November is just around the corner and campaigns are gearing up for the ad war, with an ever-increasing focus on the digital space. As campaigns start to shift from a fundraising focus to a strategy focused more on persuasion and finally to get-out-the-vote, we should start to see a larger volume of political ads filling up our digital screens. Last month, the Trump campaign’s Google Ad Spend averaged out to about $670,000 a day. This is almost on par with the Trump campaign’s daily spend in the November 2018 midterm elections. Trump’s team didn’t stop there, they also have revved up their Facebook spending in the last month, averaging about $629,000 per week according to Facebook’s ad library. For three days in May, the Trump campaign spent over $100,000 to promote the bad faith message that Joe Biden is for mass incarceration & against Black lives. This ad ran in seven battleground states with a heavy focus in Florida, targeted mostly towards women & folks over the age of 45, which makes it an early persuasion ad to a swing demographic in a swing state. To no surprise, Trump has been outspending Biden on digital across Facebook and Google platforms. The Trump campaign has spent $52 million combined on tracked digital, mostly on fundraising efforts, while Biden stands at about $23 million so far.

Let’s put this into context: according to a political spending projections report produced by Cross Screen Media & Advertising Analytics, spending of the two presidential campaigns is to account for the largest segment of 2020 spending at $2.7B. $800M (29%) of this total will likely be spent on digital video. About two-thirds of this spend will come from President Trump’s campaign alone. For congressional campaigns, the projected total is $1.0B in 2020, with $242M in spending on digital video. Spending on Senate campaigns is expected to be down 14% to $789M in 2020, but with a record of $176M projected to go towards digital video online. Across all parties, $1.6B is the estimated digital video spend, much of which will be on Facebook and Google.

Republicans are more flexible about their type of content and modes of reaching people. In the past, Democrats have promoted video ads mainly for donation asks and fundraising, I’m pretty sure we have all seen those direct-to-camera iPhone videos every candidate produces, asking voters for ‘$1 before midnight’. The Republican playbook however tends to use videos extensively to promote awareness over their controversial hot button issues, trying to rile up their base. You also tend to see Democratic videos sticking manly to the same platforms -- Facebook, Youtube/Google with some CTV thrown into the mix, while Republicans have been better at using video in a more innovative way, targeting people based on their behavioral profiles and online usage, and retargeting segments across not just Facebook & Google/YouTube, but multiple platforms such as programmatic, CTV, and PMP deals. This allows Republicans to target people on their preferred media outlet, on issues they care about, with messages Republicans want to see amplified online.

A way to ensure that Democrats start to think more about online-video-first campaign strategies: Think pre-roll inventory first and TV inventory second. Further, campaigns must always think of the user experience in the ad, especially when it comes to digital. In DMA-level media buys, the message is shared across multiple audiences promoting a more general messaging, while with digital, audiences need to be addressed on the issues that the viewer values, or else you are just wasting impressions and views.

With digital spending climbing up, targeting becomes the key topic of discussion. Again, Republicans are leading the charge, and many conservative campaigns are preparing for larger addressable media campaigns, starting with an “audience-first approach” to persuasion advertising on TV and digital media: Instead of buying TV ads with a separate agency, these campaigns are now identifying 10,0000 to 20,000 key potential voters or supporters on a certain issue within a given district. They then create video messages targeting that particular segment. It’s a tactic reminiscent of platforms such as Google and Facebook, but applied to OTT platforms and linear TV.

If there is one thing to take away from all this: A key advertising tool that every campaign will need for the 2020 elections is digital video, and not just on Facebook & Google, but CTV, mobile video, and programmatic. Do your research where people are spending their time and how they consume media (such as our own research on LatinX media consumption). Curate your audiences, using social media and programmatic video inventory to target specific people based on message and behavior to create a truly “audience-first approach”.Organize your media plans such that video is the primary medium for your persuasion campaign. Think about investing the money you raise on digital to continue digital outreach to your audiences. The truth is: If you are not thinking of digital-video-first and audience first, you are already behind!

Intrigued by all these facts but don’t know where to start? PredictWise is launching a new product this August that will cater to all your campaign needs, no matter your goal. Signal by PredictWise allows you to take control of your audiences and create fully customized audience segments leveraging real people-based data, at the intersection of attitudes and behaviors, of hundreds of millions of Americans. Stay up to date with all our newest products and services by subscribing to our newsletter at http://predictwise.com/.

You might have heard the word mobile Ad IDs floating around a lot in digital advertising spaces recently.  And, as our readers are certainly aware of by now, PredictWise audiences are keyed to MAIDs, allowing us to achieve better match rates. In fact, PredictWise is tracking 250 political, psychographic, electoral, and economic metrics tied to more than 250 MM MAIDs, updating weekly. Mobile Ad IDs have been called by many names: device IDs, MAIDs, unique device IDs, and mobile identifiers, to name a few, but what is it? Simply put MAIDs are identifiers helping advertisers understand whether a user of a particular phone has taken an action like a click or an app install. Below are a few simple tips to help you understand them better:

What is a mobile ad ID?

MAIDs (Mobile Advertisement Identifiers) are a sequence of random numbers, letters, and symbols that are shared with various app servers the user has on their phone to track their user’s journey and “remember” their choices. The two most important and widely used MAIDs are attributed to phones at the software level by Apple (IDFA) & Android (AAID).

Does everyone have a MAID?

Most popular phone manufacturers, phones running on Apple & Android, for example, have MAIDs in their operating systems.

Does the MAID tied to a user last forever?

MAIDs do get refreshed, either by the phone manufacturers operating system, by the user manually, or with the replacement of a new device. A user can also manually reset one’s device ID on their phone, whenever they want and also see what applications are currently using their device ID. This is why it is important for advertisers to not continually use the same audience ID list & keeping your lists updated, although the average life-span of the MAID is about 630 times longer than the average cookie.

Why are MAIDs important?

As I mentioned earlier, MAIDs help identify and track mobile users. This helps advertisers to ingest certain user behavior, which can be aggregated to identify trends. Tracking MAID data allows advertisers to better understand how users interact with their ads or app on a unique device level, helping campaigns find out who might be exposed to a particular ad, who clicked it, and who installed the app, allowing for a better understanding of conversion and ROI. MAIDs help campaign managers serve their ads to a more precise audience network, creating an increased custom ad environment for the users and improving the advertising experience for every unique user overall.

How Does PredictWise use MAIDs to help our clients run better ad campaigns?

PredictWise provides completely tailored MAID-level audiences to supply your campaign with a maximum targeting advantage. PredictWise custom targeting solutions are built on ID Resolution Tech, finding people where they are, regardless of whether that is mobile, online, or home, meaning we hit our targets with the highest likelihood possible.

