The below post is from July 31st and is being posted publicly today to serve as a reference for a separate post that provides an update for August of the same figures.
====================
This is part of a series of posts covering the state of the 2020 US Presidential Election. Currently, Joe Biden is far ahead in national and many swing state polls. Betting markets place him as the clear favorite with The Economist putting the odds of a Biden victory at 91% for electoral college and 99% for popular vote.
Our models, which are based on analyzing on-line and social media activity, show a different story. We see Trump with an advantage in terms of both general interest and enthusiasm – two underplayed factors in election forecasting.
No single model or method is perfect, not even polls. However, using a ‘weight of evidence’ approach, Trump continues as our forecasted winner in the 2020 election.
Alexa ‘Global Internet Engagement’ Trends
One of the most established internet popularity rankings is done by Alexa. Their ranking system is described by the company as:
“The rank is calculated using a combination of average daily visitors to this site and pageviews on this site over the past 3 months. The site with the highest combination of visitors and pageviews is ranked #1.”
A trend of each site’s rank is provided so that you can see if the site is improving or declining.
Analysis Format
With just over three months to go before election-day, voters are likely paying more attention to news and political sites. We can logically assume that once every four years, traffic to such sites receive a boost in the months prior to voting.
Our analysis is extremely straight-forward. We look at the trends over the previous two months (since June 1st) for a variety of news and political sites and determine if their ranks have been improving or not.
We chose 40 sites, 20 with apparent biases towards conservative and 20 towards liberal messaging.
The Results
Republican-centric news and political sites have fared much better over the past two months. Of the 20 sites, 18 have trended higher.
Of the 20 Democratic-centric news and political sites, only four have trended higher.
Table 1: Alexa Site Ranking Trends from June 1, 2020 to July 29, 2020, for Liberal and Conservative Centric Sites
| Liberal-Centric Sites | Rank Trend | Conservative-Centric Sites | Rank Trend |
| Slate | up | The Gateway Pundit | up |
| Daily Kos | up | Breitbart | up |
| Salon | up | Daily Caller | up |
| FiveThirtyEight | up | Fox News | up |
| CNN | down | Daily Wire | up |
| NYTimes | down | Drudge Report | up |
| Washington Post | down | PJ Media | up |
| Mother Jones | down | Free Beacon | up |
| Think Progress | down | OANN | up |
| Vox | down | Red State | up |
| NBC News | down | Townhall | up |
| CBS News | down | Twitchy | up |
| The Atlantic | down | Hotair | up |
| NPR | down | True Pundit | up |
| New Yorker | down | Newsmax | up |
| MSNBC | down | The Blaze | up |
| The New Republic | down | Zero Hedge | up |
| Jacobin Magazine | down | Western Journal | up |
| The Nation | down | Wall Street Journal | down |
| The Intercept | down | Washington Examiner | down |
Source: Alexa, ZettaCap
There is definitely something unusual going on here.
Sites that have a predominately left-leaning bias have been seeing their rankings decline whereas those with a generally right-leaning bias have seen the opposite.
The results are so obvious and conclusive that it is almost a waste of time to review them. This is not the type of data that is nuanced. It is what it is. Within a handful of months to election-day, conservative sites are seeing traffic increasingly move in their direction.
Implications
Do generally increasing rankings of conservative / Republican-centric sites and generally declining rankings of liberal / Democratic-centric sites imply more rapidly increasing interest on the right as we approach the November election-day? Yes, it implies that Republicans are showing increasing interest while Democrats are not.
Does this, by itself, mean that Trump will win the election? Not necessarily, but it does provide important insight into the actions of the electorate. The fact that this is not widely covered or followed makes this information of greater interest as well.
Do rapidly improving interest levels normally forecast the prevailing candidate? In our experience, yes, rapidly improving interest levels concerning the candidate normally forecast the eventual winner.
In the end, this is just one more datapoint and it should not be blown out of proportion. Just because conservative sites currently have the momentum does not infer a Trump victory. When placed in context, however, we see that site rankings tend to confirm trends in other areas such as social media, Amazon political themed product sales, Google Search Trends, and others. The weight of evidence, based on on-line data, point towards a Trump victory.
Our biggest current issue is trying to square these observations with polls and betting markets which essentially show the opposite – that Biden, and not Trump, has the most interest and support, and will surely win in November.
Our current and unfortunate interpretation is that, due to the contentious political environment in the US, people are perhaps not being as forthcoming in polls and in public. Their apparently private on-line activity does not appear to coincide with public polls. This is extremely unfortunate as it implies we are entering the realm of ‘hidden voters’.
Current data makes 2020 potentially worse than during 2016 when we covered the ‘hidden voter’ phenomenon. We will continue to observe these trends and how they relate to polls, but currently it looks like the American electorate might be holding a few surprises.