Overview
This past week, the Democrats held their 2020 national convention where the party officially declared Joe Biden as its nominee. Just prior to the convention, Biden announced that Kamala Harris would be his running mate, ending months of fairly intense speculation.
Collectively, these two events were major boosts to the ticket, as was expected. However, the level of interest and implied support was not evenly distributed across the country and this is the main point of this post. We found some rather unanticipated areas of weak interest. Though more data will be needed to confirm this trend, initially it looks like more states than are normally listed as swing states by the consensus are in fact teetering.
Further, many of these states are not in risk of switching from Republican to Democrat but from Democrat to Republican hands.
Democratic chances, judging from our data derived from on-line activity, for winning back Pennsylvania look reasonable. And, North Carolina and Georgia appear to be within reach. These observations are well within consensus expectations and therefore are not that impactful.
More relevant, however, are states where our data show obvious weakness for Democrats during this key period between the Harris VP announcement and +3 days after the Democratic National Convention. Multiple states including Rhode Island, New Hampshire, Minnesota, Oregon, and New Mexico showed decidedly little on-line interest or support for the Democratic ticket. The implication is that Republicans have an unexpected opportunity in these potential swing states. These observations are well outside consensus forecasts.
Upcoming key events, not least of which the Republican National Convention this week and the fast approaching Presidential Debates, will help to verify these emerging trends. For now, you should know that the consensus view of a Democratic blowout victory in November is exaggerated.
Kamala Harris as VP
We begin by looking at Biden’s announcement of Kamala Harris as his VP. It was a pivotal moment in the election cycle. Not only had there been so much speculation over Biden’s choice but it was historical as Harris became the first minority Vice Presidential candidate from a major party (note that the Green Party has already nominated various). It should have generated and did generate significant on-line interest following the announcement.
The distribution of interest and support, however, was somewhat unusual. We had expected a major boost of on-line interest and support from predominantly Democratic states and more muted interest from predominantly Republican states, with the potential exception of those states with relatively high African American populations.
The broad strokes followed our expectations with two major exceptions.
The first should be seen as generally positive for Democrats. The on-line reaction was generally strong in the South, much greater than expected. Her outperformance was specifically noted in Alabama, Louisiana, Mississippi, South Carolina, Georgia and North Carolina. This reaction tends to confirm the possibility of Georgia and/or North Carolina being states that could swing from Republican to Democrat in 2020 as many consensus forecasts predict.
The second, viewed as more surprising, is clearly more positive for Republicans. Many states, which in most consensus forecasts are solid Democrat in 2020 and which Hillary Clinton won in 2016 such as Minnesota, Oregon, Rhode Island, New Mexico, Maine, New Hampshire, and Colorado, showed surprisingly little interest or support for Harris. Furthermore, two states, Wisconsin and Arizona, that voted for Trump in 2016 but which many forecasts currently have as leaning Democrat in 2020 also showed relatively low interest for Harris.
The headline figures, when consolidated on a national level, make Biden’s VP pick look strategically strong. However, when looking at it from a state level, Republicans appear to benefit more so than Democrats.
By far the most significant boost for Democrats came in the South. With the exception of Georgia and North Carolina, such a boost does not appear to be enough to swing these states to Democrat in 2020. In contrast, the relative lack of interest in Harris among many states that the Democrats won in 2016 by less than 10 percentage points is a major red flag that seems to be going unnoticed.
Biden at the Convention
Normally, candidates receive a major boost during their party’s national convention. And, as expected, Biden received a surge in interest and support.
However, like the case surrounding Biden’s VP announcement, the on-line reaction varied considerably according to location. And, the general pattern appeared strikingly similar by state and region.
For instance the following states, which voted for Hillary Clinton in 2016, showed not only weak interest in Biden but unusually strong interest for Trump; Minnesota, Oregon, Nevada, New Hampshire, Washington, Rhode Island, Michigan, and New Mexico. Furthermore, Wisconsin, a state won by less than a single percentage point by Trump in 2016, showed a similar pattern.