The Future of MAIDS

With the digital ad industry keeping aware of the latest updates with the downfall of cookies and their future availability, most industry professionals are focusing their attention on MAIDs. As of now MAIDs are replacing the third party cookie as a more reliable user ad tracker and a form of user measurement, as the space shifts we might see things shift more toward mobile fingerprinting, devive fingerprinting or SKAdNetwork frameworks but for now, it looks like MAIDs are here to stay.

Extra Reading:

https://support.google.com/admanager/answer/6274238?hl=en

https://www.facebook.com/business/help/570474483033581

https://adint.cs.washington.edu/ADINT.pdf

https://www.iab.com/wp-content/uploads/2017/06/Mobile-Identity-Guide-for-Marketers-Overview.pdf

https://www.cs.cornell.edu/~shmat/shmat_ndss16.pdf

https://arxiv.org/pdf/1903.09916.pdf

https://marketing.wharton.upenn.edu/wp-content/uploads/2018/03/03-22-2018-Yoganarasimhan-Hema-PAPER-TargetinhPrivacy_2018.pdf

1. Connected TV & Digital Video Content
2. The Rise of TikTok
3. Mobile Ad Spend
   
   

Connected TV & Digital Video Content

What is connected TV? Simply put connected TV is digital TV content that is connected to the internet rather than a cable box, satellite, or cable chord. Content for connected TV is being streamed into an internet-connected app on a smart TV or another similar device. CTV (connected TV) includes Smart TV's, Apple TV's, devices like Tivo and Roku, gaming consoles like X-Boxes and PlayStations to name a few. In the past, ad-buying for CTV mostly involved private marketplace deals with providers, publishers, and advertisers. Now more and more we are seeing connected TV inventory becoming available on DSPs programmatically to buy, which makes buying CTV spots more accessible & cheaper for most buyers.

CTV can also arguably make the ad watching process more enjoyable compared to traditional TV, as spots are targeted based on individual-level audience data as opposed to being broadcast. That level of detailed targeting plus the added precision of measurement of being able to count who has seen the ad, on what device, and how many times is something you can never get on traditional TV. An example of this trend ticking up can be seen with The Trade Desk, as consumer reports document that COVID is responsible for the worst period in advertising history, programmatic platform The Trade Desk is trading at an all-time high, especially when it comes to CTV spending.

The Rise of TikTok

TikTok is a video sharing social media platform that allows users to create short videos & sound bites. The video-sharing platform is owned by ByteDance, a Chinese based company. Last month, the app reported two billion downloads globally, 130 million of those downloads being in the United States. TikTok is the first app that isn’t part of the Facebook social network (Facebook, WhatsApp & Instagram) to surpass 2 billion downloads since 2014. Here are some quick facts about the demographics & usage of TikTok:

• 41% of TikTok users are aged between 16 and 24
• 56% of TikTok users are male and 44% are female
• Roughly 50% of TikTok’s global audience is under the age of 34 with 26% between 18 and 24
• TikTok users spend an average of 52 minutes per day on the app
• 90% of TikTok users visit the app more than once per day
• It’s currently the third most downloaded non-gaming app of the year
• 52% of TikTok users are iPhone users.
• TikTok’s average engagement rate is 29%
  

While TikTok does not allow political ads, 70% of TikTok users in the US are of voting age, so campaigns should still be reaching out to the influencers in the TikTok community for organic content/earned media. TikTok videos have the ability to carry hashtags & political hashtags are definitely being used. The Wall Street Journal discovered that over the last three weeks of 2019, videos that carried a Trump2020 hashtag were viewed over 200 million times. As of now only 4% of marketers use TikTok, but with the rate that TikTok is growing in popularity among the Gen Z population & younger millennials I think it’s safe to assume that more and more brands will start to use TikTok to reach audiences.

Mobile Ad Spend
Due to COVID-19, ad spend is down across the board, however, mobile advertising is still evolving. Before coronavirus, marketers expected US mobile ad spend to grow 20.7% in 2020, accounting for more than two-thirds of digital ad spend. About 60% of mobile ad spend is expected to go to Google & Facebook ad networks, with Facebook accounting for 1 in 3 of mobile ad spend. Mobile ad spend is reported to account for 50% of digital ad spending by 2021 and finally surpass desktop in 2022. Since COVID, apps have seen an increase in ad spend to news apps, parenting apps & lifestyle food and drink apps. Marketers have needed to change advertising tactics quickly with the impacts of COVID-19 on advertising, more than a third of advertisers are adjusting their in-market tactics, and are increasing their use of custom audience targeting, OTT/CTV device targeting, mobile/tablet device targeting and programmatic buying. Brands and marketers are shifting their plan to mission-based marketing & cause-related marketing to the right audience, to ensure their ad dollars aren’t going to waste especially right now when every dollar is so closely being looked at.

People are donating less right now, with good reason. They don’t know where the economy will be in a few weeks or might have lost their main source of income. So every dollar you have on hand as a campaign needs to be spent in the most efficient way. Whether you are focusing on fundraising, GOTV, or persuasion, digital should be your number one spending priority going forward (truthfully it always should have been). Let’s start with mail, right now mail is behind not only figuratively - it is literally behind: People are expecting a delay in post for anything that is considered “nonessential”. Even before COVID, there was always issues with people getting mailers at the wrong time (i.e after they voted), getting them delivered to the wrong address, or not ever getting them at all. There is also no way to fully measure the RoI of mail -- aside from carefully designed experiments. How do you know if someone received it? If they did receive it, how do you know it made it past the front door and not straight into the recycling box, not to mention that mailers aren’t very environmentally friendly.

Moving on to TV. When I say TV, I mean linear (read: traditional) TV. More and more Americans are becoming “cord-cutters”, this refers to individuals who are canceling their subscriptions to multichannel television services available over cable or satellite. A MoffettNathanson report stated that the first quarter of 2019 was the largest ever for cord-cutting, with an annual drop rate of traditional TV of 3.8%. Traditional TV is also one of those mediums where you can’t ensure anyone really saw the ad, you can’t personalize it & you can’t show your ad to a custom audience, yet linear TV ads are one of the most expensive forms of advertising that still has a large grip on political advertising. TV ads are great if you have one blanket message you want to get to everyone, and all you really care about is exposure and brand awareness.

Facebook Breakdown

Every campaign should think of digital as the most effective way to reach their constituents, now more than ever. With the use of custom audiences, campaigns should be able to target individuals on most platforms & personally cater the messaging per audience. This allows a campaign to fully utilize digital in many different forms, such as community organizing, digital canvassing, persuasion, awareness, acquisitions, and GOTV.

