination of both. The remaining four Wednesday candidates had combinations of low volumes or low sentiment in their mentions.
Thursday provided a few surprises as Marianne Williamson and Andrew Yang were the two candidates that attracted significant positive attention. Of all of the candidates, Williamson’s performance generated the most attention of the candidates that had previously been polling in the 0-1% range. Her challenge will be to maintain the attention and translate it into support. The remaining four Thursday candidates had combinations of low volumes or low sentiment in their mentions.
Over the two nights of debates, we analyzed comments from approximately a quarter million social media accounts. The nature of real-time, short-text comments means that the vast majority of online mentions focused on the candidates, as opposed to the issues they are discussing. However, we sampled our dataset for comments with specific mentions of both candidates and issues to get an idea of how the public reacted to each of the candidates speaking on the key issues.
The healthcare discussions online most mentioned the two candidates that are most closely aligned with a Medicare for All stance: Sanders and Warren.
The sentiments of the Sanders and Warren mentions were average among all the candidates (slightly positive), but the sentiment in mentions of Buttigieg, Delaney, Castro, Ryan, and Yang was found to be higher, possibly indicating the public is more comfortable with a Public Option approach.
The highest sentiment was found in comments mentioning Julian Castro, albeit on a relatively small volume.
Castro, Klobuchar, and Buttigieg had the best combinations of high volume and high sentiment over the two nights on Immigration. The issue proved to be one of Biden’s trouble spots as he was mentioned in the highest number of comments, but with a negative average sentiment. It is an area that could cause him problems moving forward.
Both Warren and Sanders also had negative reactions, but at a lower volume. Andrew Yang immigration mentions had the highest sentiment but on very low volume.
3) Climate Change
Jay Inslee was successful in making sure his voice was heard on the climate change issue as he generated the most mentions, and his message was well received with his mentions being slightly positive.
Buttigieg, Yang, and Castro mentions had the highest sentiment, but the striking metric on this issue is the negative reaction to Kamala Harris. With the average sentiment from her mentions well below the average, it could be an indicator that her endorsement of the New Green Deal could be an issue for her campaign.
4) Reproductive Rights
As we found over the first three issues, the basis of Julian Castro’s strong overall performance was policy, and his message on reproductive rights was no exception as he had the strongest combination of mention volume and sentiment in those discussions.
Both Warren and Gillibrand generated significant mentions but failed to resonate as the sentiment was slightly negative.
The issue also shows another key issue on which Harris did not perform well with both low volume and negative sentiment.
5) Racial Equality
The most discussed exchange from either debate was the Harris/Biden discussion regarding Biden’s past stances on school busing. A lot of the post-debate narrative revolves around the belief that the exchange was good for Harris and bad for Biden, but our numbers don’t reflect that line of thinking. Instead, the sentiments for both Biden and Harris mentions were in negative territory, indicating that it likely will not have a dramatically positive or negative impact for either candidate.
And again, one of the best combinations of volume and sentiment went to Julian Castro.
Our issue-based analysis of the debates found several key points that indicate the race will become tighter over the next few months.
1) Issue-based messages from candidates currently outside the top six were much better received than those from the current top six candidates. Julian Castro performed the best of all candidates on the issues as he had positive reactions on large volumes of mentions across all of the issues. Andrew Yang had some of the most positive sentiment in his mentions, but he struggled to generate much volume. Of note is that the two perceived winners of each debate, Warren and Harris, had mostly average to below-average sentiments expressed in their issue-based mentions, putting into question the importance of policy platforms or stances at this stage of the race.
2) Candidates perceived to be more on the left wing of the Party registered average to below average performances on the specific issues. Warren on Immigration and Sanders on Climate Change are two examples of messages received with very negative sentiment on strong volumes of mentions. The top sentiments expressed on the issues came from candidates taking moderate stances on each of the issues.
3) Several candidates took full advantage of the opportunity the debate provided them. Castro, Yang, Booker, and Williamson all performed in a manner that caught the public’s attention, and now they have the opportunity to spread their message to a much wider audience than had been previously listening. Delaney, Bennet, Inslee, Ryan, DeBlasio, and Hickenlooper did not and will have to rethink their approach in the July debates.
As the Democratic Primary candidates head toward the first debates, the top 6 candidates have maintained their positions with the remaining candidates struggling to break into the lead pack. Within this lead pack, however, we can still find some interesting social media trends to compare and contrast with what polls have been reporting lately.
As in the polls, Biden has been maintaining a significant lead over the other candidates online. While tracking from social media results in much more volatility than you would see in polls, we have witnessed Biden's support trending downward for the last number of weeks. The trend continued this week as questions regarding his recent comments (cancer, segregationist colleagues) have dominated Biden discussions. Despite his frontrunner position, of concern for Biden is the clear downward trend in sentiment over the past number of weeks as more and more online conversations about him are negative in tone.
We can also see longer-term trends reflected in Elizabeth Warren's numbers from the past week, as she has been on a short-term upward trend. We say short-term because Warren has been fluctuating between second and third place in our overall support metric for some time now, while the polls have only recently recorded strong increases in her support.
At the same time, the recent decreases in Bernie Sanders' polling numbers are no surprise since there has been little evidence of his campaign increasing support online for the past couple of months. With a relatively consistent sentiment and volume, it appears most online activity is generated from the Sanders base and is struggling to grow.
Kamala Harris' numbers have been relatively consistent as well, but in contrast to Sanders, her trend is slightly upwards and has been for the past month.