The fact that this occurred during the week of the Democratic National Convention is very significant. This was the showcase week for Biden and the Democratic Party in general. Interest in Biden and Harris was supposed to be through the roof during the week. The fact that Biden generated minor interest in these states should again be a major red flag but has been mostly ignored. Or, perhaps better stated is that most political analysts and forecasters simply did not observe this trend as they do not, to our knowledge, spend any significant time analyzing or even incorporating on-line data. (Re-read that last sentence while pondering how important search, social media, news consumption, and other on-line activity is to our modern society in 2020.)
On the positive side for Democrats, Biden generated relatively strong interest in such key states as Pennsylvania, Colorado, and North Carolina. (Of course these were not the only states where support for Biden was evident. For states such as Maryland, Delaware, and District of Columbia, Biden looked extremely strong, which is really not surprising but highlighting the obvious is not our goal here.)
Again, we saw the same basic pattern emerge during the week of the convention as we saw around Biden’s VP announcement. There was surprisingly strong interest for the Democratic ticket in a handful of potential swing states while there was weaker than expected interest in many more potential swing states, making the Democrat’s efforts in the past week look relatively unproductive.
Biden Post-Convention
In order to confirm these trends, we then took two separate periods after the convention – the day after (August 21st) the last day of the convention and the Sunday (August 23rd) between the week of the Democratic National Convention and Republican National Convention. The basic idea is that the Biden / Harris / Democrats would still be top of mind and still not be overshadowed by the Republican National Convention starting on August 24th.
The general results clearly reinforce those from the previously mentioned periods, as we will see as outlined in Table 1.
State Summary
On a state basis, the on-line reaction to the Democratic ticket in general during the important period between the VP announcement and just after the Democratic National Convention was mixed. Democrats tended to positively surprise in the South while clearly underperforming in multiple other regions.
Table 1: Implied Interest / Support for the Biden / Harris Ticket during Key Periods in August 2020, highlighting states that showed unusually high or low on-line metrics for the ticket
| Interest / Support | Region | Harris as VP | Week of Convention | +1 Day | +3 Day |
| Unusually High | East | Pennsylvania | Pennsylvania | Pennsylvania | |
| South | Virginia | Virginia | Virginia | Virginia | |
| Georgia | Georgia | Georgia | |||
| North Carolina | North Carolina | North Carolina | |||
| Tennessee | Tennessee | Tennessee | |||
| Unusually Low | East | Rhode Island | Rhode Island | Rhode Island | Rhode Island |
| New Hampshire | New Hampshire | New Hampshire | |||
| Midwest | Minnesota | Minnesota | Minnesota | Minnesota | |
| Iowa | Iowa | Iowa | Iowa | ||
| Wisconsin | Wisconsin | Wisconsin | |||
| West | Oregon | Oregon | Oregon | Oregon | |
| New Mexico | New Mexico | New Mexico | New Mexico | ||
| Nevada | Nevada | Nevada |
Source: ZettaCap
Note: In order to appear in the table, the state had to be highlighted in at least three of the four time periods as having unusually high or low on-line interest or support.
The strong point for the Democrats was clearly the south of the country.
Though we still want to see more data before coming to conclusions, it looks like Democrats will most certainly maintain Virginia given the unusually high level of on-line support. Also, Georgia and North Carolina showed unusually strong activity as well which seems to support the consensus view that they could flip Democrat in 2020 from Republican in 2016. Tennessee was admittedly a surprise and we feel more data is necessary prior to commenting further on it.
Pennsylvania was expected to generate significant interest in favor of Biden which was confirmed by our data. Joe Biden was born in PA and his wife, Jill Biden, grew up there. The fact that on-line activity favored Biden and Democrats was to be expected and does tend to confirm the possibility of PA flipping in 2020.
More than offsetting such pro-Democrat observations was the weakness shown by Democrats and relative strength of Republicans in extremely unexpected places in portions of the West, the Midwest and East.