Yet, Facebook spend comes with pitfalls that could turn your digital strategy quickly into a marketing disaster. We break this down below:

Facebook Do’s and Don’ts

• Don’t upload your email list to Facebook and call it a day. And most importantly: Don’t create a lookalike audience of that list. After all those list swaps, it feels like all campaigns are basically using the same lists of donors over and over again. The 1% lookalike audience has been viewed as something of a “go-to workaround” (as opposed to custom lists) for many campaigns, which has led to campaigns unfortunately assuming these audiences will automatically perform better than other audiences. But, remember this lookalike is built on the same or at least a very similar seed audience. In effect, your 1% lookalike are the same folks, and this means over-crowding: When campaigns are fighting for the same audiences, ads become less relevant, have a lower impression rate, and the cost per impression (CPM) rises. Using Predictwise audience technology, built on proprietary data, allows campaigns to keep their audience uber targeted while still maintaining their reach & casting a wide net on Facebook. PredictWise audiences integrate into Facebook with a market-leading match rate.
• Place ads on Facebook, but don’t dedicate your whole ad budget to Facebook. Facebook is great when it comes to acquisition (though we have seen Cost Per Acquisition rise recently, but that is for another day), awareness, and information/education campaigns. Facebook still continues to be one of the top 3 widely used platforms, with adult users age 65 and up being the largest demographic group on Facebook. Instagram, however, has a much younger audience with 75% being 18–24-year-olds & 57% 25–30-year-olds. Just as you wouldn’t fixate on one political message for an entire campaign, why would you only spend your ad buy on one platform?
• Don’t combine your Facebook and Instagram buys unless you plan on spreading just one big universal message. The demographics for Facebook & Instagram are different, as mentioned above, meaning that the messaging, creative and overall goal should be different. People don’t interact with ads on Instagram the same way they do on Facebook, so why would you target them the same way? Also thinking of them as two separate buys will help with reporting & analysis when digging into which methods worked & which did.
• Do test multiple creatives, messages, targeting & ad formats when available, especially when targeting diverse communities of color. Latinx folks, for example, are expected to make up the largest nonwhite ethnic voting bloc in 2020. A study we conducted together with Equis Labs showed a glimpse of exactly how diverse the Latinx community is. With a community this large & this diverse, you wouldn’t just target one Latinx audience with one message and call it a day. Instead, you are better off creating multiple custom audiences based on age, language preferences, ethnicity, and geographic location. To summarize, the strategy you should use to target a Cuban American in Florida should not be the same one you use for a second-generation Mexican American in California and so forth.  
  
  
  

Have questions about any of the strategies or methodologies mentioned above? If so please contact  feedback@predictwise.com & someone from our team will reach out to you!

A lot has been written about impeachment of President Trump, and the shift in public opinion. For example, the 538 tracker has about a 4 ppt. advantage for support, a massive shift from late September when Don’t support was up by almost 10 percentage points. And, our polling is no outlier in this array of increasingly worry-some (for Trump!), or increasingly encouraging (for defenders of democracy!) recent polls: In a recent poll throughout late October (our polling relies on Random Device Engagement – randomly targeting cell-phone identifiers with in-app surveys – and advanced analytics, more here), we have Impeach up by 10 percentage points, just a little bit higher than a recent YouGov poll. 

Can we expect more change?

As always, our polls go further than other top-line impeachment polls. Our mission was to understand deep-rooted values and sentiment that might be predictive of further swings down the line. First up: perceptions. Pluralities of likely 2020 voters believe that the whistle blower who originally reported on the originally reported on Trump’s phone call with the Ukrainian president Zelensky  is a patriot (41% vs. 26%), but, only 17% of Republicans believe so. But, a strong majority of all likely voters (59%), and a strong plurality of Republican likely voters (34%) believe it to be outside the norm for a US president to ask foreign countries to investigate their political opponents.

And, knowledge around key events are quite substantial – a clearly worrying sign for Republicans and Trump. Massive majorities of all voters (84%) and Republicans (81%) know that the US president is subject to US law. Likewise, majorities of all likely voters (69%) as well as Republicans (57%) know that Trump asked Ukrainian president Zelensky to investigate the Bidens.

In total, these signs are quite worrying for Republicans. Clearly, Americans are tuned in and quite knowledgeable around the impeachment inquiry, and public hearings could be a real catalist in translating this engagement into further changes in public opinion.

Do not mistake the forest for the trees!

But, it is also important to remind people of the fact that ultimately, the majority of likely voters is not driven by Impeachment. While 17% of all likely 2020 voters, and more than a quarter of Democrats, indicate that impeachment of Trump is their number one concern for 2020, this number is dwarfed by healthcare, which was the central issue for voters in 2018, and is shaping up to be the central issue once again in 2020.

Could not help myself, so my future reference I made a quick list of how Nate Silver is a really great data journalist and publisher, but a really bad pundit and not a progressive. My best guess is that he a moderate, possibly right-of-center on policy, who is who just has a really bad understanding of politics? Or, he is a right-winger who spent years cultivating the trust of progressives, to use that power against them? Or, he is just a publisher trying to make right-wingers feel good about consuming his product? Some examples of very unfortunate public positions:

1) Nate Fought Against Boycotting White Supremacists on TV: Despite Nate’s advice, these boycotts proved quite effective in making White Supremacists less profitable and de-legitimizing their platform (Fox News!) and positions. But, easy to see how boycotts over political positions would spook the publisher of a politically oriented company. (relevant tweet).

2) Nate Fights for More Horse-Race Coverage: He continuously argues for more horse-race oriented coverage of elections. This would be great for his business, but terrible for democracy and Democrats. Normatively we want an informed electorate, and Democrats benefit from debates of substance around policy and suffer in a horse-race coverage that implicitly creates false legitimacy for Republicans.

3) Nate is against Impeachment Hearings (early): He argued against Democrats holding President Trump accountable for crimes, for having public decide in elections. This would be terrible for our Constitution, but also bad advice for Democrats who want to investigate and expose Republican crimes against US. (relevant tweet).

4) Nate is Pro-Howard Schultz: He argued for Howard Schultz’ independent run for president, claiming he may actually help Democrats which is just absurd. (relevant tweet).

5) Nate is Pro-Filibuster: He argued that Democrats should disingenuously support the filibuster than destroy it in power. (relevant tweet).

6) Nate is Pro-Liars on TV: He argued for putting President Trump, and other known liars, on TV and fact-checking them. This does not work. (relevant tweet).

7) Nate is against Impeachment Hearings (late): He argued that the Democrats should hold on impeachment even after Ukraine scandal broke, because it would hurt them. He was proven wrong within minutes. (relevant tweet).

