September 6, 2016
Clinton’s core hashtags are exceeding Trump’s for impression count despite one-half the volume.
Clinton activity is quite democratic with over 90% participation from outside the top 100. The number one clinton tweeter by volume is a furniture hustler named Martelle. It seems that Martelle is using words related to the election and football to sell furniture. Interesting.
Trump activity has a similar reach with fewer total tweets. Activity from outside the top 100 is lower at just over 84%. I am suspicious of this activity as the peak time for Trump tweeting was 4am pacific. Folks on the east coast tend to get a later start to their day and tend to use Twitter once they arrive at work. Trump timeline deliveries have very high kurosis, they are very peaky. The top Trump related tweeter seems to be rocking between supporting either candidate in an extreme form about every ten minutes. Who would schedule a tweet burst at 4am? That is something to ponder. By impressions, the number one tweeter in the trump dataset was a firm promising to help you with crowdfunding.
Surely in the spring we will have better formulas devised to work with these datasets. As always, I tend to think that the garbage and the noise can tell us more about the communication ecology than the intentional and clear messages of the candidates.
September 3, 2016
Clinton is clearly ahead as per the polls. In the longer run the key will be to understand the ways in which the valence of posts impacts their total effect. When I began this blog, the total volume of Trump posts mattered. Now it seems that total exposure for Trump the candidate is not tied to his polls. This speaks to a key idea about the q-score calculation. Increasing increasing awareness is useful, but the impact of increased awareness decreases asymptotically toward 100% awareness.
What is the upshot here?
If you have very low awareness, the benefit of publicity outweighs the damage of bad publicity. Once you pass a certain threshold, the benefit of awareness fades and the candidate just takes the hits. Now, as we have seen in this cycle, candidate support seems unlikely to drop below a certain floor.
January 20, 2016
Trump, Clinton, Sanders coming on fast. Also people on Twitter like Rand Paul. Not sure if that means anything.
Horserace journalism, if you clicked, I guess you love it.
9/28/15
Trump is winning.
Clinton is strong.
Fiorina and Sanders have compartments of support.
The core Tea Party tweeters definitely don’t like Rubio or Bush, and they just don’t care that Jindal is dropping out.
8/31/15
As of this point, the Twitter activity is cooling. Heatmaps suggest that the first Republican debate has to this point been the highpoint of activity. Furthermore, the impact of robots and other such devices has made it difficult to map Twitter activity actual election outcomes.
My leaders on Twitter: Trump, Clinton, Sanders. It is unclear if we should be considering the other candidates at this point.
8/17/15
1. Trump.
Consider the activity distribution of #makeamericagreatagain
An interesting distribution of activity, but surprisingly skewed toward weekend use. This is troubling as these are typically not high use time periods.
The relative democratization of the distribution is quite extreme, although the raw retweet percentage of 68% is too high. I take this as a likely sign of sockpuppetry.
On the Trump primary hashtag, the only meaningful activity for another candidate is #Cruzcrew. This is a bad sign for challengers.
1a. Clinton. The traffic and reach of the Clinton hashtag is remarkable, perhaps a rival to Trump. This will be a key factor going forward – the frontrunners can fairly easily disintermediate the legacy media. There are clear traces of the Clinton e-mail server and the Libya situation, although these are not particularly critical. Clinton cadence activity is normal, nothing to report there. Little activity during traditionally twitter light times, more during the workday, the main platform is the Twitter web client. Perhaps most important, one of the key nodes in this dataset is clearly a sockpuppet, with four thousand tweets regarding #HillaryClinton2016 in the past week. That’s right 4,000, the top list on this hashtag is clearly spiked with Tea Party activists, which would make the relatively positive valence of the Twitter activity impressive.
The rest is unclear. The impact of sockpuppets is really pretty intense. This will be logged in my main entry.
This update was from 8/10/15
1. Trump, Trump, Trump. His reach is enough to eclipse the field. The winner of the first debate was Mr. Trump, Ms. Fiorina was clearly second. It is far better to WIN the undercard than to lose the over-card. This dendrogram is a pretty good look into the twitter world of the election so far.
As you can see from this massive computer reading, the other candidates names only appear deep into the clustering pattern. If
1 a. Clinton. She is right there, absent the heat and light of Fox’s attempts to take down Trump, she would be a clear number one.
2. Sanders, although this could easily be an artifact of the age and computer savvy of his supporters.
3. Fiorina. Big pop from winning the undercard. This sort of reminds me of the pop that X-Pac got in undercard bouts in the 90s.
4. Carson, he at least appears.
Losers:
Walker, basically a non-starter, not even from his enemies. Jeb lost big, the polls show that as well. At the same time, this could be a win for him as he will only win the nomination through attrition and a good ground game.
Paul was not nearly as strong as I expected in this model.
Others – may be more popular off twitter, we will see if that matters.
Here is the current visual model:
This topic model can show clear activity spikes in and around the debate, just look at the area around the big red chunk on the lower right hand side.
Pre 8/10
Republicans:
In terms of Twitter activity, Trump is way dominant with a total reach nearly 16 times that of Jeb bush. Hashtags for the other candidates are in some cases, not even captured on the preview mode for the service I am using.
Democrats:
Clinton is winning. Although, a look at Instagram traffic suggests that this group is really into Sanders and the idea of a Biden run.