Of the top six, Pete Buttigieg has the most to be concerned about based on our trends. While the polls have been showing incremental increases for Mayor Pete, online activity has stagnated, and sentiment has fallen off from the very positive levels we recorded in March and April. The fact that Buttigieg is the only candidate in the top six that was not considered among the favorites early in the race certainly provided a honeymoon period of sorts through March and April, but the current online trends indicate that growing his support base is becoming increasingly challenging. These trends have resulted in the most significant percentage decrease of online support over the past month among the top tier candidates.
Rounding out the top six is Beto O'Rourke, and similar to Buttigieg, he has to be concerned about his lack of traction as evidenced in both our online analysis and in recent polls. Based on our trends, O'Rourke is at risk of dropping out of the top tier as his online support levels are approaching those of Gillibrand and Booker. While O'Rourke's sentiment has remained consistent, it appears his main challenge has been his inability to maintain his initial announcement levels of attention and excitement which has resulted in lower exposure both online and in traditional media.
Of the remaining candidates, there has been little evidence that there is another Pete Buttigieg who will emerge from the pack into the top tier. However, one potentially interesting trend has been the recent volume patterns for Kirsten Gillibrand. It appears her campaign has hit upon a strategy to get their candidate into the news cycle more often and as a result, more attention. Whether or not this will translate into support remains to be seen but the fact remains that coverage = online attention = a campaign's message spreading among voters.
With two new candidates joining the Democratic Primary race over the past couple of weeks, we are looking at how online activity for the leading candidates is shaping up and in particular, the striking correlation between this activity and the polls.
Before diving into the numbers, it is important to define our data. When Eyesover analyzes online conversations (from Twitter, Facebook, Reddit, YouTube, and news sites), we are focused on narrowing the mentions down to original content. As a result, we ignore retweets, likes and shares since there is no new content produced through those actions. We also eliminate bots, fake accounts, and spam through a variety of checks and aggregate multiple mentions of the same topic from a single account, so there is equal weighting from each account in our datasets. In effect, our "mentions" are a count of individual accounts that are "talking about" a candidate or topic. All of these actions leave us with data that we are confident will provide an accurate view of the public's opinion.
The above graph demonstrates how the volume of people talking about each candidate correlates to the candidate's standing in recent polls. Biden is well ahead of the rest with Warren, Harris, Sanders forming a second group.
An interesting additional metric is the average sentiment expressed in the conversations about each candidate. While most are in a relatively neutral range (scoring between -0.1 and +0.1 out of a -1 to +1 range), Eric Swalwell and Kristen Gillibrand are the two candidates that have scores firmly in negative territory while Jay Inslee is the only one with a clearly positive score. Swalwell's volume and sentiment combination indicates his heightened volume consists of significant criticism and as a result, is not leading to support in the polls.
Another takeaway from our monitoring is that it seems the window for entering the race and being the focus of the news cycle has closed. Throughout the spring, almost all candidacy announcements gained enough media coverage to provide candidates with an opportunity to earn significant name recognition. Recent entrants into the race such as NYC Mayor Bill DeBlasio and Montana Governor Steve Bullock, have not been afforded either opportunity, likely due to the crowded field, and as a result, neither candidate recorded any large increases in mentions online.
Small movements in this week's New Brunswick numbers, with most of the change coming from a PC decrease in support and an increase for the Green Party.
The PCs endured slightly lower sentiment over most issues, but in particular, negativity over the flood response has been increasing for the government. Last week was also difficult for the government as negative feedback from issues such as NB Power's Joi Scientific investment and NB Liquor/Cannabis NB results caused a drop in support during the week, from which they recovered over the past few days.
The Liberals fared well with positive sentiment on most issues except economic growth due to the 2015 GDP revision, while the Greens continued to increase their support with particularly strong sentiment expressed towards their climate change positions.
Increased negativity was expressed in comments discussing the PANB's approach to the nursing home issue which has been a consistent trend over the past month.
One of the most intriguing aspects to the Democratic Presidential Primary has been the candidate's fight for media coverage. With 20+ official candidates, many campaigns are struggling because they have been unable to capture the attention of the public in such a crowded field.
The ability to gain media attention (both traditional and online) over the past few months has created a clear separation of candidates into tiers. Biden, Sanders, Warren, Buttigieg, Harris and O'Rourke have succeeded in capturing national media attention and as a result have moved into one tier, with the remaining candidates a step below in both Eyesover's support scoring and traditional polls.
The path into the top tier has varied. Biden, Sanders, and Warren have been there from the start due to name recognition. Buttigieg required events and announcements to capture enough sustained attention to join. O'Rourke and Harris could be deemed hybrids as they had enough name recognition to be in the neighborhood, but their announcement bumps put them firmly in the top tier.
With all of these candidates, their announcements caused large spikes in attention online and led to increased support in post-announcement polls. Buttigieg was able to build on his spike and grow his presence, while others like Harris and Warren were able to sustain the attention they gained from their announcements. Many others, including all in the lower tier of candidates, could not capitalize on their announcement bumps and have struggled to build on their support base. Of this latter group, it has been O'Rourke who until recently had the largest decrease in attention and support from the peak that came with his candidacy announcement.
Which brings us to Biden. His announcement bump was similar to O'Rourke's in that is dominated the news cycle for 48 hours+ and caused a huge spike in mentions, but it was also similar in a negative manner since the bump was not sustained, and he is back to his previously attention levels. What does this mean for his support? If the patterns we have seen repeat, we would expect Biden's polling lead to start decreasing in upcoming polls. He will remain at the top of the list due to the name recognition factor and the indisputable fact he has significant lead provided by a legitimately strong support base. But the online patterns would indicate there is some weakness in his numbers and he is likely to start falling back to the rest of the field over the next few weeks as others reclaim the spotlight he was not able to maintain.