8) Nate is wants everyone to praise Trump for killing ISIS leader: Raid was successful despite President Trump. (relevant tweet).

9) Nate is all in for Joe Biden: He makes srawman attacks on anyone who discounts the possibility that Biden will be the nominee. He may be the nominee, but he is not the front-runner. Biden’s issue is not policy, but age, competency, etc. (relevant tweet).

A little bit of personal background: Two very different feelings dominated on the morning after the 2016 election: anger/shame/frustration/sadness regarding the State of the Union (Not Strong!), and a sense of validation, namely that large-scale continuous data collection via disparate modes, paired with the right analytics, can yield actionable insights able to help us dissect the American mindscape. We were motivated to share these methods with the progressive ecosystem, but first we needed to understand it. What we learned about the progressive ecosystem in the wake of ’16 did not make us sleep better.

The Progressive Ecosystem post 2016 – non-continuous, monopolistic, non-collaborative

Progressive investment, both financial and soft, is inefficiently geared toward elections – period. Thus, the days after any election almost all campaign-related data is lost: data repositories insufficiently transfer from campaign to campaign and turnover rates at central Democratic organizations make it hard to develop sustainable infrastructure and institutional knowledge. Then, data collection, data analytics, content creation, and distribution come to a halt for extended periods of time, guaranteeing a massive hole in our: understanding of voters, development of infrastructure and content, and communication and messaging for vast periods of time between elections. Any inefficiencies in progressive methodology, conditional on a campaign being operational, are dwarfed by these dormant periods, but add to a baseline of negative spill-overs: In campaign times, data and analytics are created for specific clients, without regard for the general progressive cause, oftentimes focus on rival primary candidates, and is hardly shared between camps. All of this is exacerbated further by an ecosystem divided into small monopolies or duopolies, stymieing both competition and cooperation.

Conversely, Republicans have pushed toward continuous, hierarchical/vertical and cohesive messaging campaigns for decades, run by Koch, Mercer, Murdoch, and the RNC. In light of the non-existing Democratic defense – virtually no positive progressive content has been created and distributed on a continuous basis –  this onslaught has led to predictable results. The outrage over the hideously framed estate tax is now an almost anachronistic example. In this day and age, Republican messaging campaigns, with their focus on misinformation, register everywhere, even among Independents and Democrats. Two prominent examples, from the PredictWise stock: On immigration, only 26% of Democrats know that fewer immigrants have come to the US since 2009. On healthcare, only 45% are aware that the percentage of Americans without health insurance has increased under Trump.

The problem does not stop there. On many issues. Republicans have convinced a sizable chunk of Americans that a Republican administration serves their interests better, while in fact their policy preferences are much more in line with Democrats. Example:  81% of Americans support Medicare Buy-in, 57% of Americans disapprove of a healthcare market allowing insurance polices NOT covering pre-existing positions. But, only 41% prefer the Democratic healthcare plan, as opposed to 39% preferring the Republican plan.

Toward a Permanent Campaign: 20-For-20

To combat these dynamics, our PredictWise 20-For-20 vision centers around a permanent campaign, focusing on positive progressive messaging.

Data Collection/Analytics: Continuous surveys with ad-hoc additions combined with behavioral data should be used to develop key attitudinal insights and value frames, and should be paired with continuous monitoring of media agenda setting and exposure. Data should be homogenized and aggregated centrally and linked individually where appropriate.

1. CONTINUOUS collection of baseline data. This data could include aspects of public opinion, fact knowledge, concern intensity, and psychographics. As the repository grows, this data will get more precise over time, while being able to shed light on important dynamics. For now, this kind of data has to be projected onto voter files via Machine Learning models, but the goal is to reduce the modeling component more and more as new data flows in, with the ultimate goal of curating a ground-truth attitudinal layer on top of the voter file that is not (or only very lightly modeled). Currently, PredictWise is stocking data of 300,000 unique respondents on more than 200 economic/psychographic/political attitudes.

2. AD-HOC data collection. This data should be used to rapidly inform political elites, media, and mass about public opinion on emerging issues. Example: A plurality of Americans believe that the Trump administration encourages non-democratic regimes to clamp down on dissidents and the free press, per PredictWise data.

3. COMBINATION of behavioral, survey and other data. We routinely build our models on top of a mix of behavioral and survey data. For example, ambient cell phone data can help us achieve scale; behavioral data can help account for measurement error/social desirability bias inherent in attitudinal data, for example when it comes to media consumption etc..

4. MEDIA AGENDA Tracking. We need to track agendas of mainstream media: (which still produces the bulk of news that people consume) (a) broadcast/cable media, (b) online as well, in order to react to emerging discourses. This requires independent scraping efforts, or deals and with the Internet Archive, NewsBank, or other aggregators of transcript-level data.

5. MEDIA EXPOSURE Tracking. We need to track individual-level exposure to (a) broadcast/cable media, (b) online as well, with the ultimate goal to create single-dimensional, dynamic exposure profiles combining exposure patterns and patterns of exposure content.This requires new partnerships with Nielsen, Commscore/Rentrak, etc.

6. Data collection HOMOGENIZED as much as possible across modes to create positive spill-over effects.For example, matching question formats of canvasing data with question formats in ongoing surveys allows various organizations to conduct more granular analyses, with more statistical power.

7. ID LINKAGE. In a world in which our content dissemination strategies move to the digital realm, we need to invest in ID linkage technology allowing us to reach who we want to reach with highest possible accuracy. For example, match rates of voter-file-based PII into Facebook, other Demand-Side-Platforms or Addressable TV are dismal, and can be significantly increased if we move to mobile-first identifiers such as MAIDs.

8. CENTRALIZED ANALYTICS layer. As opposed to sharing top-line data across the ecosystem, modern machine learning tools can yield much more powerful results when raw data are combined first.

Content Creation: Continuous content creation informed by data should be created cheaply with crowd-sourced labor and directed at both earned and paid media, focusing on low cost and high reach. Message testing needs to be externally valid and focused on where it has the highest marginal lift.

9. Content creation INFORMED BY BASELINE data. Data on targeted Americans – both on policy preferences and psychographics – should be used to inform (and cut down on) content dimensions considered for testing.

10. CROWD-SOURCED viral content. As opposed to relying on boutique ad shops that create curated ads for $$$$, we can leverage record engagement on the left for this task.

11. Content creation aiming for both, EARNED and PAID media. Content should be designed with two distribution channels in mind: organic, through social networks, AND paid media.

12. LIMIT MESSAGE TESTING to as few dimensions as possible. We believe that baseline data provides much more stable information regarding what political/psychographic content to focus on. Mote limited content details – color-schemes, placing, sizes of images etc. are much better suited for testing.

