Keeping tabs on the ever changing environment

The amount of information and content that the average consumer receives online and on our phones today dwarfs previous generations—and you’d be hard pressed to find any marketing expert that anticipates this trend abating.

In fact, it was estimated that about $107 billion was spent in America alone in 2018 on digital advertising. For campaigns last cycle, that amount is estimated to be between $900 million and $1.8 billion. This is clearly a big range. But it’s a tough number to pin down based on the way expenditures are reported. For comparison, in 2014 digital spending was estimated at about $250 million.

The challenge in the political world is how to measure what’s being done and the effectiveness of that effort. Digital consultants have their metrics such as impressions, clicks, et cetera. But are there other ways to understand the reach and effects of the money being spent?

While campaigns have been a little slow to make adjustments to digital, the same can be said of the political research world. Traditional survey research isn’t set up to measure small changes. Back in the days of three major networks and 1,000 points on these stations being a near guarantee of reach around the country or state, a survey before and after the run of the TV ad would suffice. But digital advertising works much differently, and research needs to adjust accordingly.

Understanding how humans react to information in our digital age is an ongoing effort across the globe, which admittedly is having a tough time keeping up with the changes in usage and behavior. That being said, this is an incredibly exciting time in the areas of research and measurement. It’s allowing us to increase our understanding of what makes people tick, and more importantly what makes people change their minds after being exposed to ads.

Trendency fortunately allows us to track these events, and when looking at the most recent election, one item was able to break through and create an impact on voter enthusiasm and electoral choice: the Judge Kavanaugh hearings.

As you can see below the chances of voting did not change much from mid-July through August. We saw some movement right before and after Labor Day. Then, the hearings for now-Justice Kavanaugh began.


One of the post-election questions we asked our panelists was which party they voted for in the race for House of Representatives in their district. Because the Trendency data is continuous and these same panelists have answered questions throughout the cycle, we can append this new data to old questions. This allows us to analyze just when opinions shifted and potentially what the cause of that shift might have been.

When we look back at this generic Congressional ballot test, it becomes clear that the Kavanaugh hearings had a large impact on voters, but one that differed by party and gender.

The graph below shows the percentage of voters who offered very strong support for a particular party, broken down by different time periods around the Kavanaugh hearings. For example, before the hearings began in early September, 76 percent of women who ended up voting for a Democrat gave strong support for Democrats. At the same time, only 64 percent of women who voted Republican in November gave strong support to the Republican Party candidate.


Before Kavanaugh, Republican candidates had a real problem with potential female voters. But, as the hearings progressed and became more contentious, that level of strong support increased (as it did with Republican men). By the time Kavanaugh was approved by the Senate on October 5, the percent of Republican women with strong support was up 8 percentage points.

Strong support from voters on the Democratic side showed much more volatility. The initial hearings seemed to galvanize them, but the time period between Dr. Ford’s article in the Washington Post and subsequent hearings (from September 16th to October 5th) coincided with a drop in strong support levels. It certainly appears that the hearings animated eventual Republican voters while it had mixed results at best for eventual Democratic voters.

If you looked at the data from early in the fall, there was some clear softness in the minds of voters, especially with Republican women. All that changed by Election Day, when partisanship reigned supreme again. There were other factors at play beyond Kavanaugh, including the caravan of migrants as well as a general pattern of people returning to their respective partisan bases as November approaches, but the Kavanaugh hearings broke through the noise and shifted the vote choice in real and meaningful ways.

Enthusiasm Takes a Hit

Changing levels of candidate support does not tell the whole story of the Kavanaugh hearings. The whole episode may have increased partisanship, but it also temporarily dampened voter enthusiasm to an alarming degree, unprecedented in past Trendency data.

Throughout the election cycle, Trendency measured respondents’ likelihood of voting, asking them to rate their chance of voting on a scale from 0-100. Obviously, this is a self-reported measure and therefore may not reflect true voter turnout but tracking this measure over time can demonstrate shifting levels of voter enthusiasm.


The trendline above tracks two groups from July through Election Day: those who reported voting for Democratic candidates post-election and those who voted for Republicans. The vote likelihood levels were steady from July through late August. Once the Kavanaugh hearings began, volatility increased dramatically. For eventual Democratic voters, their enthusiasm initially went up and then dropped quickly through the final confirmation vote in early October. Future Republican voters essentially decreased their enthusiasm to vote throughout, including an especially precipitous decrease during what appears to be around the time Senator Jeff Flake led to a week-long delay in the vote.

Once the vote was finished and Brett Kavanaugh became Justice Kavanaugh, Republican voters immediately reclaimed their enthusiasm, eventually returning to a similar level that they reported in the summer. It took a little longer for Democrats to find their footing, but a late surge took their chance of voting a couple points higher on Election Day than it had been throughout the summer and fall.


The common public perception during and after the Kavanaugh hearings was one of galvanization for conservative voters, renewing in their minds the importance of midterm elections. In one sense, the Trendency data appears to back up this claim; support for their candidates among Republicans (especially women) was notably soft before the hearings began, but support for Republican candidates solidified as the proceedings continued. Counterfactuals are impossible in these situations, to know what might have occurred had Kavanaugh withdrawn or without a Supreme Court fight at all. These softer supporters of Republican candidates may have inevitably moved to the right, but the hearings clearly sped up this transition.

While Trendency confirms this preconceived notion, it also provides a previously unseen element of voter disgust with the whole process, one that affected supporters of both parties. A dampening of voter enthusiasm among Democrats might be expected, but a similar drop for Republicans, just as they are increasing the strength of their support, is quite interesting.

Simply due to the timing of the hearings, this drop in voter likelihood had time to regain lost ground, but this data shows just how much impact a large-scale political event right before an election can have on voter choice and voter enthusiasm – something to keep in mind during what is sure to be another chaotic presidential campaign in 2020.

Brendan Gleason