Many of the states highlighted in Table 1 should be seen as potential swing states regardless of trends of on-line data. Any state where the margin of victory was single digits in the previous election should be considered a potential swing state. Of the states that are listed in Table 1 as having had unusually weak interest in the Democratic ticket (and relatively strong interest in Republicans) that fit this category are New Hampshire, Minnesota, Iowa, Wisconsin, New Mexico, and Nevada. The other states listed included Oregon (Democrats won with an 11 percentage point margin in 2016) and Rhode Island (16 percentage point margin) and seem to be more of a stretch for Republicans, though not inconceivable.
In the West, Nevada, New Mexico, and Oregon stood out as potential Republican surprises. Nevada, which was won by Democrats in 2016 by only 2 percentage points, is clearly a potential swing state, though rarely mentioned by consensus forecasters as such. Not highlighted in Table 1, other regional states such as Arizona and Colorado showed similar though not as obvious trends, which tend to back-up our observations for Nevada and New Mexico.
Oregon seems to be somewhat of a unique situation in that there have been consistent protests, which have often turned violent, in the state during the previous months. The fact that our on-line data shows weakness for Democrats and strength for Republicans should be a major red flag for Democrats. We have also noticed similar yet slightly more diluted trends for Washington, which has had its own share of protests and disruptions. The fact that these neighboring states depict similar trends tends to confirm our observations.
It seems as though the general stretch from the Southwest (Arizona and New Mexico) up through the Mountain region (such as Colorado, Utah and Nevada) and ending in the Pacific Northwest (Oregon and Washington) are showing unusually low interest/support for Democrats and unusually strong for Republicans. As mentioned, these trends vary considerably by state, but the general underlying trend is the same, is surprising, and is going completely undetected by others. There is no clean reference name for this area other than the western most portion of the nation minus California.
Again, we hesitate to make too broad of conclusions while there are so many important events coming up that will offer further data-based insights. If our initial observations are correct, however, we could be witnessing a seismic shift in the political landscape of the country.
The Midwest, which was likely the source of the largest 2016 electoral surprise, also follows a similar pattern of having various neighboring states reflecting a similar trend. In this case, it is Wisconsin, Iowa and Minnesota. Of these three, Minnesota was the only one to go for Hillary Clinton in 2016, but with a paper thin margin of 1.5 percentage points. As such, it will not take much for this state to flip and the current trends of on-line data as well as regional trends support that theory. Not included in Table 1, it is interesting to note that Michigan also showed a similar but more diluted trend, which helps to support the regional trend as well.
Lastly, we saw unusual activity in the portions of New England. Specifically, New Hampshire, which Clinton won by less than 0.4% in 2016, showed weakness for Biden. Our initial impression is that NH could very well flip to Republican in 2020. A more aggressive call surrounds Rhode Island which is normally a deep blue state. However, our data clearly show it to be at least flirting with the idea of voting for Trump in 2020. Maine, to a lesser extent, depicted a similar trend, which helps to confirm the regional move.
Regional Trends
Though already mentioned, it is worth repeating — the data shows regional trends and not just state trends. These do not appear random or spotty. Regional trends hold more weight and tend to self-confirm.
The Democrats clearly outperformed in the South while underperforming in the northern portion of New England, the Midwest and West minus California.
On a national level, the figures do not look so terrible for Democrats as the large population centers appear to be holding firm. States such as California, New York and Illinois continue to support Biden / Harris / Democrats.
Recall that the election is based on the Electoral College, so state level analysis is paramount. And, our state level analysis of on-line data leans towards surprising Republican support in unexpected states. The fact that many of these state trends are confirmed by similar overarching regional trends helps to confirm them.
Why Not Disregard these Observations
Let’s face it, these observations are not mainstream. And, almost every political forecaster currently has potential swing states going Democrat in 2020. Few if any even mention the possibility of a swing state flipping from Democrat to Republican in 2020. Why on earth would you even consider any of these out-of-consensus forecasts?
Our service was the only one to correctly call the 2016 US Presidential Election. In fact, we are the only known data-based forecast to have stated that Trump would win Michigan, Ohio and Pennsylvania. This is not to say that our current observations are infallible. However, if our data and systems were able to pick up underlying trends and make such bold calls in 2016, it seems to follow that the same data and systems should be heeded in 2020.