13. Message testing limited to ORGANIC ENVIRONMENTS. This is key to getting real treatment effects (i.e., externally valid), especially if treatment groups can be targeted weeks or even months post treatment, given what we know about decay of communication effects.

Content Distribution: Time to focus on digital, targetable media, with long-term strategy to win people over years, not weeks or days, with continuous and repeated exposure tested externally, i.e. outside the distribution platforms).

14. Content distributed DIGITAL FIRST. As we have pointed out in our academic work, the marginal value of pouring $$$ into DMA-level cable ad buys diminishes (a) over time of the campaign, (b) as $$$ spent in the DMA increases. Americans natively spend time online and on their phones, and we need to reach them there. And while Addressable TV is a new attractive alternative, we need to be mindful of the lack of marketplaces governing addressable TV, which means that we have to buy full DMAs, even if we are only interested in targeting a segment in that DMA. Being outspent by Trump’s campaign – especially when his content is geared toward the general election and ours is not, is a real disadvantage!

15. Different content distributed to different individuals. Targeting is good, but remember not to over-target and account for blow-backs, or boomerang effects, from mistargeted content. Boomerang effects stemming from mis-targeting certainly have the potential to depress overall treatment effects that are already quite low to begin with in the real world.

16. Content distributed with a LONG-TERM STRATEGY. As opposed to lining voters up on the horse-race dimension and going after the median voter, we need to use our data repositories to talk to Americans – no matter whether they voted or not, whether they fall in the middle of the horse-race distribution or not – about the issue they care about, and do so in a positive light. This creates long-term positive effects, and reflects our underlying belief that American voters hold different, but meaningful views on different issues,and that we can use this potpourri of policy preferences to our advantage, if we address the right ones with the right folks. For example, Republicans taking progressive stances on healthcare but conservative stances on LBTQ should be targeted with persuasive appeals on healthcare, while Republicans taking conservative stances on healthcare but progressive stances on LBTQ should be targeted with persuasive appeals on LBTQ.

17. Content distributed via NATIVE IDENTIFIERS. We cannot continue to use offline identifiers to target in the digital realm. Instead, we need to build repositories stored by MAID identifiers allowing a flawless integration into digital platforms and enabling cross-device targeting, while maintaining privacy.

18. Content distributed CONTINUOUSLY and REPEATEDLY. We need to hit Americans with the same content in various forms over time to combat the decay of effects stemming from one-time-exposure interventions.

19. Data used to generate EARNED MEDIA. For example, baseline data can be geared towards keeping mainstream media accountable regarding the unpopularity of major Republican polling initiatives. We can achieve this by distributing content written around timely public opinion data both organically and to paid channels (shout-out to the folks at Data for Progress who have internalized that strategy).

20. Large-scale data repositories used to TEST INTERMEDIATE EFFECT and ADJUST. Attitudinal data on targeted segments can inform treatment effect estimates that (a) include effects among non-targeted segments who might register effects because content has been shared with them via their social networks – both online or offline, (b) offer an intermediate estimate of movement among target demographics in-between post-intervention tests and elections.

INVESTMENT: This only happens if progressives shift spending from campaigns to continuous operations. We started with financing, and we shall close here: We need to sensitize our donor base to a different spending-culture. One reason progressive donors have been focusing on electioneering: elections provide measurable RoI. Continuously updating data repositories can help replace elections as the only RoI in our space, and can incentive progressive donors to move away from an electioneering-centered spending model.

GOP misinformation campaign

Republicans and president Trump have been engaging in full-blown misinformation campaigns for years now – everything from crowd-sizes at Trump speeches to illegal immigrants voting to promises of universal coverage and rhetoric around a stronger healthcare amidst dropping numbers of insured. No question, building a policy argument substantially based on blatant lies is intentional GOP electoral strategy. And, it has massive effects, in that likely Republican voters readily believe the misinformation spread by the Republican Party. The surprising element: Not only Republicans have internalized the GOP misinformation campaign, so have a sizable chunk of Democrats. Let’s look at the data (this is based on a slightly under-powered PredictWise poll, N=600; 03/27/2019; full data here; Methods and validation here).

Republican bias of misinformation: immigration and healthcare

First up immigration. Only 16% of all likely 2020 voters know that there are less undocumented immigrants in the US than 2009, 79% believe that the number of immigrants has stayed the same, or even increased (64%). Among Democrats, only 26% know that the number of illegal immigrants has declined, with 61% (!!) believing the number has increased.

Second, health insurance: A good chunk of overall voters believe that the number of uninsured decreased or stayed the same during Trump’s presidency (54%), while of course it has gone up. Among Republicans, the vast majority believes the number of insured to have at least remained the same, with 31% (!!) believing the number has gone down, not up. Again, numbers are not so dissimilar for Democrats: 45% believe that the number of insured has at least remained the same, with 23%, almost a quarter of all Democrats, wrongfully believing the number has gone up.

Republican bias of misinformation: entitlement spending and gun violence

The data looks similar when it comes to entitlement spending. Only 20% of all voters know that more Federal money is spent on the old, as opposed to on poor, in absolute terms. Among Republicans, 71% believe that the poor receive at least as much Federal spending as the elderly, with 47%, almost half of all Republicans, wrongfully believing that the poor receive more Federal spending than the elderly. Again, a majority of Democrats is wrongfully convinced that the poor receive at least as much Federal spending than the elderly (60%).

Lastly, most voters believe that car crashes kill at least as many Americans every year as gun violence (62%), although car crashes kill about 38,000 Americans a year, compared to 40,000 Americans being killed by gun violence, on average. Among Republicans, 75% believe that car crashes kill the same amount or more, with 35% believing that car crashes kill twice the amount of Americans compared to gun violence. Even among Democrats, a majority believes car crashes to be at least as fatal as gun violence (52%), with 40% believing that car crashes kill substantially more Americans than gun violence.

Republican bias of misinformation: even among Democrats?

Without a doubt, the Republican propaganda machine spread misinformation to the Republican base quickly and effectively. The big question to us: Why does Republican misinformation register among likely Democratic voters as well? Our working hypothesis: mainstream media does not do enough to combat the onslaught of strategic Republican misinformation campaigns.

The Big Question and our Larger Vision

When we were approached by The Atlantic in early 2018, we developed a bold idea together. Develop a map (both geo-spatial and geographic) of what we as researchers call affective polarization, or political tolerance. The method: collect a (fairly) large survey of about 2,000 respondents, isolate demographic and neighbor-hood-compositional characteristics predictive of a uni-dimensional measure of tolerance toward out partisans, and assess how these traits vary county-by-county – I will not review the methods again, spelled out diligently here. Some caveats we have made clear from the beginning: To assess true (as opposed to modeled) variation in tolerance by county, one would need access to a hundred-Million-N survey. Even if we had access to this kind of survey (which we don’t), there is still a trade-off when it comes to cost and time. Do we gain anything when we interview Americans in every county, especially considering trade-offs with cost and time to collect data? The answer, I think, depends on the outcome of interest. Additionally, political tolerance is a latent phenomenon – difficult to measure: simply asking “are you politically tolerant” would lead to severe measurement error, which is why we rely on a multi-item-scale capturing this latent trait derived from years of research on this topic (for some of my work on antecedents of – and potential cures for – affective polarization, see here, here).

In some ways, this effort represents the larger vision of PredictWise: Using sharply declining cost of data collection (structured survey data AND unstructured ambient data), access to large-scale voter file and ambient data, as well as advances in computation and statistics, to answer questions pertaining to important phenomena beyond the horse race, with the goal of ultimately giving progressive campaigns ammunition for data-driven message selection, targeting, and tracking of RoI. Data collection around our core effort has been ongoing since 2017, putting the raw data base of PredictWise at 300,00 respondents, and Millions of ambient data points (read: application usage behavior etc.) per respondent. We have deployed this technology at the aggregate – identifying congressional districts we believe to be amenable to long-running progressive campaigns despite dismal-looking horse race estimates (here), and at the individual, providing superior and faster ways to cut lists of persuadable targets for digital ad buys (here). For The Atlantic story, we collected custom data in April 2018, and scored counties by estimated political tolerance (full methods: here). The goal of this undertaking: combining quantitative methods with qualitative reporting, or what I would call data-driven ethnography. Ultimately, we tried to find both quantitative and qualitative evidence in support of positions toward the top or bottom of our tolerance scale. Here is a preview:

Suffolk County, MA:

• Population: 797,939 (2015)
• Density 13,758/sq mi (5,312/km2)
• PredictWise Polarization Score: 65.39
• Mean Partisan Identification Homogeneity Score at Census Block level, where -100 is 100% Dems, 0 is 50% Reps and 50% Dems, and 100 is 100% Rep ): 80.50
• Identified couples: 55,854
• Agreement rate among couples on party identification: 89.99%

In contrast:

Jefferson County, NY:

• Population: 117,635 (2015)
• Density 92/sq mi (36/km2)
• PredictWise Polarization Score: 46.13
• Mean Partisan Identification Homogeneity Score at Census Block level, where -100 is 100% Dems, 0 is 50% Reps and 50% Dems, and 100 is 100% Rep ): -26.97
• Identified couples: 38,708
• Agreement rate among couples on party identification: 75.68%

Plus, please read the fantastic piece by Amanda Ripley on more qualitative evidence suggesting Jefferson County stands out: here.

Critique: Fair and unfair

This story has received a ton of attention, and, not surprisingly, a ton of critique. One stream of criticism relates to obsessing over perceived precision in all 3,000 counties (“hey, I live in McCormick county, you guys cannot be serious!”). I address this critique at some length our Methods addendum. The concerning part of this critique: as far as I know there is no ground truth map of political tolerance – too often we quickly discard non-intuitive results based on a set of preconceived ideas (in fact, PredictWise is in the business of changing this). One element that this critique gets especially wrong is identifying the counterfactual to our work, which is not a perfect US map of political tolerance, but no data at all.  Another stream of criticism attempts to reverse engineer our algorithm on the spot (which is public: here) – and not surprisingly gets it wrong every single time. This is a concerning form of engagement (mostly because it reveals a very low ability-to-arrogance ratio), but nothing we need to seriously concern ourselves with.

Of course, there has also been fair critique. Why do we have sharp discontinuities among some states? We have pointed out the trade-off between including party in our models, which necessitates counting partisans at the county level – a notoriously difficult task, and not including it, which would omit an important driver of tolerance. Yes, our counting rules can differ from state-to-state (which is the only way in which artifactual discontinuities at state borders can arise), but again the ground truth is not known. If you live in Florida, you might reside in the same media market as your peers across the state border, but you still receive a different dosage of digitally targeted political content, and your neighborhood conversations are likely different as well.

Going forward

Like it or not: We have opened a debate about what is driving political tolerance, and how drivers vary across the US. By no means do we argue that we have presented a definitive answer on the subject. Instead, we hope to kick off the virtuous cycle of academic-style research: replication, testing, improving. For what it’s worth: looking into how we count partisans off of voter file data would be a great place to start. Of course, I would be very much honored being a part of these efforts going forward, and I want to thank all of you who have provided substantive critique so far. If you are interested in the raw data, click here, and please do get in touch for the password: tobi@predictwise.com.

Identifying the most and least politically tolerant county in the US – PredictWise analytics and reporting by The Atlantic

Today, The Atlantic published a story identifying the least and most politically tolerant counties in the US and along the way ranking every county on a political tolerance scale, using analytics by PredictWise. The story is great for multiple reasons (yes, I am biased), not the least of which is the combination of rigorous quantitative analyses with qualitative methods, or what I would refer to as data-driven ethnography. I am not aware of many journalists that get at incredibly complex problems with a structured methodological approach. It opens a healthy debate and beats every New York Times Ohio diner story! Of course, every exhaustive ranking opens up numerous debates, and some are more fun than others: Residents of Suffolk County might feel the county’s top spot is undeserved: after all, many residents there have conversations with folks, even friends, identifying with the out party. Here is an example: Susy (name changed) is outraged: After all, she has always prided herself with fostering friendships across the aisle – her friend who she gets breakfast with once a week is a Republican, and while she disagrees with her politically, she loves her friend dearly. Her question: Does the model pick this up? The answer: No, it does not, it cannot, and it never will.

Every model is a representation of some unknown outcome that offers estimates of that outcome. Frustratingly, error is inherent in all models, and not knowing the true outcome (in this case political tolerance) means we can never know the true error. Using analytics, surveys and Big Data to reflect on microtrends rather should be a virtuous cycle in which researchers lay bare their assumptions and decisions, and other researchers check, correct and ultimately improve upon the first model. In this spirit, here is a more detailed breakdown of PredictWise analytics, including transparency re decision rules, data, etc.

The Survey Instrument

First, PredictWise collected 2,000 survey responses across the country, using a sampling technique called Random Device Engagement (RDE). For more background, read here, but the gist of it is that we use advertising networks on mobile devices to engage random people where they are to answer our surveys. RDE has a good coverage of 7,000,000 respondents in the US (much deeper than most panels), and allows us to collect ambient data on top of survey responses: most interestingly a rich history of highly precise device-based geo-location coordinates. We then surveyed our unique respondents on 14 questions – the full survey instrument is below:

1. How would you react if a member of your immediate family married a Democrat?
2. How would you react if a member of your immediate family married a Republican?
3. How well does the term ‘Patriotic’ describe Democrats?
4. How well does the term ‘Selfish’ describe Democrats?
5. How well does the term ‘Willing to compromise’ describe Democrats?
6. How well does the term ‘Compassionate’ describe Democrats?
7. How well does the term ‘Patriotic’ describe Republicans?
8. How well does the term ‘Selfish’ describe Republicans?
9. How well does the term ‘Willing to compromise’ describe Republicans?
10. How well does the term ‘Compassionate’ describe Republicans?
11. How do you feel about the Republican Party today?
12. How do you feel about the Democratic Party today?
13. How do you feel about Democratic voters today?
14. How do you feel about Republican voters today?

In addition we collected demographic information, partisan identification and matched respondents back to the full voter file using our history of geo-coordinates, taking as the home latitude-longitude pair the modal data entry between 07 pm and 05 am local time, and relying on address+demographics fuzzy matching. Finally, we have to drop all self-declared independents – after all, how do you determine tolerance for the out party of somebody who has no partisan affiliation (although we do count self-declared Independents who consistently score one party very low and the other party very high on Feelings Thermometers we collected as partisans)?

Methodology: High Level

PredictWise routinely relies on highly evolved variants of Mr.P. (or, spelled out: multi-level regression and post-stratification). The method allows us to take relatively low-N-survey data and derive small-area estimates. In fact, we have driven the methodological debate related to Mr.P. for years, and published extensively on it (here, here, here). In short, we model the outcome of interest (here: political tolerance) based on: urbanicity based on home address, age, gender, education, household composition, race, party affiliation, and two variables we use to describe the neighborhood: age variation and variation in partisan identification at the census block where the individual resides. Our model is a multiplicative multi-level model producing estimates of political tolerance for Millions of demographic combinations. These models are powerful because every single response is used to train all parameters. So, the political tolerance of a married Republican over 55 with a college education, living in suburban neighborhoods with a high mix of partisan affiliation and age (measured at the census block) increases precision in all separate parameters simultaneously! Our full Mr.P. model is complex, evolved over years, and spelled out at the bottom for methods geeks (any/all feedback welcome!).

The last step of this kind of analysis means weighting estimates for all (in this case many Millions of) demographics by the fraction of the demographic of interest in the target population, and that is the crux: Of course, we have to identify the fully interacted count table at the county level. So, we have to know: how many white, married Republicans over 55 with a college education, living in suburban neighborhoods with a high mix of partisan affiliation and age are there, really? That data is unknown. We do our best to use augmented full commercial voter file acquired through TargetSmart , and impute the many missing records with records from the ACS at the census block group, which is done probabilistically. There is one further difficulty: we have to identify partisans. It is crucial that we identify partisans in the model, as there is some strong support for asymmetric polarization – members of different parties feeling differently about political tolerance – so not including it in the model introduces error (which we call poor model fit), but including it in the model means identifying the exact number of partisans by county – a notoriously difficult task. In short: it is a trade-off. We follow a pre-defined decision rule to identify partisans: 1) relying on partisan registration , 2) relying on primary vote in case party registration is not available, 3) relying on voter file models calibrated such that the national distribution of Republicans and Democrats matches the national Gallup average. Can mistakes happen? Yes, we do suspect some of the sharp state differences can be artifacts of how party data is collected at the state level. Counter argument? We only run into these differences in a handful of states, and the most blatant examples, South Carolina and Florida, are states in which we find partisans to be very insulated in their neighborhoods by age and partisan affiliation. So, we decided to let the data speak for itself, instead of smoothing (read: fudging) our results ex post to avoid controversy.

The Bottom Line: Replicate, Debunk, Further the Debate (But Be Transparent!)

PredictWise prides itself in transparency. I am beyond happy to see the kind of healthy and lively debate the Atlantic story created – that is a good thing! And, more data and better models will always improve existing research. In that sense, I cannot wait for these results to be replicated, shared, debunked and improved. Of course, the perfect solution is a survey answered by 100s of Millions of Americans – but that remains elusive. Instead, a combination of novel survey methods, analytics and computing – made possible by recent advancements in statistics and computer science – allows us to get some, any handle on geo-spatial variation in and understanding of phenomena like political tolerance more important than ever in today’s political climate. In that sense, let this be the first of many analyses on this subject. And to all applied researchers: happy replicating, improving, debunking!

Methodology: Details

Here is our variant of Mr.P used in  (Warning: it gets technical). We model the survey data based on a Bayesian quasi-IRT model assuming every outcome we are interested in can be explained by an item-specific discrimination parameter or slope, and a set of item-and category-specific difficulty parameters or intercepts plus a single latent trait (read: our respondents’ standing on the tolerance dimension). In essence, these are sequential Bayesian ordered logits assuming the same underlying latent trait eta. Here is the Bayesian spirit of the models applied to two outcomes – whether you are OK with your offspring marrying an out party spouse, and how selfish you think members of the out party are:

.
.
.

The latent trait itself is defined by a set of individual-level and census-block-level predictors: urbanicity based on home address, age, gender, education, household demographics race, party affiliation, and two variables we use to describe the neighborhood: age variation and variation in partisan identification at the census block where the individual resides (see below). All predictors are themselves drawn from a prior normal distribution with mean 0 and variance learned from the data patterns.

Based on this equation, we create outcomes for each of the demographics we care about – notably the full set of interactions of all variables above (so, Millions of demographics!). The last step of this kind of analysis means weighting estimates by the fraction of the demographic of interest in the target population (see above for more discussion on the projection or target space).

In the Atlantic visualization, the color scale relates to the county-level percentile of eta hat!

February 24, 2019 at 7:59 PM ET: Here are the final market predictions (not much changed):

February 23, 2019 at 8 PM ET: I am going to blog from here on Oscar night. Here are the latest predictions. This will update live during the presentation! Normally markets get 19-20 binary “correct”. But, I should warn everyone that the average top prediction is the lowest I have ever seen, just shy of 70 percent. Which is means that if these predictions are well calibrated, there will be just 16-17 correct tomorrow.

Segmentations powering digital ad targeting:

Targeting advertising on the basis of individual-level segmentations is not a new idea: other organizations have collected custom survey data – time- and cost-intensive, to create static segmentation of persuadables, as a one-size-fits-all solution for digital targeting of the entire campaign. PredictWise offers a radically different approach. We have created technology than can create completely customized segmentations for a host of diverse use-cases, with the click of a button and almost in real time. In short, our AI draws on API calls layering analytics on top of our massive data base of tracking all Americans on more than 250 political, economic and psychometric dimensions. Given the low cost of computation, segmentations can be developed for specific creatives within hours, beat current much more time- and cost-intensive solutions on all relevant behavioral metrics such as click-throughs, viedeoplays, or engagement (more below), and allow campaigns to evaluate RoI in real time.

Disruption:

1. Custom segmentation for custom use-cases: In the old world, creating segmentations was expensive and time intensive, due to custom data collection. With the PredictWise method, segmentations can be created with a click of a button, within hours. This is possible because we do not collect custom data. And, that means it is scalable to produce segmentations matching the content of specific creatives, yielding much higher return on investment. This has virtually no variable cost to us!
2. True persuadability scores: As supposed to only looking at horse-race data, our segmentations fold in attitudinal data matching the content of the creative and additional measures of party strength. So, if candidate A wants to target a digital ad on expanding Medicare, our persuadables segmentation identifies soft Republicans who are tolerant toward the out party, likely to vote, and support expanding Medicare
3. Updates: We can produce regular updates that reflect true shifts in the underlying sentiment. As campaigns progress, people should move between segments, such that a static segmentation misses the key progression of the campaigns. And, we maintain records of movement such that month-over-month changes are easily tracked, providing convenient RoI for large-scale interventions, e.g. has an issue-centered ad campaign in a certain congressional district succeeded in moving likely voters from Persuadable to In Our Camp?

Validation:

In late 2018, we were approached by OpenProgress, doing work for the CA-45 (Katie Porter) campaign. The campaign was in need of a segmentation identifying the most persuadable voters to power digital ad-buys on Facebook. Two rival segmentations were available: (a) Generic segmentation using voter file data, and (b) Custom segmentation based on custom, cost- and time-intensive survey data (collected over the course of a month, with 10x our raw-cost). PredictWise, instead, created a custom segmentation for matching the primary message content of the Katie Porter’s campaign (healthcare; taxes) to likely voters inhibiting a progressive position on these dimensions, as well as tolerance toward the Democratic party and soft support for the Republican candidate, based on the PredictWise back-end data tracking 250+ dimensions for all Americans. Each creative was targeted to all audiences with a similar budget. Our segmentations beat the rival segmentations handily and significantly on all relevant behavioral metrics of engagement (Plot1). Differences are especially stark on comments, with PredictWise providing a 2x lift over the custom audience and a 3x+ lift over the generic audience.

Plot 1: RoI of Ad-buys on Facebook powered by PredictWise segmentations and rival segmentation in CA-45

These differences are robust to day of week or overall $ spent on each creative and audience. Another way of describing the PredictWise advantage: for every $1 spent additionally, you would have gotten 0.015 comments and 0.74 engagements more had you used PredictWise versus the rival audience, all else equal (significant effects; Plot 2).

Plot 2: “Premium gains” in RoI per Dollar Spent when PredictWise vs. rival segmentation in CA-45

Of course, behavioral metrics are no fool-proof indicator of persuasion – in fact there is no one fool-proof metric. For example, randomized control trials across exposed and non-exposed have (a) a hard time controlling for decay of effects, and (b) say little about your ultimate goal, which is moving demographic strata from Persuadable to “In our camp”. In 2020, our offering of audiences will include segmentation updates (see above), offering the best metric of externally valid RoI to date.

Methods:
Our segmentations are built on our massive data tracking all Americans on 250+ political/economic/psychometric dimensions. The data is built on the basis of 300,000+ survey respondents and more than 30 Million behavioral data points over the last year, with close to 100 Million data points in total. Models derived from this survey data – relying on state-of-the-art PredictWise machine learning algorithms – are projected onto a specially curated target universe. Segmentations are created from there via API calls, within minutes.

Just this weekend the mainstream media highlighted three ways that its unwritten rules actually encourages extreme and non-normative political behavior.

1) If the mainstream media sees details of a policy, it will tear it apart. But, if it does not see details of a policy, it will take its supporter’s word for what it does, even if it is impossible and otherwise absurd in the context of the supporter. Example from this weekend was the Green New Deal, which can be read here. I fully support some of this and have serious concerns with some of it. It is ambitious and necessary, even if it needs to evolve to work. But, the mainstream media spent most of the weekend focused on the most extreme aspects of the deal or some side debate over talking points that may or may not have been on a website. Seeing the details should signal the seriousness of the effort, but the mainstream media gets bogged down in the negative interpretation or salaciousness of details. Conversely, Republicans spent eight years claiming to have a magical healthcare plan to replace ObamaCare. It would cut costs, raise coverage, and there would be a unicorn in every driveway. Of course, they never showed anyone this magical healthcare plan, so the mainstream media just reported on what the Republicans said the plan would do. Thus, Americans were legitimately surprised when Republicans took control of House, Senate, and President and had absolutely no plan for replace ObamaCare.

This policy encourages non-normative behavior: better to have no plan get favorable press than to have a plan so the press can tear it apart. And, this policy encourages extreme behavior: no need to moderate your position if you can have extreme policies, but mainstream media will report whatever claim you are doing, not what you are doing.

2) The mainstream media holds Democrats to a higher standard than Republicans, because they know Republican leadership does not care about integrity or ethics. This asymmetric accountability is absurd, and allows Republicans to become more extreme and cruel without a necessary check. This weekend Republican President Trump mocked the genocide of Native Americans, then his son joined in with even more direct commentary. A Republican leader, communications director for Turning Point USA, glorified Hitler’s early work (she is totally against his post-September 1, 1939 expansionist policies). There is justifiable coverage of two very credible claims of sexual assault on the lieutenant governor of Virginia, but no one bothers to mention the 19+ credible claims against the Republican president. Mainstream media does not bother to report on Republican ethics, because they know Republican leaders will not take action. But, they will never take action if the mainstream media continues to give them a free-pass.

This policy encourages non-normative behavior and extreme behavior: Republicans are free to lack ethics and integrity because mainstream media is not interested in reporting on them.

3) Mainstream media still does not know to report on President Trump, they cannot help but drag the target of his lies or harassment into their coverage. Giving him a win, and incentivizing lies and harassment. This weekend he went after Senator Warren by degrading Native Americans. Why report the target? Just report on the racism and racist policies, without providing that context. There is a deep academic literature that shows that reporting a lie, even with clear context that it is a lie, just spreads it further. President Trump rose to the top of the Republican Party by leading the charge of racist comments against President Obama. The media repeated this false accusations ad-nauseum, helping to make them mainstream. They could have just reported that Donald Trump is making racist lies. It is that simple.

This policy encourages non-normative behavior: mainstream media rewards lying and harassment by repeating it over and over.